1929 Stock Market Crash: How Long Did Recovery Really Take?

The Dow Jones Industrial Average closed at a record 381.17 on September 3, 1929. Eight weeks later it was 40% lower. The worst days were Monday, October 28, when the Dow fell 12.8%, and Tuesday, October 29, when it fell another 11.7%. Those are the days known as Black Monday and Black Tuesday.

The crash was only the start. From there the index kept falling, unevenly, for almost three years, to 41.22 on July 8, 1932, 89% below its record. The Federal Reserve's history of the crash describes what followed as "our nation's longest, deepest depression." The Dow did not close above 381.17 again until November 23, 1954.

The broader S&P composite index peaked within two weeks of the Dow. Its highest weekday close was 31.86 on September 16, 1929. S&P's own records put the high at 31.92 on Saturday, September 7, because the New York Stock Exchange also traded on Saturday mornings then. Either way, it fell 86% to its own bottom on June 1, 1932. In money terms, $1,000 in it at the September 16 close was worth $641 at Black Tuesday's close and $138 at the bottom. The recovery figures below use the S&P composite because it is broader than the Dow and Robert Shiller's long-run dividend and price data cover it. In 1929 it tracked 90 stocks; the 500-stock S&P 500 only launched on March 4, 1957, according to S&P Dow Jones Indices. So these are reconstructed historical estimates, not the returns of any fund a 1929 investor could buy.

By price alone, through the Great Depression and beyond, 25 years: the Dow regained its 1929 high on November 23, 1954, and the broader S&P composite on September 22, 1954. It is accurate for what it measures, which is the index level with no dividends and no adjustment for prices.

Measured as a reconstructed broad-market series, with dividends reinvested and no taxes or trading costs, the answer is different, and it depends on two choices: whether you count dividends, and whether you adjust for prices.

• Price only: back to even September 1954, 25 years after the peak.
• With dividends reinvested: back to even in January 1945, about 15 years. This is called total return: it counts the dividends companies paid as if they were reinvested in the market.
• With dividends, after inflation: first back to even in November 1936, about 7 years, and back for good in July 1949, about 20 years. This is the real return version, and it got there first because consumer prices fell sharply in the early 1930s.
• Price after inflation: first back in December 1958, 29 years. This measure leaves out dividends, so it shows only what the index level itself could buy. It slipped below again many times and held for good only from late 1985, as consumer prices rose during and after World War II and again in the 1970s.

So "25 years" is too long by one measure and too short by another. To see how inflation changes a return over any period, try the inflation-adjusted return calculator, which uses rates you enter rather than historical data. Each date in the table below is the first close at or above the 1929 peak level.

Only by price. Counting dividends and falling consumer prices, the reconstructed S&P composite was back to its 1929 level twice before 1954: briefly in 1936, and for good in 1949.

The November 1936 recovery was brief. In our daily reconstruction, this series first crossed its 1929 starting level on November 4, 1936 and fell below it again by November 13. Exact crossing days depend on how the monthly dividend and inflation figures are spread across trading days, so read them as model estimates. The market fell back below its 1929 level that same month, dropped sharply again in 1937 and 1938, and in April 1942 the 1929 starting value, with dividends and inflation counted, was down to about half. It got back above for good in July 1949 (July 12 in our reconstruction), almost 20 years after the peak, and stayed above in every later observation we have. The price-only recovery wobbled too: the S&P composite slipped back under its 1929 peak in October 1954 before holding from November 3.

The start date matters as much as the measure. A reconstructed series that starts at the close on October 31, 1929, after the crash rather than at the peak, briefly returned to its starting level in the rally of early 1930, then fell far below. On the same basis, with dividends reinvested and adjusted for inflation, it first returned after the 1932 bottom in November 1935, about 6 years later, then slipped below again until 1943. When you see another figure, check which start date and which measure it uses.

Black Thursday was the day the panic began, but it was not the day the market fell most. You may read that the market "lost 11%" that day. That was roughly the fall at the worst point during trading, not at the close.

According to economist Harold Bierman's account on EH.net, the Dow dropped from about 306 to 272.32 during the session, on about three times the normal trading volume, and then recovered to close at 299.47. That close was down 2.1% on the day. The next morning's Wall Street Journal, as Bierman notes, credited New York bankers with stopping the decline with $1 billion of support.

The support bought two quiet days. On Friday the Dow closed slightly higher, and in Saturday's session slightly lower. The heavy losses came the next week.

On Black Monday 1929, the Dow fell from 298.97 to 260.64, a 12.8% drop from Saturday's close. On Black Tuesday it fell again, from 260.64 to 230.07, down 11.7%. The Federal Reserve's history describes the two days the same way: down "nearly 13 percent," then "nearly 12 percent." In two sessions the Dow lost almost a quarter of its value.

By the close on Black Tuesday the Dow was 40% below its September record. It kept falling to 198.69 on November 13, 1929, 48% below the record, before a rally in early 1930 and then the long slide to 1932. To see those weeks from the inside, one decision at a time, try our 1929 stock market crash simulator, a free replay with pretend money.

The name "Black Monday" also belongs to a later crash: on October 19, 1987 the Dow fell 22.6% in one day. That crash and its recovery are covered in our post on how long the market took to recover from Black Monday 1987.

