In March 1929, the National Bureau of Economic Research published its first estimate of the start and end dates of what it called “contractions” or “recessions,” going back to 1855. The private think tank, then based in New York, continued to assign dates to peaks and troughs of the U.S. business cycle in subsequent years, while the NBER researcher who compiled the first set went on to design a system of national income accounts for the Commerce Department. In 1942 Commerce supplemented these for purposes of better understanding the wartime economy with a measure called gross national product, with regular quarterly reporting of GNP beginning in 1947.

If you’ve been wondering why recessions in the U.S. are more or less officially determined by a panel of economists at NBER and not, as in many countries, by the simple rule-of-thumb of two consecutive quarters of declining real gross domestic product (which supplanted GNP as the headline economic growth metric in 1991 in the U.S. and well before then elsewhere), here’s the reason: Some guy at NBER was determining when recessions started and ended before GDP existed, after which he went on to help invent GDP.

The NBER is thus largely responsible for the modern understanding of and nomenclature surrounding the business cycle, as well as some of the measurement apparatus. It may even be partly responsible for the strong political salience of recessions in the U.S., where they have often accompanied or immediately preceded changes in party control of the White House (in contrast to say, Germany, where incumbent Chancellor Angela Merkel won resoundingly in the wake of the Great Recession in 2009). That doesn’t mean its recession determinations are right, especially since the business cycle ceased to be the focus of its research decades ago. But the organization’s pioneering role in determining what recessions are does seem relevant to the discussion over whether we’re in one now.

The NBER was founded in 1920 by a conservative AT&T Inc. executive and a liberal Hickey-Freeman Co. executive who’d been a Socialist Labor Party activist in his younger days. They hoped that better data could resolve some of their economic disputes. The first research director they selected was Wesley Clair Mitchell, a Columbia University economist who bore the external markings of a lefty intellectual—he had been the star student of famed capitalism-critic Thorstein Veblen, lived in Greenwich Village and was married to a famed progressive educator—but was resolutely neutral in his academic work, which centered around the business cycle.

Mitchell offered no grand theories, just the hope that accumulating more and better data about the economy’s ups and downs could increase understanding. At NBER he was able to put smart young researchers to work on this, notably Simon Kuznets, who had emigrated to the U.S. from Ukraine in 1922 and immediately distinguished himself as Mitchell’s student at Columbia. It was Kuznets who, near the beginning of a 34-year affiliation with NBER, compiled a list of recession dates under Mitchell’s supervision in 1929 and then, at the behest of the Commerce Department in the 1930s, devised the system of national income accounts that, with a nudge from innovators across the Atlantic, eventually spawned GDP.

For the latter work, Kuznets received the third Nobel Prize awarded in economics, in 1971. Meanwhile, Mitchell’s handpicked successor at NBER, Arthur Burns, went on to great prominence as an adviser to Republican presidents and then chairman of the Federal Reserve. Milton Friedman and Anna Schwartz’s influential 1963 “Monetary History of the United States” was an NBER project.

The organization’s empiricist obsession with the business cycle began to fall out of favor among academic economists after World War II. In a famously withering review of Burns and Mitchell’s 1946 book, “Measuring Business Cycles,” mathematical economist Tjalling Koopmans (winner of the economics Nobel in 1975) complained that in it “the movements of economic variables are studied as if they were eruptions of a mysterious volcano whose boiling caldron can never be penetrated.”

He had a point! It’s not just that the why of the business cycle is mostly ignored in the book, the what isn’t exactly made crystal clear either. “Our definition presents business cycles as a consensus among expansions in ‘many’ economic activities, followed by ‘similarly general’ recessions, contractions, and revivals,” Burns and Mitchell write. “How ‘general’ these movements are, what types of activity share in them and what do not, how the consensus differs from one cyclical phase to another, and from one business cycle to the next, can be learned only by empirical observation.” Got that?

At one point the pair do suggest that the “simplest method” of determining the timing of a business cycle “would be to mark off the months in which the specific cycles of an acceptable measure of aggregate economic activity reached successive peaks and troughs,” and note that quarterly and even monthly estimates of just such a measure, GNP, were under development. We’re still waiting on official monthly GNP or GDP numbers, though, and the quarterly ones are subject to revisions years and even decades afterwards that can flip positive quarterly changes to negative and vice versa.

