What is a recession?

The definition most people use is not the definition

“Two consecutive quarters of falling GDP” is the rule of thumb you will hear on the news. It is a reasonable rough signal and it is not how recessions are actually dated in the United States.

The dating is done by the Business Cycle Dating Committee of the National Bureau of Economic Research, a private non-profit, not a government agency. Its definition is a sentence, not a formula:

a significant decline in economic activity that is spread across the economy and that lasts more than a few months

Three words in that sentence do the work. Significant is depth. Spread across the economy is diffusion. More than a few months is duration. A downturn has to satisfy all three to a reasonable degree, and the committee explicitly trades them off: an unusually deep contraction can qualify even if it is short, which is how the two-month contraction of 2020 was dated a recession despite failing the two-quarters test entirely.

What the committee actually looks at

Not GDP alone, and not quarterly data as its primary input. GDP is measured quarterly and revised substantially. The committee leans on monthly measures:

• Real personal income less transfer payments
• Nonfarm payroll employment
• Real personal consumption expenditures
• Household-survey employment
• Real manufacturing and trade sales
• Industrial production

No single one decides it. The committee has said it puts the most weight on real personal income less transfers and on payroll employment, and it looks at everything together.

The announcement always comes late

This surprises people, and it is deliberate. The committee waits until the data are firm enough that it will not have to revise a call. In practice that means:

• The start of a recession is typically announced six to twelve months after it began.
• The end is often announced more than a year after the trough.

The 2020 recession is the extreme case in both directions. It began in February 2020 and the committee said so in June 2020 — fast, because the collapse was unmistakable. It ended in April 2020, and the committee announced that in July 2021, fifteen months later.

So “are we in a recession right now?” is a question nobody can answer authoritatively in the moment. Anyone who tells you with confidence is making a forecast, not reporting a determination.

What it has to do with the Fed

The Fed’s dual mandate is maximum employment and stable prices. A recession is, more or less by construction, a failure on the employment half. That does not make it an act of weather, and it does not make the usual response — cut the federal funds rate, and in severe episodes buy assets — a free lunch.

A slump is often the real adjustment after money was already too easy for too long. Credit was cheap; spending ran ahead of output; the price level rose. Tightening to restore money to output is how that sequence is closed. The unemployed pay in the labour market. Leaving money easy would send the same bill through inflation instead. See what causes inflation.

The hard case is when both halves of the mandate point in opposite directions — falling employment alongside high inflation. Cutting rates supports jobs and works against price stability; holding or raising them does the reverse. There is no setting that satisfies both. The Committee still chooses. See the dual mandate, explained.

Not every recession begins as a monetary mistake. A pandemic, a financial collapse, a war can interrupt activity on their own. The dating committee does not assign a cause. The question worth asking of the Fed is whether money accommodated the boom that preceded the slump, and whether the response is another round of the same.

What a recession is not

• Not a stock market fall. Markets price expectations and are frequently wrong about them. Several large declines have not been followed by recessions.
• Not one bad quarter. Duration is part of the definition.
• Not one bad industry. Diffusion is part of the definition. Manufacturing can contract while the economy expands.
• Not a depression. There is no official definition of a depression, but the term is reserved for contractions far deeper and longer than any post-war US recession.
• Not proof that tightening was a blunder. A boom that was financed by cheap money has a real cost when the money is withdrawn. Calling that cost a recession does not make the prior ease free.

How to read the coverage

• “We’re in a recession” in the present tense is a prediction. The determination comes later.
• “Two quarters of negative GDP” is a real signal worth knowing about, but it is not the standard, and GDP gets revised.
• “The recession is over” means output stopped falling, not that conditions feel good. The trough is the bottom, and employment usually keeps falling for months after it.
• “Recession odds are X percent” is somebody’s model. Ask whose, and what it has predicted before.
• “The Fed caused a recession by hiking” — ask what money was doing in the years before the hikes, and who was paying then.