Economists still disagree. Harold Bierman, writing for EH.net, notes that "no consensus exists about its precise causes." The factors most accounts point to are these:

• A long, fast run-up. According to the Federal Reserve's history, the Dow rose sixfold from 63 in August 1921 to 381 in September 1929. Economist Irving Fisher said prices had reached "what looks like a permanently high plateau."
• Borrowed money. Investors bought stocks on margin, paying part of the price with a loan. Bierman points to the growth of margin buying and of investment trusts and utility holding companies that had themselves borrowed heavily, which made them fragile when prices turned.
• Tighter money. In August 1929 the Federal Reserve Bank of New York raised its discount rate, the rate it charges banks to borrow, to 6% to rein in speculation. The Fed's history says the increase "had unintended consequences."

Bierman's own argument is that efforts by officials and the press to stop speculation, and bad news for heavily borrowed utility stocks in October 1929, helped trigger the fall. Once prices dropped, investors who had borrowed had to sell to repay their loans, which pushed prices down further.

Two things helped that you won't see on a price chart.

Consumer prices fell (deflation). The U.S. consumer price index fell 27% from 1929 to spring 1933, so a dollar in 1933 bought about 37% more than a dollar in 1929. A portfolio that was still down in dollars was down less in what it could buy. Consumer prices did not return to their 1929 level until April 1943.

Dividends kept coming, even as prices collapsed. At the June 1932 low, the S&P composite's dividend yield was about 13.8%, because prices had fallen much further than payouts. Reinvested, those dividends bought shares at very low prices.

Taken together, $1,000 at the 1929 peak was worth $138 by price at the 1932 low, but $164 with dividends and $208 with dividends after inflation.

These are index figures, and they describe the market, not people. During the Depression many investors lost income or had to sell, and those who had bought on borrowed money were forced to sell as prices fell. An index that was back to even in buying power did not mean its owners were.

The 1987 crash was faster, and 1929's was far deeper. On October 19, 1987 the Dow fell 22.6% in one day, yet the S&P 500 was back above its 1987 high by price 21 months after the crash (23 months after its August 1987 peak), against 25 years after 1929. Unlike 1929, the 1987 losses were not followed by a recession or a banking crisis, according to the Federal Reserve's history. The 1987 recovery, with its own table, is covered in our Black Monday 1987 post. The Panic of 1907 sits in between: the Dow fell 48.5% and passed its old record by price 9 years and 8 months after it, as our Panic of 1907 post shows. After the dot-com bubble burst in 2000, the S&P 500 needed 7 years by price to first get back above its record, and the Nasdaq 15, as our dot-com bubble post shows.

The difference was what came after. The Federal Reserve's history traces part of it to the Fed's own 1929 rate rise: under the international gold standard it forced other central banks to raise rates too, and tight money "tipped economies around the world into recession." Problems elsewhere in the financial system then "turned what may have been a short, sharp recession into our nation's longest, deepest depression." In 1929 the economy was already shrinking before the crash: the National Bureau of Economic Research dates the recession's start to August 1929 and its end to March 1933. The crash was followed by the Great Depression and a long bear market, and the Dow fell another 82% below its Black Tuesday close (89% below its September record) before it found a bottom in 1932.

For how markets recovered from other shocks, see how stocks performed during past wars.

• Dow Jones Industrial Average, daily closes 1928 to 1956: MeasuringWorth (Samuel H. Williamson). This series includes 1929's Saturday sessions.
• S&P composite, daily closes from December 1927: Yahoo Finance, ^GSPC history. This series has no Saturday sessions.
• Dividends and monthly prices: Robert Shiller's monthly S&P data. Inflation: BLS CPI-U, all items, not seasonally adjusted (CUUR0000SA0).
• Recession dates: NBER, US Business Cycle Expansions and Contractions.
• History: Federal Reserve History, "Stock Market Crash of 1929" and Harold Bierman Jr., "The 1929 Stock Market Crash," EH.net.

Method. "Back to even" is the first daily close at or above 31.86, the September 16, 1929 close and the highest in our data, which has no Saturday sessions. S&P's records show a Saturday high of 31.92 on September 7; measured from that level instead, every recovery date in this post comes out the same. Shiller's dividend figure for each month is an annual rate (the trailing four-quarter total), so we divide it by 12, spread it evenly over that month's trading days and add it to the daily price change. Inflation-adjusted values divide each day's level by that month's CPI-U. Our Dow figures are for a price-weighted index of 30 companies whose members changed over the years; the S&P composite tracked 90 stocks in 1929, and the 500-stock S&P 500 launched in March 1957 (S&P Dow Jones Indices). Dividends are reinvested in proportion to the index level each day, compounding, and the price series used has no dividends of its own, so nothing is counted twice. The total-return figures assume dividends were reinvested at no cost and with no tax; no index fund existed in 1929, so they describe the market, not any one investor. A second check that uses only Shiller's monthly averages gives the same month for three measures and one month earlier for price after inflation. "Sustained recovery" dates use daily closes through 1960, then Shiller's monthly averages through September 2023, where our downloaded copy of his data ends. Dates that depend on dividends or inflation are model estimates, accurate to about a month.

Historical data shown; past performance does not indicate future results. StockCram is an educational platform, not a broker-dealer or investment adviser, and nothing on this page is investment advice.

Every post in this series measures crashes the same way; our market-history method explains how.