The NBER determinations of recession dates, while they don’t ignore GDP, have thus tended to focus on more-timely and less-prone-to-revision monthly data series. An ever-shifting array of data series, it turns out. As University of California at Berkeley economists Christina Romer and David Romer described in a historical review presented at the January 2020 annual meeting of the American Economic Association (video here, pdf download here), the unemployment rate has gone in and out of favor as a recession metric while NBER researchers have through the decades cited such measures as bank debits outside New York City, freight-car loadings, business failures, corporate profits after taxes and imports, as well as more obvious ones like nonfarm payroll employment, industrial production and personal income.

This hodge-podge of justifications didn’t seem to diminish the impact of the NBER’s recession determinations, with Harvard economist Jason Furman recently dredging up an array of media mentions from the 1960s and 1970s that make clear that its verdicts—in those days often the work of a single staffer, Geoffrey Moore—were seen as definitive.

When Harvard’s Martin Feldstein became NBER’s president in 1977, he moved the organization’s headquarters to Cambridge, Massachusetts, and repositioned it from a focus on business cycles and a few other macroeconomic topics to policy-relevant academic research on all manner of economic questions (three-quarters of its $34 million budget in the 2020-2021 fiscal year came from government research grants, with the U.S. Department of Health and Human Services and National Science Foundation the chief funders and the GDP-measurers at the Commerce Department not represented at all). But Feldstein, who took a leave from NBER and Harvard to serve as chairman of President Ronald Reagan’s Council of Economic Advisers from 1982 to 1984 and remained a leading Republican economic sage until his death in 2019, was publicity-savvy enough to retain the organization’s responsibility for determining recessions. The Business Cycle Dating Committee that he created in 1978, composed mostly of outside economists, has if anything attracted more attention than its predecessors, in part because it announces its decisions in a consistent way while earlier recession calls were sometimes tucked in the back of NBER publications.

Right now the committee is holding fire despite two consecutive quarters of shrinking GDP, partly because it always waits a while before making a recession call but also because, as several current and former members made clear to Bloomberg’s Steve Matthews last week, the negative GDP readings are so far at odds with most key monthly economic indicators and those indicators carry more weight.

By now I think it should be clear that this reticence to declare a recession is not the result of some plot organized by the Biden administration and the liberal media. But yeah, it is kind of confusing and possibly unnecessary. As Christina and David Romer—who happen both to be members of the Business Cycle Dating Committee—put it in their January 2020 paper:
If a modern time-series econometrician with no knowledge of the history of business cycle dating or the concept of a recession were handed the time series for postwar U.S. GDP growth, they would conclude that the data pointed to a division of short-run fluctuations into two types of periods that correspond closely to the recessions and expansions identified by NBER researchers over the past century.

That speaks well for the NBER researchers, but it also raises the question of why we can’t just leave the work to an econometric model. The Romers point to a much-cited 1989 paper by University of California at San Diego economist James D. Hamilton that used a statistical technique known as Markov switching to differentiate between two different regimes of GNP growth in the U.S., identifying periods of negative growth that pretty much coincided with the NBER recession dates. In an August 2020 revision of their paper, the Romers updated the model with GDP numbers through 2019 with similar results and got an even better fit by adding nonfarm payroll employment and the unemployment rate to the mix.

They also proposed an approach that aims to identify periods of economic slack rather than decline, in part because periods of negative growth will become more common in developed countries as population growth slows or turns negative, while if Japan’s experience so far is any guide these won’t always be accompanied by rising unemployment or other signs of distress. In this case, GDP growth, employment and unemployment are compared to trend rates estimated by the Congressional Budget Office.

I’ve asked the Romers for the current recession probabilities churned out by their models, and will update here if I get them. But with employment growth still strong as of June and unemployment nearly a percentage point below the CBO’s noncyclical (formerly known as natural) rate, any model that includes those metrics is unlikely to conclude that the U.S. is in a recession—yet.

Justin Fox is a Bloomberg Opinion columnist covering business. A former editorial director of Harvard Business Review, he has written for Time, Fortune and American Banker. He is author of The Myth of the Rational Market.