When are we in a recession? When these 8 economists say we are.

Get the context and find out the "why" behind the stories shaping our world


When are we actually in a recession? It’s when these 8 economists say we are.

There’s something odd about the economy right now. If you look at the pace of hiring, it’s historically strong. If you look at the output of the overall economy as measured by the gross domestic product, it’s contracting. Consumers are about as negative about the economy as they’ve been in years, while manufacturing is still growing.

Hear more from Matthew Zeitlin about this story:

Perhaps the most perplexing, at least for some observers, is the difference between the economy’s output and its labor market.

According to the Bureau of Labor Statistics, the U.S. has had 18 straight months of job growth, with 372,000 jobs created in June, which would have been one of the best months of job growth in the post-2008, pre-covid period. Another “real time” indicator developed by former Federal Reserve economist Claudia Sahm, which is supposed to detect recessions in real time by looking at how quickly the unemployment rate is growing, does not indicate a recession happening now or particularly soon.

Yet reputable sources like the Commerce Department’s Bureau of Economic Analysis (BEA) and the Federal Reserve Bank of Atlanta reported that the gross domestic product shrank by more than 1 percent in the first quarter of this year.


Former Obama administration economic adviser and Harvard University economist Jason Furman has argued that this gap — between overall economic growth going down while job creation is going up — is the largest it has been since the late 1940s.

But whether this means a recession — a sustained period of economic contraction — is on the way is harder to figure out. While a popular rule of thumb for a recession is two quarters of negative growth or GDP shrinking, in the U.S. at least, it’s the judgment of a committee of economists. They look at the data certainly, but it is also a judgment call.

Grid takes a multi-lens approach on this question of a possible recession in the coming months by looking at the data, the institution in charge of designating a recession and the history of why declaring a recession is not always an easy call.

The data is confusing or perhaps even wrong

One thing that could be going on is that the data is just wrong. Economic data can be heavily revised, meaning the apparent disjunction between the two may shrink or even disappear. Revisions are particularly common — and large — when the economy is at a turning point. In the final three months of 2008, the economy shrank at a 6.2 percent annualized rate, far greater than the initial repot of 3.8 percent negative growth. In June, 2020, the Bureau of Labor Statistics reported that job losses in March were 1.4 million, not 881,000, while the massive April job loss was 150,000 great than initially reported.

“It strikes me as very hard to believe that these things are simultaneously true, a world where we’re seeing non-trivial declines in output and the labor market that adds jobs,” Guy Berger, LinkedIn’s principal economist, said. “I just don’t think we’re going to look back in five years and think this was a period where GDP contracted by a non-trivial amount and the labor market kept charging ahead.”


The survey used to measure payrolls — which is quoted in the media as the “jobs number” — showed 372,000 new jobs. But this is different than the measure used for the unemployment rate, which is based on a survey of households. This is used to measure the labor force — the portion of the population that is either working or looking for work — and then what portion of that population is unemployed. To do this, it needs to also measure employment, and the June figures showed an employment decline. It is typically jumpier than the establishment survey and is not often cited for the level of total employment.

The employer survey can also be revised substantially in the coming months, but Furman pointed out that even with fairly large revisions, it would not be at a level that indicates a recession is coming soon.

“What’s unusual about this time, people are very on edge for any divergence,” Berger said. “In a few months, we will know one way or the other which one is right. We should be patient.”

American Enterprise Institute Director of Economic Policy Studies Michael Strain insists the data is not really that hard to figure out. “This is not as unusual as everyone is making it out to be. GDP and unemployment can move in different directions,” Strain told Grid.

Specifically, he noted the “nowcasts” of GDP are not as dire as two straight negative quarters of growth would appear — the Atlanta Fed tracker shows consumer spending and business investment growing, while the notorious “change in private inventories,” is falling. This component of GDP measures how much businesses add or subtract from their physical stocks of stuff they use to make goods or to sell. This could mean the stocks of items retailers maintain to sell (like, say, sweaters at the Gap), or even the steel used to make cars. It is, the Commerce Department says, “one of the most volatile components of gross domestic product (GDP), giving it an important role in short run variations in GDP growth.” When GDP is acting “weird,” change in private inventories typically play a role.

“If you look at core components of domestic demand, get rid of inventories, all of the sudden things start to make sense,” Strain said. “Inventories are weird because we’re pulling out of the pandemic.”

But just where the economy is pointed, according to the summary statistics, is not even so clear. Another figure the BEA reports, gross gomestic income, which is theoretically supposed to match up with gross domestic product, actually grew 1.8 percent, while GDP fell 1.6 percent.

