What is a recession?

A recession, in the definition the National Bureau of Economic Research (NBER) uses for the United States, is a significant decline in economic activity that is spread across the economy and lasts more than a few months. The NBER Business Cycle Dating Committee chooses the US peak and trough months. It weighs depth, diffusion, and duration, and it reads monthly indicators together with real GDP and real gross domestic income (GDI). In the United Kingdom there is no equivalent committee. The Office for National Statistics (ONS) acknowledges the convention that a technical recession is two or more consecutive quarters of contracting real GDP. This page is educational only. It is not investment advice.

Four ways people decide

The same headline can mean four different tests. The table keeps them apart. The NBER row follows the committee's business cycle dating page and its procedure FAQ. The UK row follows the ONS methodology note Communicating the UK Economic Cycle, revised on 11 November 2022. The Sahm row follows the notes on FRED series SAHMREALTIME. FRED is the data site of the Federal Reserve Bank of St. Louis.

How the NBER dates a US recession

The committee keeps a chronology. It does not publish a forecast. An expansion runs from a trough to a peak. A recession runs from a peak to a trough. The peak month is the last month of the expansion, and the trough month is the last month of the recession. Counting then starts in the month after the peak and stops in the trough month. Expansion is the usual state. The dating page says most recessions are brief, and that getting back to the previous peak level of activity can take much longer than the recession itself.

Depth means the decline is significant. Diffusion means it shows up across the economy, not in a single sector. Duration means it lasts more than a few months. The committee treats the three as somewhat interchangeable. Each has to be met to some degree, and an extreme reading on one can partly offset a weaker reading on another. That is the reason the collapse after February 2020 still counts. The trough was only two months later, in April 2020, and the fall was deep enough and widespread enough that the short length did not keep it off the chronology.

Monthly dates come from monthly series published by US statistical agencies. The dating page lists six. They are real personal income less transfers, nonfarm payroll employment, employment as measured by the household survey, real personal consumption expenditures, manufacturing and trade sales adjusted for price changes, and industrial production. Real personal consumption expenditures are the broad monthly measure of household spending. There is no fixed rule for which series enter or how they are weighted. In recent decades the two series given the most weight are real personal income less transfers and nonfarm payroll employment.

Real GDP cannot pick a month, because it is quarterly. The committee makes a separate decision about the peak or trough quarter. The two quarterly measures that are not available every month are real GDP, built from spending, and real GDI, built from incomes. The procedure FAQ says that, when the committee looks at production over quarters, it gives real GDP and real GDI equal weight. The gap between them is the statistical discrepancy. The FAQ says that gap was particularly important in the recessions of 2001 and of 2007 to 2009. Quarterly averages of the monthly indicators, especially payroll employment, are in the same discussion.

The peak quarter usually contains the peak month. The latest cycle is an exception, and the chronology marks it. The peak month is February 2020, which falls in the first quarter of 2020, but the peak quarter is the fourth quarter of 2019. The trough month is April 2020, and the trough quarter is the second quarter of 2020. A sentence about quarterly GDP can therefore point at a different slice of the calendar from the monthly recession date.

Recent US recessions, and how long the call took

Dating is retrospective. The committee waits until it is confident that a turning point happened, and until routine revisions have had a chance to land, so the chronology does not need a major rewrite. The procedure FAQ, revised on 7 October 2025, says there is no fixed timing rule. In the announcement table on that page, the shortest gap from a turning point to the announcement was four months. That was the February 2020 peak, announced on 8 June 2020. The longest was 21 months. That was the March 1991 trough, announced in December 1992. The three latest recessions are below. Length is the chronology's peak to trough duration. The lags are the elapsed months in the FAQ table.

The 2001 recession is the clearest split between the GDP shorthand and the chronology. The FAQ states that this recession did not include two consecutive quarters of decline in real GDP. The committee still dates a peak in March 2001 and a trough in November 2001.

