The Official NBER Definition of a Recession
The National Bureau of Economic Research's Business Cycle Dating Committee is the official arbiter of US recessions. Here is exactly how they decide, who decides it, and why the announcement comes months or years after the fact.
Who Declares a Recession?
In the United States, recessions are officially dated by the NBER Business Cycle Dating Committee, a committee of academic economists convened under the National Bureau of Economic Research, a private nonprofit research organisation based in Cambridge, Massachusetts. The NBER was founded in 1920; the Business Cycle Dating Committee was formally constituted in 1978, though NBER had been publishing business cycle analyses since 1929.
The committee consists of eight economists, typically senior academic macroeconomists from leading US universities. It has no formal government affiliation, but its recession dates are universally accepted by the Federal Reserve, the Bureau of Economic Analysis, the Congressional Budget Office, and financial institutions worldwide. When a journalist, policymaker, or textbook says “the recession began in December 2007,” they are citing NBER.
The committee makes two separate types of announcement: (1) the declaration of a peak in economic activity (the start of a recession) and (2) the declaration of a trough (the end of a recession). Both are made retrospectively, after sufficient data has accumulated.
The NBER Definition: Three Requirements
NBER's formal definition of a recession has three components, each of which must be present:
The decline must be significant - a minor statistical wobble in one indicator does not qualify. NBER looks for a broad-based decline visible in multiple indicators simultaneously.
The decline must spread across the economy. A contraction confined to one sector (say, manufacturing while services are booming) does not constitute a recession under NBER's definition.
The decline must last more than a few months. Extremely brief contractions (2-3 months) typically do not qualify unless the depth is extraordinary - as in the COVID recession of 2020, where the sharpness was historic.
“A recession is a significant decline in economic activity that is spread across the economy and that lasts more than a few months, normally visible in production, employment, real income, and other indicators.”
NBER Business Cycle Dating Committee
The Six Monthly Indicators
The committee primarily uses six monthly economic indicators to identify turning points. Notably, quarterly real GDP is not one of the primary indicators - it is quarterly, subject to significant revision, and is treated as a supplementary cross-check. The six monthly indicators are:
| Indicator | Source | Weight | April 2026 Status |
|---|---|---|---|
| Real personal income less transfers | BEA | Primary | Rising 0.8% year-over-year |
| Nonfarm payroll employment | BLS | Primary | 214k jobs/month (3m avg) |
| Real personal consumption expenditures | BEA | Primary | Rising 1.4% year-over-year |
| Wholesale-retail sales (inflation-adjusted) | Census | Secondary | Flat to slightly positive |
| Household employment (CPS survey) | BLS | Secondary | Broadly consistent with payrolls |
| Industrial production | Federal Reserve | Secondary | Negative year-over-year |
Why GDP Is Not the Primary Indicator
Real GDP is the most widely cited economic statistic, but NBER explicitly treats it as supplementary rather than primary. There are three reasons:
- Quarterly frequency: GDP is measured quarterly. Monthly indicators provide turning-point identification four times more frequently, allowing earlier and more precise peak/trough dating.
- Revision magnitude: GDP is heavily revised, sometimes by more than 1 percentage point, in subsequent releases. A contraction that looks definitive in the advance estimate may disappear after two revisions. Monthly indicators are revised less dramatically.
- Single-indicator problem: A recession is definitionally a broad-based decline. Using only GDP would miss cases where output is temporarily negative but employment, income, and spending are strong (as in 2022 H1), or where output is slightly positive but employment is collapsing.