“Revisions in both GDP and GDI could largely erase that contraction in output,” Christopher Waller, a member of the Federal Reserve Board of Governors, said in a speech Thursday. “I am very cautious about acting on the GDP estimates and on projections of an output slowdown.”

While Berger and Strain both agreed the data, as of now, does not show the economy currently in recession, it does show slower growth, which can explain why different reports are a bit all over the place.

“Once you start moving from a labor market that’s growing really fast to a little slower one, more like the 2010s, some data will look good, some will look bad,” Berger said.


“It’s not this amazing aberration that everyone thinks it is,” Strain said. “We are 200,000 jobs short of where we have been [earlier this year].” It’s just that since the labor market was so hot, a 200,000-job-a-month slowdown puts the economy at 300,000 jobs created per month.

Who decides if we’re in a recession?

The National Bureau for Economic Research (NBER), an over 100-year-old nonprofit economic research institution, is the organization responsible for “dating” when a recession starts and stops.

This decision is made by an eight-member committee, appointed by the NBER. This committee is made up of macroeconomists who are well-known and eminent in the field and tend to work at elite universities. Among its members are Robert Hall, a Stanford University professor who has been on the committee since its inception; Christina and David Romer, a pair of University of California at Berkeley professors, the former of whom was the head of President Barack Obama’s Council of Economic Advisers; and Robert Gordon, one of the leading scholars on the business cycle and economic growth who has been on the committee with Hall since 1978.

The committee is supposed to take in a range of economic data including income, employment, consumer spending, retail sales and industrial production and then, looking backward, determine when there has been a “significant decline in economic activity that is spread across the economy and lasts more than a few months,” and be able to define contractions and expansions by identifying the peaks and troughs of the economy.

But none of these guidelines are ironclad; for example, the committee determined that in 2020, the economy peaked in February and bottomed out in April, a contraction that lasted all of two months. That announcement came in June and was relatively easy, former committee member and Harvard University economist Jeffrey Frankel explained in a column at the time, “looking at the numbers gave the same answer as ‘looking out the window.’”


Usually things don’t work so quickly. A more “normal” recession might have been in 2001, when the committee announced in November that a recession had begun that previous March, when employment had peaked and begun declining, while industrial production had peaked the previous September.

The committee noted “the determination of the date of the peak in economic activity was as challenging as usual,” and that “major indicators peak in different months.”

But even by then, there hadn’t yet been a reported figure showing two consecutive quarters of negative GDP growth. The committee pointed out in a memo that only in July 2003 had the final figures showed that GDP had actually shrunk for three quarters. “The two-quarter-decline rule of thumb would not have allowed the declaration of the recession until August 2002, let alone a declaration that it had begun early in 2001, as in the statement that the committee made in November 2001,” the committee said.

Historically, recessions are a judgment call

It’s possible the committee today faces the opposite problem it did in 2001 — preliminary data showing two consecutive quarters of negative growth, a common rule of thumb for a recession that is the definition in several countries, while the higher-frequency data shows something else. Typically, the GDP data comes after the other data, collected more frequently, turns.

While the NBER has identified peaks and troughs of economic activity going back to 1857, it only started doing so in its current fashion in the late 1970s after the formation of the committee. The process is largely useful for researchers who all have the same agreed-upon dates to work with when looking at past business cycles.


While in 1979 they released memos explaining why they hadn’t yet determined a recession, they made their first determination of one in June of 1980 even though they had “observed that no cyclical decline in real GNP has yet been recorded,” because the data came out quarterly. According to a history of the NBER’s business cycle dating by Christina and David Romer, the precise conflict analysts are pointing to now, where growth appears to be negative while employment is expanding, is familiar to the NBER committee.

While the NBER has a reputation for nonpartisanship and reflects the broad mainstream of the economics profession — current and past members of the committee have served in Democratic and Republican administrations, as well as in the Federal Reserve — its upcoming decisions (or nondecisions) are sure to bring on controversy, especially if the initial report of GDP growth in the second quarter is negative and a recession call is not forthcoming.

The behavior of GDP and employment has often been quite different: “An important issue the committee faces in choosing turning points is how to balance the various series when there are conflicts. This has been particularly difficult in recent years because the behavior of GDP and employment has often been quite different,” the two wrote in their history of business cycle dating.

Thanks to Alicia Benjamin for copy editing this article.

  • Matthew Zeitlin
    Matthew Zeitlin

    Domestic Economics Reporter

    Matthew Zeitlin is an economics reporter at Grid focused on the domestic impact of major stories such as coronavirus, the supply chain and economic volatility.