The FAQ also says that, since the committee was created in 1978, it has not changed a turning point after announcing it. The expansions and contractions table is marked last updated on 14 March 2023. The business cycle dating page still identifies the latest peak as February 2020 and the latest trough as April 2020.

Two negative GDP quarters in 2022

In 2022 a lot of commentary treated two falling GDP quarters as a finished recession call. The estimates published that year did show two declines. On 29 September 2022 the Bureau of Economic Analysis (BEA) third estimate said real GDP decreased at an annual rate of 1.6 percent in the first quarter of 2022 and 0.6 percent in the second quarter. The same release said real GDI increased 0.8 percent in the first quarter and 0.1 percent in the second quarter. The average of real GDP and real GDI decreased 0.4 percent in the first quarter and 0.3 percent in the second. Spending and income were pointing different ways. That split is the statistical discrepancy the NBER says it weighs when it looks at quarters.

No new peak was added to the chronology. The latest recession on that list still ends in April 2020. Writing in November 2022, the ONS described the same US episode as one in which GDP had fallen for two quarters and a recession had not been officially called.

Later BEA updates changed the GDP path. The summary of the 2024 annual update of the National Economic Accounts says revisions from the first quarter of 2019 through the fourth quarter of 2023 left the direction of real GDP unchanged in every quarter except the second quarter of 2022. That quarter was revised up by 0.9 percentage point, from a decrease of 0.6 percent to an increase of 0.3 percent. The current BEA vintage on FRED series A191RL1Q225SBEA, retrieved on 8 October 2026, shows a 1.0 percent decrease in the first quarter of 2022 and a 0.6 percent increase in the second quarter. The units are the percent change from the preceding period, at a seasonally adjusted annual rate. The two-quarter decline was a property of the 2022 estimates. It is not a property of the current GDP vintage. The missing NBER date is older than that revision. No 2022 peak was announced while both quarters were still declines in the published GDP figures.

Early GDP prints move. The BEA versus ONS GDP page is the place to see how a US annualised quarter on quarter rate differs from a UK quarter on quarter percent, and why the vintage of the release matters. Open the BEA table for the quarter you mean before you repeat a 2022 headline.

The UK convention

The ONS note, revised on 11 November 2022, says the ONS will acknowledge the widely held convention that a technical recession comprises two or more consecutive quarters of contracting output. Contracting output here is a fall in volume GDP, which is real GDP. The rule is easy to state on the day a quarterly estimate is published. The same note lists what it leaves out. It does not ask how large the fall was. Early estimates are revised, and a change close to zero can switch sign. A downturn can also be serious without arriving as two consecutive negative quarters.

The note says the UK does not have a committee responsible for the economic cycle. In its own GDP communications, the ONS says it will look at the level of GDP as well as the quarterly change, keep saying that early estimates can be revised, and bring in other indicators along with the sign of volume GDP. It describes the NBER ideas of depth, diffusion, and duration as part of that wider reading. Those sentences explain how a statistical office talks about a number. They do not create a UK list of peak and trough months.

Use the ONS GDP bulletins for the figures. Country context is on the United Kingdom and United States pages.

The Sahm rule is a real time signal

The Sahm rule does not choose NBER dates. FRED publishes it as the Real-time Sahm Rule Recession Indicator, series SAHMREALTIME, and names Claudia Sahm as the source. The series notes say it signals the start of a recession when the three-month moving average of the national unemployment rate (U3) rises by 0.50 percentage points or more relative to the minimum of the three-month averages from the previous 12 months. This real time version uses the unemployment rate that was available in that month. The Bureau of Labor Statistics revises seasonal factors later, at the start of the next year. Aside from those seasonal factors, the unemployment rate does not revise in the way GDP does.

On SAHMREALTIME, retrieved on 8 October 2026, the July 2024 observation is 0.53 percentage points, so it met the 0.50 line. August 2024 is 0.57 and September 2024 is 0.50. The NBER chronology did not gain a peak after February 2020. The September 2026 observation is 0.00 percentage points. FRED dates that update 2 October 2026. A reading at or above 0.50 is the signal described on the series. It is not a peak month, and it is not a reason to buy or sell anything.

