Skip to content
Ze Finance
← All the news

#MARKETS & INVESTING

"In wartime the stock market collapses." From September 1939 to August 1945 the US stock market gained 28.3% in real terms, dividends reinvested and inflation removed. Of seven major conflicts since 1914, six ended in a gain.

Published on September 3, 2026

Ze Finance fact card: Markets & investing
+28.3% the real total return of the US stock market between September 1939 and August 1945, dividends reinvested and inflation removed, measured on the S&P Composite index in Robert J. Shiller's series

The calculation runs on the monthly S&P Composite series of Robert J. Shiller. From September 1939 to August 1945, one dollar invested becomes $1.28 in constant purchasing power, or 4.30% a year — while the nominal price rose only 16.1% and US inflation 28%. Recomputed on an independent series, the whole US market from Kenneth R. French's data library deflated by the Bureau of Labor Statistics price index, the same interval gives +58.0%. The two measures differ on the size, never on the sign. Of seven major conflicts involving the United States since 1914, only one is negative: the First World War, at −14.8%.

Do stock markets collapse during wars?

Not the US market: +28.3% in real terms between 1939 and 1945, and six winning conflicts out of seven since 1914.

In other words: what is measured here is the market of the side that won, never that of the losers.

Five caveats, because they matter

  • The path is nothing like a straight line: between September 1939 and April 1942 the market loses 36.3% in real terms. It regains its starting point only in July 1943, 46 months later.
  • The two independent series do not give the same figure: +28.3% for Shiller's S&P Composite, +58.0% for Kenneth R. French's whole US market. The gap comes from coverage — small caps are absent from the first — and from monthly averages against month-end readings.
  • The United States was neither invaded nor occupied, and won. The German, Japanese and French exchanges of those years — closures, requisitions, destroyed capital — appear in neither series.
  • Start and end dates are conventions. Shifting the starting point by a few months changes the size of the result.
  • Seven episodes do not make a law. This count describes what happened; it says nothing about the next conflict.

Why it still matters: the intuition that war means a crash does not survive the count — but only because the available data is that of the side that won.

For you

  • When you are shown how markets behave in wartime, which country is being described?
  • Would you have held from September 1939 to July 1943, watching a third of your capital go before it came back?

Source: reproducible Ze Finance calculation on the monthly S&P Composite series of Robert J. Shiller, Online Data (Yale), September 2024 vintage; cross-checked on the independent US stock market series from Kenneth R. French's data library (Tuck School of Business, Dartmouth), deflated by the Bureau of Labor Statistics consumer price index (CPIAUCNS).

View the source →

Download the fact card in high resolution (1856 × 2304) ↓