Economics & Money · Growth, inflation & crises

Who calls a recession

US recession dates come from a committee at the National Bureau of Economic Research, and its verdict arrives months after the economy has turned.

4 short lessons · about 20 minutes · 6 sources · updated November 2026

Three paper dioramas of a small town with a trench: one very deep, one spanning the whole town, one stretching far into the distance.

What you will learn

  • Apply the NBER's criteria of depth, diffusion and duration to judge a downturn
  • Explain why two straight quarters of falling GDP are neither required nor enough for a recession
  • Calculate how many months a recession lasted from its peak and trough dates
  • Explain why an expansion can begin while jobs and income are still below their earlier peak

Lesson 1 of 4 · 5 min

What makes a slump a recession

Before you readWho sets the dates for when a US recession starts and ends?

Correct.Not quite. The answer is A private research group.

The dates come from a committee at a private research organization, the National Bureau of Economic Research. No government agency sets them, and this lesson shows the three tests the committee uses.

No US government agency decides when a recession starts or ends. The dates in the history books come from the Business Cycle Dating Committee of the National Bureau of Economic Research, or NBER, a private research organization founded in 1920.

In 1961 the Commerce Department began reprinting the NBER's dates in a monthly report, without endorsing them. The government has never compiled a list of its own, so the committee's dates serve as the quasi-official record.

The committee's definition has three parts. A recession is a significant decline in economic activity, spread across the economy, that lasts more than a few months. The committee calls these tests depth, diffusion and duration.

Each test has to be met to some degree, but an extreme reading on one can make up for a weak reading on another. In 2020 activity fell so far and so widely that the committee called it a recession while allowing that it might prove brief. It lasted two months, from a February peak to a low point in April.

Because a recession has to reach across the economy, the committee reads economy-wide measures every month, such as consumer spending and industrial production. It has no fixed formula for combining them. In recent decades it has leaned most on two, payroll jobs and real personal income less government transfers. That's income adjusted for inflation, with benefit payments taken out.

Check yourselfThe 2020 US recession lasted only two months, far shorter than most. Why did the NBER's dating committee still count it as a recession?

Correct.Not quite. The answer is Its extreme depth and spread outweighed its brevity.

The committee weighs depth, diffusion and duration together, and an extreme reading on one can offset a weaker reading on another. The 2020 drop was so deep and so widespread that it qualified despite being brief. Duration still has to be met to some degree, and a typical recession must last more than a few months.

You've probably heard a simpler rule. A recession is two quarters in a row of falling real GDP, the economy's total output adjusted for inflation. The financial press uses it often. It usually agrees with the NBER, because most recessions do include two falling quarters.

It can miss in both directions. The 2001 recession, which ran from a March peak to a November low point, never had two falling quarters in a row. And in 2022 the official figures showed GDP falling in each of the year's first two quarters. The NBER dated no recession, and revised figures now show output rose in the second of those quarters.

The committee gives four reasons for not using the shortcut. It looks at a range of indicators, and GDP is only one of them. It judges depth, so two small dips needn't add up to a significant decline. Its dates are monthly, and GDP comes out quarterly. And when it does look at quarterly output, it gives equal weight to real gross domestic income, or GDI, which measures the same output by adding up incomes.

Try to recallA commentator says the US can't be in a recession because real GDP hasn't fallen for two quarters in a row. Explain why the NBER's approach disagrees, using 2001 as evidence.

The two-quarter rule is a press shortcut, and the NBER's committee doesn't use it. It dates recessions from a range of monthly indicators such as payroll jobs and real income, gives real GDI the same weight as GDP, and asks whether a decline is significant, widespread and lasting. The 2001 recession, from a March peak to a November low point, never had two straight quarters of falling real GDP, so a recession can happen without the rule being met.

So in the US, a recession is the stretch the committee eventually marks between a peak month and a low-point month. Its test is a fall that's deep, widespread and lasting, judged from many monthly measures.

Two falling quarters of GDP is a rule of thumb. It's usually right, and it has been wrong in both directions.

Lessons in this topic

  1. What makes a slump a recession

    Judge a downturn by its depth, spread and length, and see why two falling quarters of GDP don't settle it.

    Essentials · 5 min

  2. Counting a recession from peak to trough

    A recession runs from a peak month to a trough month, and jobs and income can take years longer to climb back.

    Essentials · 5 min

  3. Why the recession call comes late

    The NBER waits months to name a peak or trough, so economists also watch a faster signal built from the unemployment rate.

    Deeper · 5 min

  4. US recessions used to last twice as long

    Compare how long US recessions and expansions have lasted since 1854, and see how the NBER handles double dips and depressions.

    Deeper · 5 min

Three quick stories from this topic

Two falling quarters, no recession

In 2022 US GDP was reported falling two quarters in a row, and no recession was declared

Two falling quarters is the popular definition of a recession, but the committee that dates US recessions doesn't use it. It weighs monthly jobs and income figures too. Revised data now show output rose in the second of those quarters.

The rule misses the other way too

The 2001 recession, which the NBER dates from a March peak to a November trough, never had two quarters in a row of falling real GDP.

What the committee asks instead

It looks for a significant decline in activity, spread across the economy, lasting more than a few months. It judges that from a range of monthly measures, and GDP is only one input.

A recession announced late

The US was in recession for almost a year before it was officially announced in 2008

Activity peaked in December 2007, and the announcement came on December 1, 2008. The committee that dates US recessions waits until it's sure and the data have been revised. Its slowest call, the end of the 1991 recession, took 21 months.

Why the committee waits

Once it names a peak, the date stands. Even an immediate upturn would count as a new expansion, so it waits until it's confident a recession has occurred.

The fastest call

The quickest announcement came four months after the fact, on June 8, 2020, for the February 2020 peak. Troughs have taken from 8 to 21 months.

The two-month recession

The shortest US recession on record lasted two months, in the spring of 2020

It ran through March and April 2020 and ended the longest expansion on record, 128 months. A recession normally has to last more than a few months. This fall was so deep and so widespread that it counted anyway.

Why a short slump still counted

The NBER judges a recession on depth, diffusion and duration together. Each has to be met to some degree, but an extreme reading on one can offset a weaker reading on another, and the 2020 drop was extreme in both depth and spread.

It ended a record run

The recession cut short the longest expansion in the NBER's chronology, which ran 128 months from the June 2009 trough to the February 2020 peak. The next-shortest recession on record, in 1980, lasted six months.

Sources

This topic is part of From one sale to a whole economy, a guided path through related topics.

  • Inflation and purchasing power

    Read price headlines by separating the level of prices, their rate of change, and your own spending basket.

    4 lessons · 17 min

  • GDP and what it measures

    One number ranks economies and marks recessions. See what gross domestic product adds up, why growth figures strip out prices and divide by people, and what the number leaves out about how people live.

    4 lessons · 17 min

  • The Great Depression

    Between 1929 and 1933 US output fell by about a quarter and one worker in four had no job. See how a stock market crash became a worldwide slump, what ended it, and why economists still argue over its cause.

    4 lessons · 16 min

  • How banks create money when they lend

    Follow a new loan onto a bank's books and a borrower's, then see what happens when the money is spent or repaid and what stops banks from creating it without limit.

    4 lessons · 17 min

  • Why bond prices fall when rates rise

    When interest rates rise, a fixed-rate bond loses market value, but an owner who holds it to maturity still collects what it promised if the issuer pays.

    4 lessons · 20 min

  • Why prices change

    See how shortages and surpluses push a price, tell a change in demand from a change in supply, and find out what happens when a price isn't allowed to move.

    4 lessons · 22 min

All economics & money topics · Take the quiz