Given the size and economic importance of the US, the world is watching where the US economy is going. It peaked--and the 1990s boom came to an end--around March 2001, when America's remarkable economic expansion gave way to the first recession in ten years. But did that recession end when output started to grow in December 2001? Or is it still continuing? The National Bureau of Economic Research (NBER) and its Business Cycle Dating Committee, the semi-official arbiter and tracker of the US business cycle, remain silent. Recent economic and monetary policy changes have exposed a crucial ambiguity that had always been implicit in the way the NBER thought about business cycles.
Since the Great Depression, it was almost always clear when a recession ended: industrial production grew strongly, total sales reversed their decline, and the unemployment rate fell. All of these trend reversals occurred simultaneously or within a very few months of each other. Thus, it never really mattered how one identified the end of a recession, or even whether one had a definition at all. As a Supreme Court justice once said of obscenity, "I may not know how to define it, but I know it when I see it."
Recently, however, this rough-and-ready approach has begun to prove inadequate. In retrospect, the first warning sign was the so-called "jobless recovery" of the early 1990s. Production and sales bottomed out in March 1991. But the unemployment rate continued to rise--by more than a full percentage point before peaking at 7.6% in June 1992. If, as I believe, the most important business-cycle indicator is workers' justified anxiety about losing a job and the difficulty of finding a new one, then the worst cyclical moment for the American economy came a full fifteen months after the recession's semi-formal end.
And things are worse this time. As measured by trends in production, the recession that began in March 2001 was one of the shortest and shallowest ever: over in less than nine months, and amounting to an extremely small decline in gross domestic product. The same appears true when measured by sales.
But, as measured by employment, this is one of the worst, if not the worst, recession since the Great Depression: 2.1 million fewer people are at work in the US today than at the peak of the business cycle two years ago. If one includes normal growth of the labor force, the employment shortfall today relative to what it would have been had the 1990's boom continued amounts to 4.7 million jobs.
So why the breakdown of the old business-cycle indicators? One reason is that the Federal Reserve has acted differently in the past decade and a half. Most earlier recessions were not entirely undesired: higher unemployment helped reduce inflation when it exceeded the Fed's comfort level. When the Fed concluded that inflation was no longer a threat and shifted its primary short-term task from ensuring price stability to boosting production, most earlier recessions ended. Lower interest rates caused all business cycle indicators--production, sales, employment, and the unemployment rate--to turn upward.
A second reason is that roughly from 1970-1995, the underlying trend of productivity growth was relatively slow. When productivity grows slowly, it is extremely unlikely that rising production will be accompanied by falling employment. Since 1995 or so, however, the
productivity growth trend underlying the US economy has been quite rapid: not 0.7% per year--the average during the preceding 25 years--but 2% or 3%, perhaps even more. And so far, there is every sign that rapid underlying potential productivity growth persists.
With these two considerations in mind, the NBER's current dilemma becomes obvious. Unlike in previous recessions, this time there was no sudden shift in Fed policy to give all macroeconomic business cycle indicators the same turning point. Moreover, rapid growth in underlying productivity means that a respectable, if subnormal, recovery in terms of output growth is associated with falling employment and a rising unemployment rate.
The lesson is that if we are to continue to use the word "recession," we finally have to define exactly what one is. Is a recession a period of falling output alone? Or is a recession a period when the labor market becomes worse for a typical worker? Whether the US economy is still in recession depends on how we answer this question.
The most important point is not whether the US economy is in recession, but that the old categories simply do not fit. The US is still in an employment recession; but in the past, employment recessions were accompanied by falling output, which is not the case now. The US is in a production recovery; but in the past (except for 1991-1992), production recoveries were accompanied by improving labor markets, which is not the case now.