A recession is a broad, lasting drop in economic activity — not a bad week in the stock market, and not one weak jobs report. In the United States, no government agency declares one. The call belongs to the National Bureau of Economic Research, a private nonprofit, and it usually arrives months after the downturn has already begun.
The confusion matters now because the signals are mixed. Bankruptcy filings topped 600,000 in the 12 months through June, up 12% from the prior year, per the US Courts office — the highest since the pandemic. Yet economists at the Atlanta Fed estimate GDP grew about 4% in the third quarter. Both can be true at once, and both feed the recession debate.
So economy news coverage often gets this wrong: "recession" is a description applied after the fact, not a threshold that trips an alarm. Here is what the word actually means, which numbers the people who decide it watch, and why the timing is always contested.
What is a recession, exactly?
The most quoted rule of thumb — two straight quarters of falling GDP — is not the official definition. The NBER describes a recession as a significant decline in economic activity spread across the economy, lasting more than a few months, visible in output, employment, real income and industrial production, according to The Economic Times' definition of the term, which summarizes the NBER framework.
Two words carry the weight: "broad" and "lasting." A decline confined to one sector, however painful, does not qualify. A one-month dip does not either. The definition is deliberately vague because the economy does not move in clean quarters — it slides in some measures while holding up in others.
The practical effect: a recession is a judgment, not a calculation. The same data can support different calls, which is why the word gets argued about even when the numbers are agreed on.
Who decides, and how late do they decide?
The NBER's Business Cycle Dating Committee makes the call, usually with a long lag. The committee waits for data to be revised and confirmed before dating a peak — the point where expansion turns to contraction. That means the official start of a recession is often announced six months to a year after it began.
This is not bureaucratic caution for its own sake. Early reads get revised. A jobs number published in one month can look different three revisions later, and the committee would rather be right than fast. As The Economic Times notes, recessions are often recognized only after they have happened, because some signs are unclear at the start of a downturn.
For readers, the lag has a consequence: when you hear "the US is in recession" from a commentator, that person is making their own call, not citing an official one. The official call does not exist yet.
Which indicators do economists watch?
The committee leans on a small set of monthly measures:
- Non-farm payrolls — the monthly jobs count. Job losses spread across sectors are the clearest recession signal.
- Real GDP — total output, adjusted for inflation. Read our guide to why GDP reports are drafts, not verdicts.
- Industrial production — factory, mine and utility output.
- Retail sales — a direct read on consumer spending, which drives most of the economy.
- Real income — what households earn after inflation.
Outside the committee, forecasters watch earlier, rougher signals. The PMI surveys read the factory's mood early, and the yield curve signals, then it apologizes — it has flagged past recessions that never arrived.
What are the current signals saying?
The picture is split, and the split is the story. Business Insider reports that Tuomas Malinen, an economist who studies financial crises, sees the US as close to a downturn. He points to two warning signs: corporate bankruptcies, which have climbed well above their post-COVID low, and the spread between long-dated Baa-rated corporate bonds and the bank prime rate — a private-sector yield curve that has turned positive before past recessions.
On the other side of the ledger, the ISM's manufacturing New Orders Index rose to 56.7 in July, its seventh straight month of expansion, which Malinen attributes partly to data-center construction. And most Wall Street forecasters expect growth to hold. The Atlanta Fed's estimate of roughly 4% third-quarter GDP growth puts the economy far from a technical recession.
What this means: the indicators disagree because they measure different things. Bankruptcies measure financial stress on firms. New orders measure demand. An economy can carry heavy stress in one corner while another corner booms — which is precisely the pattern Malinen describes, and precisely why the official definition insists on breadth.
Why is the timing always contested?
Three reasons, in order of how much argument they cause.
Data revisions. GDP and jobs figures are estimates that change. A quarter that looked like contraction can be revised to growth, and the reverse. Anyone calling a recession on the first print is calling it on a draft.
Uneven sectors. Manufacturing can shrink while services grow. Housing can collapse while consumer spending holds. The definition requires breadth, so a severe sector-specific slump — say, a factory recession — may never meet it. Our manufacturing coverage regularly shows that gap.
Political and market stakes. A recession call moves markets and shapes elections, so interested parties reach for the label early or resist it late. Neither the early callers nor the resisters are the referee.
The honest answer to "are we in a recession?" is usually that nobody official knows yet, and the people who will eventually decide are still waiting for cleaner data.
What the record shows — and what it doesn't
What the evidence establishes: the definition is broad on purpose, the referee is a private research group that calls recessions late, and the current signals point in different directions — rising bankruptcies and bond spreads against strong new orders and a high GDP-growth estimate. What remains unknown is whether the stress showing up in firm-level data spreads to hiring and spending. That spread, or the lack of it, is what the dating committee will eventually read.
Until then, treat every recession headline as a forecast from a named person or institution, not a fact. Check who is calling it, what indicator they are reading, and how often that indicator has been wrong before.
