Identifying Recessions Is More Art Than Science

Their work is rigorous, but it’s always been as much art as science. One of the first economists to formulate the idea that booms and busts were more than disasters — in other words, that they followed a kind of predictable sequence of discrete stages — was a French statistician, Clement Juglar. His tripartite framework of prosperity, crisis and liquidation captured the wave-like movement from booms to busts and back.

Other economists working in France and Germany elaborated on these ideas. But the most consequential theorist of the business cycle — the man whose work inadvertently fostered the belief that recessions can be pinned down like butterflies in a specimen case — was the American economist Wesley Clair Mitchell.

Mitchell was an obsessive empiricist, someone who loved collecting and tabulating raw data in the hopes of finding patterns. He was particularly interested in identifying how each stage of the business cycle in modern, industrial economies paved the way for the next stage, producing a perpetual ebb and flow.

In 1913, he published the first of several massive works, “Business Cycles,” which laid out the defining features of the stages he called prosperity, crisis, depression and revival. The word “recession” rarely appeared. When it did, it simply referred to a reduction in value; a “recession in prices,” for example.

In 1920, Mitchell founded the NBER, which became the nation’s most important center for the study of business cycles. Mitchell and his colleagues approached the problem with a key insight: economic indicators tended to move in concert, rising and falling around the same time. This enabled them to identify the economy’s peaks and troughs.

The economists Christina and David Romer have shown that these labors, though commendably data-driven, nonetheless rested on subjective criteria that changed over time. Different data series came and went in dating business cycles; so, too, did the relative importance of each data point. Much depended on what the Romers described as a “hodge-podge” of single-series indicators such as pig iron production and freight car loadings.

In 1946, Mitchell and his collaborator, the economist Arthur Burns, published “Measuring Business Cycles,” another tome on the subject. Despite working at the problem for decades, Mitchell’s methods for dating the peaks and troughs of the business cycle remained maddeningly imprecise. The Romers describe how several critical passages in the book “seem to be arguing for a somewhat judgmental approach that does not assign fixed weights to various series.”

Reviewers of the work zeroed in on these alleged shortcomings. In 1947, the economist Tjalling Koopmans lamented that Mitchell and Burns had “deliberately spurned” the “toolkit of the theoretical economist,” never invoking a single equation in support of their approach. Koopmans assailed the “pedestrian character of the statistical devices employed,” summarizing the book with a damning assessment: “measurement without theory.”

This wasn’t the only problem. The nomenclature used to describe the various stages of the business cycle remained equally vague. Mitchell and Burns explained that their “working definition” of the business cycle consisted of four sequential phases: expansion, recession, contraction and revival. But beyond this schema, the word “recession” was left undefined.

It was around this time, though, that the word began replacing “depression” as a kind of catch-all term for any sustained economic downturn. This bothered economists like Burns. In 1960, when he waded into a debate over the direction of the economy, Burns declared that “terms such as ‘recession,’ ‘adjustment,’ ‘lull’ and the like do not have any recognized scientific meaning.”

Burns grudgingly acknowledged that the term “recession” might be used to describe “a decline of aggregate economic activity which is (a) of moderate size, (b) fairly widespread, and © lasts from about eight months to a year or a little longer.”

Moderate? Fairly? What did these words actually mean? For its part, the NBER continued to shy away from using the word “recession,” focusing instead on identifying the peaks and troughs of economic life. But the financial press began referring to the period from peak to trough as a recession, and the NBER eventually followed suit.

In 1978, the NBER created the Business Cycle Dating Committee, appointing Stanford economist Robert Hall as its head. (Hall remains in place today.) In 1980, as the economy struggled with economic malaise, Hall tried to warn the Wall Street Journal that his group didn’t offer easy answers. “We’ve been criticized for not having a formula,” he warned. “But there isn’t any way to teach a computer to define recessions.”

Indeed. The challenge of calling a recession continued to depend on the sort of analysis normally associated with intelligence agencies, where disparate data is weighed and sifted in a laborious fashion, and where qualitative judgments count as much as quantitative ones. This was increasingly out of step with the pretensions of theoretical economics, which preferred to begin with equations, only using data to prove a foregone conclusion.

Yet the NBER remained faithful to Mitchell’s intuitive, data-driven approach. A press release from 2001 defined a recession as a “significant decline in activity spread across the economy, lasting more than a few months, visible in industrial production, employment, real income, and wholesale-retail trade.” Likewise, the NBER’s current website lists a comparable definition, citing a “significant decline in economic activity.”

As Romer and Romer point out, the latest NBER definition is actually “less precise than the 1946 definition about what constitutes a ‘significant’ decline.” No matter. Arthur Hall and the other members of the committee – Christina Romer is one of them – will ultimately decide whether today’s economic woes qualify.

On one level this seems outrageous. Why should a little-known group of eight economists get to date the business cycle, much less define the scope of national suffering? There must be some econometric model – an algorithm, please! – that would take the guesswork out of this business.

That isn’t going to happen, and that’s a good thing. The method that Wesley Clair Mitchell and his associates pioneered doesn’t reduce reality to theories and equations. Rather, it embraces the inescapable messiness of human economic existence, imposing order only after careful deliberation.

The results aren’t perfect. But they try to do justice to human experience, leaving mathematical models in their own tidy world.

Bloomberg News provided this article. For more articles like this please visit bloomberg.com.

Read more articles by Stephen Mihm