This dataset is free to browse and gated for download. Approval is tied to a Papers With Backtest subscription, which also covers the other datasets in this organisation and the strategy catalogue at https://paperswithbacktest.com. Plans and what each one includes: https://paperswithbacktest.com/pricing

All Monthly Risk Indicators

The Chicago Fed National Activity Index since 1967, the geopolitical risk index since 1985 with its threats and acts and its historical version since 1900, the news-based US economic policy uncertainty index since 1900 and the global one since 1997, Shiller's cyclically adjusted price-earnings ratio since 1881, and the Baker-Wurgler investor sentiment index since 1965 with its five proxies -- monthly, under their own names and in their own units.

Shape measured at the next refresh of dataset_cards/facts.json. Refreshed monthly.

Why It Matters

Business-cycle, uncertainty, valuation and investor-sentiment measures that papers use as conditioning variables, at the frequency they are published. They replace the monthly columns of All-Daily-TuringRiskIndex, where the same indicators were rescaled into VIX units, preceded by a synthetic history back to 1960 and interpolated into smooth daily paths -- an interpolation that let next month's print leak into this month's rows. Here a month holds that month's value and nothing else.

Load It

pip install --upgrade pwb-toolbox

from pwb_toolbox import datasets as pwb_ds

df = pwb_ds.load_dataset("All-Monthly-RiskIndicators")
cfnai = df.set_index("datetime")["cfnai"].dropna()

# Baker-Wurgler: keep only the months already public on the decision date
asof = "2010-12-31"
sent = df[df["bw_available_from"] <= asof].set_index("datetime")["bw_sent_orth"].dropna()
latest = sent.iloc[-1]   # December 2009, published by 30 June 2010

Columns

Column Name	Description
datetime	Last day of the reference month, not the release day (see Timing).
cfnai	Chicago Fed National Activity Index: weighted average of 85 activity indicators, zero = trend growth, from 1967-03.
cfnai_ma3	Its three-month moving average, the version the Chicago Fed reads for recessions (below -0.70), from 1967-05.
gpr	Geopolitical risk index (Caldara and Iacoviello), ten newspapers, 1985-2019 average = 100, from 1985-01.
gpr_threats / gpr_acts	Its geopolitical threats and geopolitical acts components.
gpr_historical	The historical geopolitical risk index, three newspapers, from 1900-01.
us_epu_news	News-based US economic policy uncertainty index (Baker, Bloom and Davis), as published monthly from 1900-01.
global_epu	Global economic policy uncertainty index, GDP-weighted average of national indices at current prices, from 1997-01.
cape	Shiller's cyclically adjusted price-earnings ratio of the S&P Composite: real price over the ten-year average of real earnings, from 1881-01.
bw_sent_orth	Baker-Wurgler investor sentiment index SENT⊥ (Eq. 3 of Baker and Wurgler 2006, updated): first principal component of five standardised sentiment proxies, each first orthogonalised to six macroeconomic series (since the 2026 vintage, detrended rather than year-on-year growth). Mean 0, s.d. 1 over the full sample, from 1965-07. Missing for 2025-10, where the October CPI was not published.
bw_sent	The raw index SENT (Eq. 2): the same principal component without the macroeconomic orthogonalisation, mean 0, s.d. 1, from 1965-07. Correlation with `bw_sent_orth` 0.99 in the 2026 vintage.
bw_dividend_premium	Proxy `pdnd`: value-weighted dividend premium (log market-to-book of payers minus non-payers, Baker and Wurgler 2004), in percentage points, from 1960-04. The index uses it at t-12.
bw_ipo_first_day_return	Proxy `ripo`: average first-day return on IPOs in the month, in percent (Ritter's data), from 1960-01; null in the months without IPOs. The index uses its 12-month IPO-weighted average at t-12.
bw_ipo_count	Proxy `nipo`: number of IPOs in the month, from 1960-01. The index uses its 12-month sum.
bw_closed_end_discount	Proxy `cefd`: equal-weighted discount on unlevered general-equity closed-end funds, in percent, from 1965-07.
bw_equity_share	Proxy `s`: equity issues over equity plus debt issues across the prior twelve months, as a fraction, from 1958-12.
bw_available_from	Declared first day the row's Baker-Wurgler values may be used: 30 June of the year after the reference year (see Timing). Null on rows without Baker-Wurgler values.

Timing

datetime names the month the value describes. When it becomes known:

cfnai, cfnai_ma3: released around the 20th-25th of the following month, and revised with later releases (this is the current vintage);
gpr, gpr_threats, gpr_acts, gpr_historical, us_epu_news: in the first days of the following month;
global_epu: several months later;
cape: the price is the month's average of daily closes; the earnings of recent months are estimates that Shiller revises as reports arrive;
bw_* (Baker-Wurgler): released once a year, by the authors, as one file covering every month up to December of the year just ended. The file carries no release date per month, so bw_available_from declares one: 30 June of the following year, for every month of the year. The vintages on record came out in May 2024, on 11 March 2025 and on 26 March 2026 (revised 21 April 2026), so the declared date is later than every observed release; January waits seventeen months, December six. Last observation 2025-12 (vintage of March 2026, usable from 2026-06-30); the next file, through 2026-12, is expected in the spring of 2027. Read this way, the most recent usable value is six to eighteen months old: that is the cadence, not a gap.

A strategy that trades at the end of month T on the value dated T is reading the future for every one of these columns; lag them by the release delay, and for the bw_* columns keep only the rows whose bw_available_from is on or before the decision date.

Two caveats no date can remove for the Baker-Wurgler index. Each vintage re-estimates it on the whole sample (the principal-component loadings, the standardisation and, for bw_sent_orth, the macroeconomic regressions), and the authors revise past months when the data improve: the values here are the current vintage, not what an investor saw at the time. And the authors warn against month-on-month changes: "do not use these series for measuring changes in sentiment (e.g. sentiment(t) - sentiment(t-1)) due to lag structures; these are low-frequency levels indicators". NYSE turnover, the sixth proxy of the 2006 paper, has been dropped from the maintained index.

Sources and Citation

Federal Reserve Bank of Chicago, CFNAI (via FRED series CFNAI and CFNAIMA3). Caldara, D. and M. Iacoviello (2022), "Measuring Geopolitical Risk", American Economic Review 112(4). Baker, S., N. Bloom and S. Davis (2016), "Measuring Economic Policy Uncertainty", Quarterly Journal of Economics 131(4); Davis, S. (2016), "An Index of Global Economic Policy Uncertainty", Macroeconomic Review. Shiller, R. (2015), Irrational Exuberance, 3rd edition, and the data on shillerdata.com. Baker, M. and J. Wurgler (2006), "Investor Sentiment and the Cross-Section of Stock Returns", Journal of Finance 61(4); Baker, M. and J. Wurgler (2007), "Investor Sentiment in the Stock Market", Journal of Economic Perspectives 21(2); the data file SENTIMENT.xlsx on Jeffrey Wurgler's NYU page and the code on the BWInvestorSentimentIndex GitHub repository.

Access

Browsing the card and the schema is open to anyone. Downloading the files needs an approved request, tied to a subscription: what each plan includes. The same subscription covers every dataset in this organisation and the strategy catalogue.

Elsewhere

The strategy catalogue
pwb-toolbox, the loader used in the snippet above
Every dataset in this organisation

Downloads last month: 98

Total file size:

124 kB