Unemployment and the NBER chronology move together only loosely. The procedure FAQ points out that the unemployment rate has no long run drift of the kind activity has, and that it moves in the opposite direction from most activity measures. It can rise before the peak, while activity is still growing, and it often keeps rising after the trough. The unemployment rate was 4.4 percent in May 2007 and 5.0 percent at the December 2007 peak. After the June 2009 trough it kept rising for four months and reached a high in October 2009. A threshold on unemployment can therefore flash in a different month from the activity peak the committee eventually dates.

How long recessions last, and what a depression is

On the chronology, the March 2001 recession lasted 8 months, the December 2007 recession lasted 18 months, and the February 2020 recession lasted 2 months. The summary row for 1945 to 2020 puts the average contraction at 10.3 months. That average describes past US dates. It is not a prediction.

The return to the old peak is a second clock. The FAQ gives two measures after the June 2009 trough. Real personal income less transfers did not exceed the level at the previous peak until July 2011. Nonfarm payroll employment did not exceed the previous peak until May 2014. The recession had already ended in June 2009. The level of activity stayed below the old peak for years after the trough month.

The NBER does not put depressions on the chronology as their own category. The period from peak to trough is a recession. The period from trough to peak is an expansion. The word depression is ordinary language for a particularly severe stretch of weakness. The FAQ says the latest US episode generally regarded as a depression was in the 1930s. The dates are a peak in August 1929 and a trough in March 1933, a contraction of 43 months, and a later peak in May 1937 and trough in June 1938, a contraction of 13 months. The FAQ says the contraction that started in 1929 is widely acknowledged to have been the worst in US history. There is no second official stamp to apply the next time activity falls.

Nearby measures that do not date a recession

An inverted yield curve is a pattern in interest rates. It is not a committee decision. The yield curve explainer sets out why inversions attract recession headlines, and why a spread is not a start date. The yield curve spread calculator subtracts two yields you type. It does not download a live curve, and it does not decide whether a recession has started.

The output gap is the percent difference between actual GDP and an estimate of potential output. A recession often leaves activity short of that estimate, which shows up as a negative gap. Potential output depends on the model, and it is not a published receipt in the way a GDP release is. The output gap calculator does the percent from numbers you type.

Okun's law is a rule of thumb from a GDP growth gap to a change in unemployment. It can describe how a downturn shows up in the labour market. It does not replace NBER dating or the UK two-quarter convention. The Okun's law calculator is that arithmetic, using the coefficient you choose.

The Phillips curve is the short run link between unemployment and inflation. Rising unemployment is one thing people watch in a downturn. The curve does not say whether the NBER or the ONS will use the word recession. Wider maps are the GDP and unemployment topics. The glossary entry for recession is the short version. The data guides explain how to read a release. They are not a substitute for the NBER chronology, a BEA or ONS bulletin, or the FRED series named above.

Common questions

How is a recession determined?

In the United States, the NBER Business Cycle Dating Committee looks for a significant decline in economic activity that is spread across the economy and lasts more than a few months. Depth, diffusion, and duration each have to be met to some degree, and an extreme reading on one can partly offset a weaker reading on another. Monthly dates use real personal income less transfers, nonfarm payroll employment, household survey employment, real personal consumption expenditures, manufacturing and trade sales adjusted for price changes, and industrial production. In recent decades the committee has put the most weight on real personal income less transfers and nonfarm payroll employment. Quarterly dates also use real GDP and real GDI, with equal weight. In the United Kingdom, the ONS acknowledges a technical recession when output contracts for two or more consecutive quarters. There is no UK committee that publishes peak and trough months. This page is educational only. It is not investment advice.

Who decides whether the United States is in a recession?