How the Announcement Timing Works
The committee operates on a deliberate delay. Once a recession is suspected, the committee waits for enough monthly data to confirm the turning point - typically 6-12 months for the peak announcement and 12-21 months for the trough. This delay is intentional: it prevents the committee from declaring and then retracting recessions as preliminary data revises.
| Recession | Actual Peak | Peak Announced | Peak Delay | Trough Announced | Trough Delay |
|---|---|---|---|---|---|
| 2007-09 Great Recession | Dec 2007 | Dec 2008 | 12 mo | Sep 2010 | 15 mo |
| 2001 Recession | Mar 2001 | Nov 2001 | 8 mo | Jul 2003 | 20 mo |
| 1990-91 Recession | Jul 1990 | Apr 1991 | 9 mo | Dec 1992 | 21 mo |
| 2020 COVID Recession | Feb 2020 | Jun 2020 | 4 mo | Jul 2021 | 15 mo |
| 1981-82 Recession | Jul 1981 | Jan 1982 | 6 mo | Jul 1983 | 8 mo |
Why the Delay?
The delay serves the committee's core purpose: authoritative, not provisional, dating. An early announcement that later requires retraction would damage the credibility that makes NBER dates the global standard. The committee also needs to see the full profile of the decline to confirm depth, diffusion, and duration - all three criteria must be met, and that requires observing several months of data beyond the suspected turning point.
The 2020 COVID recession represents the fastest peak announcement in NBER history (4 months), justified by the historically unprecedented speed and magnitude of the contraction. The trough announcement (July 2021, 15 months after the April 2020 trough) was faster than average for the same reason - the V-shaped recovery was equally unambiguous.
Can the Committee Be Wrong?
No NBER recession has ever been formally retracted, but initial dating is sometimes refined. The committee revises peak and trough months when additional data clarifies the precise turning point. The 2020 trough, initially assessed as possibly June 2020, was confirmed as April 2020 by the July 2021 announcement. The committee is transparent about its methodology and publishes detailed explanations of each determination.
Critics occasionally note that NBER's qualitative committee judgment is less reproducible than a mechanical rule, and that the delay makes NBER dates useless for real-time policy response. The Sahm rule, developed specifically to address this limitation, provides a systematic real-time recession signal based solely on unemployment data - though it too has acknowledged edge cases.
Frequently Asked Questions
Who is on the NBER Business Cycle Dating Committee?
The NBER Business Cycle Dating Committee consists of eight academic economists, typically including macroeconomists from leading US universities. The committee has included prominent figures such as Robert Hall of Stanford (long-serving chair), Martin Feldstein (Harvard), Robert Gordon (Northwestern), and Christina Romer (Berkeley). Membership rotates and the current composition as of April 2026 reflects the committee's tradition of selecting economists with deep expertise in business cycle research and macroeconomic measurement.
When was the last NBER recession announcement?
The last NBER Business Cycle Dating Committee announcements were for the COVID-19 recession. The committee announced in June 2020 that a recession had begun in February 2020, and then announced in July 2021 that the recession had ended in April 2020. This was notable for the speed of both announcements - the peak was announced just 4 months after it occurred, and the trough (April 2020) was announced 15 months later, faster than average due to the extreme sharpness of the contraction and recovery.
Why doesn't NBER use the two-quarter GDP rule?
NBER explicitly rejects the two-quarter GDP rule because GDP is quarterly, subject to heavy revision, and is only one of the six monthly indicators the committee watches. The committee's definition requires a decline to be significant in depth, spread across the economy (not just one sector or one data series), and lasting more than a few months. The 2001 recession was declared without two consecutive negative GDP quarters. The 2022 H1 GDP contraction was not declared a recession because employment was growing strongly throughout.
Does anyone challenge NBER's recession dating?
NBER's dating is widely accepted as authoritative, but it is occasionally critiqued for being too slow (6-18 month delay makes it less useful for real-time policy) and too opaque (the committee does not publish a quantitative formula, only a qualitative judgment). Some economists use alternative methods for real-time dating, such as the Sahm rule (which uses unemployment data) or nowcasting models. No alternative dating body has achieved comparable authority or acceptance among US government agencies, central banks, or financial institutions.
Related Pages
The Two-Quarter Rule: Why It Is Not the Official DefinitionLive Recession Indicators: Current April 2026 ReadingsIs the US in a Recession in 2026?Complete NBER Recession History 1854-2026For a companion guide on the most important macroeconomic indicator that NBER uses as a cross-check, see whatisgdp.com.