The NBER’s Recession Dating Procedure

Robert Hall, Chair
Martin Feldstein, President, NBER
Ben Bernanke
Jeffrey Frankel
Robert Gordon
Victor Zarnowitz

This memo will appear monthly on the NBER's website, NBER.org, during the period of uncertainty about the state of the economy. The purpose of the memo is to explain the Bureau's procedures for dating recessions, not to indicate whether or not the economy is in a recession. That determination is made only by the Bureau's Business Cycle Dating Committee, which has not yet met. The committee does not determine the start or end of a recession until it has at least six months of data beyond the peak date, and often even more time is required to make the determination

A recession is 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. A recession begins just after the economy reaches a peak of output and employment and ends as the economy reaches its trough. Between trough and peak, the economy is in an expansion. Expansion is the normal state of the economy; most recessions are brief and they have been rare in recent decades.

Because a recession influences the economy broadly and is not confined to one sector, the committee emphasizes economy-wide measures of economic activity. In principle, the best such measure is real gross domestic product, GDP. But GDP is measured only at a quarterly frequency and is continually revised, often decades later. The traditional role of the committee is to maintain a monthly chronology, so the committee refers almost exclusively to monthly indicators.

The broadest monthly indicator is employment in the entire economy. Figure 1 shows the movements of total non-farm employment from the payroll survey, averaged over the past 6 recessions (1960-61, 1969-70, 1973-75, 1980-81, 1981-82, and 1990-91). The figure shows the 36 months surrounding the peak month determined by the NBER. The value in the peak month is standardized at 1.00. Employment generally declines about 1.1 percent from the peak to the trough

The committee generally also studies another monthly indicator with economy-wide coverage, personal income less transfer payments, in real (inflation-adjusted) terms. In addition, the committee refers to two indicators with coverage of manufacturing and goods: real manufacturing and trade sales and industrial production. The Bureau of Economic Analysis of the Commerce Department compiles the first and the Federal Reserve Board the second. Because manufacturing has become a relatively small part of the economy, the movements of these indicators may differ from those reflecting other sectors

Figure 2 shows the Fed's Index of Industrial Production averaged over the past 6 recessions. Notice that the peak in industrial production occurs two months before than the peak date determined by the Bureau for the overall economy, on the average.

Although the four indicators described above are the most important measures considered by the NBER in developing its business cycle chronology, there is no fixed rule about which other measures contribute information to the process

A recession involves a substantial decline in output and employment. In the average recession, industrial production fell by 4.6 percent and employment by 1.1 percent. A small decline in output not matched by a corresponding decline in employment would not enter the Bureau's chronology as a recession. As a result, the Bureau waits until the data show whether or not a decline is large enough to qualify as a recession before declaring that a turning point in the economy is a true peak marking the onset of a recession. The Bureau announced in April 1991 its determination that July 1990 was a peak, and in December 1992 announced its determination that March 1991 was a trough. The particularly long lag for the trough date resulted from the slow pace of growth in 1991 and 1992-had the economy reversed course and declined below its March 1991 level during that period, the period would have been counted as a single longer recession. The Bureau could not set a date for the trough until, in late 1992, the economy had regained its July 1990 peak and a subsequent contraction would have been considered a separate recession.

The Current Economy

Figure 3 shows the recent movements of employment superimposed on the average shown above in Figure 1. Employment reached a peak in March 2001 and declined since then. The figure for October is the first to reflect the effects of the attacks of September 11. For the purpose of illustration, we have aligned the recent data with March 2001 as a possible peak, though the committee has not yet made any determination of a peak. Through October, the decline in employment has been similar to the average over the first 7 months of recessions. The cumulative decline is now about 0.7 percent, about two-thirds of the total decline in the average recession

Figure 4 shows the movements of real personal income less transfers. Though the rate of growth has slowed somewhat, there is no sign of a decline through September 2001. Figure 5 shows real manufacturing and trade sales. This measure reached a peak almost a year ago, moved downward for several months, but moved upward in July and August. Figure 6 shows industrial production. A peak occurred in September 2000 and the index declined over the next 12 months by close to 6 percent, surpassing the average decline in the earlier recessions of 4.6 percent.

The data continue to show substantial declines in real activity in manufacturing, the sector reflected in the industrial production index and in real manufacturing and trade sales. Aggregate employment has fallen substantially as well, by about two-thirds of the amount typical of earlier recessions. Among the four indicators, only income has behaved differently from recession averages over the past 7 months.

For more information, see the FAQs at the end of this memo, and also see http://www.nber.org/cycles.html.

Q: The financial press often states the definition of a recession as two consecutive quarters of decline in real GDP. How does that relate to the NBER’s recession dating procedure?

A: Most of the recessions identified by our procedures do consist of two or more quarters of declining real GDP, but not all of them. But our procedure differs in a number of ways. First, we use monthly indicators to arrive at a monthly chronology. Second, we use indicators subject to much less frequent revision. Third, we consider the depth of the decline in economic activity. Recall that our definition includes the phrase, “a significant decline in activity.”

Q: Isn’t a recession a period of diminished economic activity?

A: It’s more accurate to say that a recession—the way we use the word—is a period of diminishing activity rather than diminished activity. We identify a month when the economy reached a peak of activity and a later month when the economy reached a trough. The time in between is a recession, a period when the economy is contracting. The following period is an expansion. Economic activity is below normal or diminished for some part of the recession and for some part of the following expansion as well. Some call the period of diminished activity a slump.

Q: You emphasize the payroll survey as a source for data on economy-wide employment. What about the household survey, which showed a decline in employment in August?

A: Although the household survey is a large, well-designed probability sample of the U.S. population, its estimates of total employment appear to be noisier than those from the payroll survey. The downward jump in August, which differs from the payroll data, may be such a random movement. Data in the coming months will help resolve the discrepancy between the two sources of data on employment.

Q: How do the movements of unemployment claims inform the Bureau’s thinking?

A: A bulge in jobless claims would appear to forecast declining employment, but we don't use forecasts and the claims numbers have a lot of noise.

Q: What about the unemployment rate, which jumped 0.4 percentage points in August?

A: Unemployment is generally a lagging indicator. Its modest rise from a very low level to date is consistent with the employment data. The household survey—the source of the unemployment rate data—contains random noise that occasionally results in larger than expected changes, as in August.

Q: How do structural changes in the economy in the 1990s affect the NBER's method for dating business cycles? The Bureau notes that industrial production measures a declining part of the economy. What other substitutes for output bear watching, particularly with regard to service sector activity?

A: Economy-wide employment and real personal income are the most important monthly indicators. At a quarterly frequency, real GDP is informative. Another interesting monthly indicator is aggregate hours of work. For the service sector, the BEA publishes monthly data on consumption of services (http://www.bea.doc.gov/bea/dn/nipaweb/TableViewFixed.asp?SelectedTable=206&FirstYear=2000&LastYear=2001&Freq=Month). Interestingly, these data show that consumption of services has grown more slowly in past months than consumption of durable and non-durable goods.

Q: Regarding movements of income as an indicator of recessions, isn’t it true that real income has not fallen substantially during five of the past nine recessions.

A: That is why employment is probably the single most reliable indicator.