The NBER Business Cycle Dating Committee maintains the US chronology of peak and trough months. It is part of the NBER, not a federal statistical agency. The committee FAQ says the US government does not compile or publish an alternative business cycle chronology. The committee waits until it is confident a turning point occurred and until routine data revisions have had time to arrive. There is no fixed wait. In the FAQ table, the shortest gap was four months, from the February 2020 peak to the announcement on 8 June 2020. The longest was 21 months, from the March 1991 trough to the announcement in December 1992. The United Kingdom has no equivalent committee. The ONS says so in its methodology note on communicating the UK economic cycle, revised on 11 November 2022.

Is two quarters of negative GDP a recession?

Two or more consecutive quarters of contracting real GDP are the technical convention the ONS acknowledges for the United Kingdom. They are not the US definition. The NBER FAQ says most recessions it identifies include two or more consecutive quarters of declining real GDP, but the 2001 recession did not. A small decline can also fail the test of a significant, widespread fall. In 2022, the BEA third estimate of 29 September 2022 showed real GDP decreasing at an annual rate of 1.6 percent in the first quarter and 0.6 percent in the second quarter, while real GDI increased by 0.8 percent and then 0.1 percent. The NBER did not add a recession. Its latest peak is still February 2020 and its latest trough is still April 2020. The BEA 2024 annual update revised the second quarter of 2022 from a decrease of 0.6 percent to an increase of 0.3 percent. On FRED series A191RL1Q225SBEA, retrieved on 8 October 2026, the first quarter of 2022 is a 1.0 percent decrease and the second quarter is a 0.6 percent increase, at a seasonally adjusted annual rate.

What is the Sahm rule?

The Sahm rule is a real time US unemployment signal, not an NBER announcement. FRED series SAHMREALTIME says it signals the start of a recession when the three-month moving average of the national unemployment rate rises by 0.50 percentage points or more relative to the minimum of the three-month averages from the previous 12 months. It uses the unemployment rate available in that month. On the series retrieved on 8 October 2026, the July 2024 observation is 0.53 percentage points, which meets the 0.50 line. The NBER chronology still has no peak after February 2020. The September 2026 observation is 0.00 percentage points, in the update dated 2 October 2026. Crossing the line is not investment advice.

How long do recessions last?

On the NBER chronology, a recession starts the month after the peak and ends in the trough month. The March 2001 recession lasted 8 months, the December 2007 to June 2009 recession lasted 18 months, and the February 2020 to April 2020 recession lasted 2 months. For 1945 to 2020, the chronology's average contraction is 10.3 months. Most recessions are brief. Returning to the previous peak level of activity can take much longer. After the June 2009 trough, real personal income less transfers did not exceed its previous peak until July 2011, and nonfarm payroll employment did not exceed its previous peak until May 2014.

What is the difference between a recession and a depression?

The NBER does not identify depressions as a separate category. A recession is the period from a peak to a trough, and an expansion is the period from a trough to a peak. The word depression is commonly used for a particularly severe episode. The FAQ says the contraction from the August 1929 peak to the March 1933 trough is widely acknowledged to have been the worst in US history. That contraction lasted 43 months. A later contraction ran from May 1937 to June 1938 and lasted 13 months. Neither date is a template for calling a future downturn a depression.

Sources

• NBER, Business Cycle Dating
• NBER, Business Cycle Dating Procedure: Frequently Asked Questions
• NBER, US Business Cycle Expansions and Contractions
• BEA, Gross Domestic Product (Third Estimate), second quarter of 2022, released 29 September 2022
• BEA, 2024 Annual Update of the National Economic Accounts, summary of results
• FRED, Real Gross Domestic Product, percent change from preceding period (A191RL1Q225SBEA)
• FRED, Real-time Sahm Rule Recession Indicator (SAHMREALTIME)
• ONS, Communicating the UK Economic Cycle
• ONS, Gross domestic product (GDP)
• BEA, Gross Domestic Product