Recessions and Stocks: What the S&P 500 Did in Each One Since 1950
A recession is a significant, widespread decline in economic activity that lasts more than a few months. In the United States the official dates come from the NBER, and across the 11 recessions since 1950 the S&P 500's lowest close came before the recession's final month in all 11.
“Recession” is the word people reach for when they want to know whether to sell. This page sets the official US recession dates beside the S&P 500’s daily closes, so the relationship can be read off the record rather than guessed.
How it works
A recession is not two quarters of falling GDP in the US. That rule of thumb is common in the press, but it is not how the dates are set.
The official turning points come from the Business Cycle Dating Committee of the National Bureau of Economic Research (NBER), a private research group, which looks at a range of monthly measures of activity rather than one number.
The NBER’s table, read on 25 Sep 2026, states the convention plainly: recessions “start in the month after a peak in the business cycle, and end in the month of the trough.” So a recession is the stretch between two named months, a peak and a trough, and everything in between counts.
The dates arrive late. The committee waits until the data leave little doubt, which means months pass between a turning point and the announcement of it. Its announcements page lists, for example, that the December 2007 peak was announced on 1 December 2008, and the June 2009 trough on 20 September 2010.
The Federal Reserve Bank of St. Louis turns those dates into a monthly series called USREC: 1 for a recession month, 0 otherwise. That series is what lets a recession be lined up against a price file, and it is what this page uses.
Why the market moves first
Stock prices are a bet on future profits, and the economy is measured after the fact. Investors sell when they expect earnings to fall, which tends to be before the falling activity shows up in official numbers. They buy back when they expect the worst to be priced in, which tends to be before activity has stopped falling.
That timing gap is the whole story of recession trading. An investor who waits for the recession to be declared is reacting to information the index priced months earlier. One who waits for it to be declared over is usually buying after a large rebound.
Rates and credit are part of the chain. A yield curve that inverts, where short interest rates sit above long ones, has preceded several US recessions, which is why it is watched. It is a warning with a long and variable lead, not a date.
A worked example
Take the 2007 to 2009 recession, using S&P 500 closes. The NBER dates the peak to December 2007 and the trough to June 2009, 18 months apart.
Step 1, the market’s high came first. The highest close in the 12 months to the end of December 2007 was 1,565.15 on 9 October 2007, two months before the December 2007 peak month and three before the recession’s January 2008 start.
Step 2, the official start came much later. The index closed December 2007 at 1,468.36. By 1 December 2008, the day the peak was announced, it closed at 816.21. That is 816.21 divided by 1,468.36, minus 1, which is a fall of 44.4% before anyone could officially say “recession.”
Step 3, the low came before the end. The lowest close was 676.53 on 9 March 2009, 56.8% below the October 2007 high. By the end of June 2009, the trough month, the index was at 919.32, already up 35.9% from that low.
Step 4, the “all clear” came later still. On 20 September 2010, when the trough was announced, the index closed at 1,142.71, a further 24.3% above the June 2009 close.
So a reader who sold on the announcement of the recession sold 44.4% down, and one who bought on the announcement of its end paid about 24% more than the trough-month close. Neither date was useful for timing.
The original data
Across the 11 NBER recessions since 1950, from the July 1953 peak to the February 2020 peak, they lasted a median of 10 months, the shortest 2 months (2020) and the longest 18 (2007 to 2009). The FRED series counts 113 recession months out of 920 from January 1950 to August 2026, or 12.3% of the time.
Measured from the last close of the peak month to the last close of the trough month, the S&P 500 finished higher in 5 of the 11, with a median change of −1.4%.
The range runs from +17.9% (1953 to 1954) to −37.4% (2007 to 2009). In order from 1953, the eleven changes were +17.9%, −3.9%, +16.7%, −5.3%, −13.1%, +6.6%, +5.8%, +5.4%, −1.8%, −37.4% and −1.4%.
In all 11, the lowest close inside the recession came before its final month. In 8 of the 11, the highest close of the prior 12 months came before the peak month itself. The median fall from that high to the recession’s lowest close was −27.1%.
Month by month, recession months were weaker but not uniformly bad. The median S&P 500 month was +0.2% in the 113 recession months and +1.05% in the 806 expansion months, and the index rose in 53.1% and 61.5% of them.
The six troughs announced since 1980 came 8 to 21 months after the trough month.
The committee named the July 1980 trough on 8 July 1981, 12 months later; November 1982 on 8 July 1983, 8 months; March 1991 on 22 December 1992, 21 months; November 2001 on 17 July 2003, 20 months; June 2009 on 20 September 2010, 15 months; and April 2020 on 19 July 2021, 15 months.
On the announcement day the index was above its trough-month close in 5 of the 6, by +5.5%, +20.6%, +17.3%, +24.3% and +46.2%.
The exception was 2001, when it was 13.8% lower, because the bear market that began in 2000 kept going after that recession ended. The CSV of every recession and its S&P 500 figures has each date and close.
Every figure is price only, without dividends, and is recomputed from the daily closes and the dates above. In this site’s study of 24,971 trading and investing videos, deduplicated by video id, 3 put “recession” in the title, at a median of 1,045 views; one passed 20,000.
Reading a recession as it happens
Most of the signals arrive in the wrong order for a trader. Employment and output data are revised, sometimes heavily. Inflation can stay high into a slowdown, which delays rate cuts. And the index can fall 20% or more, a bear market, with no recession at all.
What the record does support is narrower. Stocks have tended to fall before a recession starts and to turn before it ends. So the useful question during one is not whether the economy is shrinking, which is already priced, but how much of the damage the index has absorbed.
That gap is also why recession forecasting sells so well and trades so badly. A correct call on the economy can still be a losing trade if the market got there first.
When it fails
The first failure is waiting for the official word. By the NBER’s own announcement dates, the start of a recession has been confirmed 4 to 12 months after the peak month since 1980. A plan that begins “if a recession is declared” starts after the damage.
The second is assuming a recession guarantees a fall from here. In 5 of 11 recessions the index ended higher than it began, measured month end to month end. Selling at the official start has sometimes meant selling close to a low.
The third is treating every large fall as a recession signal. The 1987 crash and the 2022 decline came without an NBER recession. A market crash and a recession are separate events that sometimes overlap.
A fourth is forgetting the sample size. Eleven recessions is a short list. The 2020 recession lasted two months and the 2007 one lasted eighteen, and a median drawn from so few, so different, cases is a description of the past, not a forecast.
And a fifth is trading the headline number. One weak GDP print is not a recession by the NBER’s method, and the first estimate is often revised. Reacting to it is reacting to noise.
Related
The bear market page covers the market-side measure of a bad stretch, which is defined by price rather than by the economy. The yield curve page explains the rate signal watched most closely before recessions.
And the market crash page times every S&P 500 fall of 20% or more since 1950, several of which had nothing to do with a recession.
Keep the recession question and the portfolio question apart. By the time a recession is official, the index has usually already fallen, and often already turned. A plan written in advance for a 30% fall does more work than any attempt to call the economy.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.