Employment Trends before and after Business Expansion Peaks

Policymakers and analysts closely scrutinize movements in labor market conditions to assess the evolving strength or weakness in the macro economy. For example, increases in employment or decreases in the unemployment rate usually signal an economic expansion, while the opposite tends to signal an economic slowdown or the leading edge of a recession. Thus, labor market conditions are often thought to be a comprehensive indicator of overall economic activity.See Kevin L. Kliesen, “How Well Does Employment Predict Output?”, Federal Reserve Bank of St. Louis Review, September/October 2007, Vol. 89, No. 5, pp. 433-46.

The close correlation between changes in employment and aggregate economic activity, as measured by the growth of real gross domestic product or real gross domestic income, is used by the Business Cycle Dating Committee of the National Bureau of Economic Research to help determine business cycle peaks and troughs. Indeed, this committee follows the two main measures of employment that are published each month by the Bureau of Labor Statistics in the employment situation report. These two measures are known as nonfarm payroll employment and civilian employment. They are derived from separate surveys of (1) nonagricultural (or nonfarm) business establishments (including government) and (2) private households. The former survey is known as the Current Employment Statistics (CES) survey, and the latter is termed the Current Population Survey (CPS).Kliesen’s 2007 Review article details the key differences between the CES and CPS.

The two tables below show average monthly changes in employment (1) four to six months prior to the business expansion peak month, (2) one to three months prior to the peak month, (3) during the expansion peak month, and (4) one to three months following the peak month. I also include the median estimates. The first table uses the CES nonfarm payroll series, and the second table uses the CPS household series. The sample period is 1960 to 2007. Note that I am not including the COVID-19 pandemic recession because its unique nature produced extraordinarily large declines in employment after the expansion peak in February 2020.Nonfarm payrolls declined by 3.31 million in March 2020 and then by an additional 22.19 million in April 2020. Nonfarm payrolls then increased by 3.95 million in May 2020 and by 5.21 million in June 2020.

A few characteristics of employment trends around business expansion peaks stand out. First, average monthly increases in employment—regardless of the series used—tend to moderate modestly prior to the expansion peak. The first table shows nonfarm payrolls increased by an average of 147,000 per month four to six months prior to the expansion peak. The average gain then slowed to 100,000 per month in the three months prior to the peak. This pattern holds for the CPS series, but the magnitude of slowing is larger—from an average of 232,000 to 67,000.

Second, the average increase in employment is appreciably higher in the peak month compared with the three months prior to the recession. Thus, there tends to be a relatively sharp increase in employment in the month before the economy falls into a recession. In only two of the eight episodes was there a decline in payroll employment in the peak month, as shown in the first table. There were three declines in the peak month using the CPS estimate, as shown in the second table. Third, and probably not surprisingly, employment growth turns modestly negative after the economy enters into a recession. Interestingly, there was a period from 1969 to 1980 when employment continued to increase after the peak month, but this no longer holds.

Fourth, there is considerable variation in monthly employment changes across episodes. The last line on each table measures the standard deviation for each horizon. For nonfarm payrolls, the smallest standard deviation (i.e., deviations of employment changes from the average) generally occurred in the one-to-three-month period prior to the peak. At this horizon, the standard deviation was 59,000. But the standard deviation at this horizon was much larger for civilian employment (CPS), at 181,000. Indeed, the standard deviation for CPS employment at all horizons was considerably larger than for CES employment. At the peak month, the standard deviation for CPS employment (494,000) was about three and a half times larger than the deviation for CES employment (142,000). The appreciably larger standard deviation for CPS employment suggests that the signal from CES employment around business cycle peaks is more reliable than the signal from CPS employment.

Finally, each table plots the most recent observation based on data through July 2024, the latest employment data available. The July employment report was much weaker than expected, as nonfarm payroll employment rose by only 114,000; the consensus estimate of economic analysts was that nonfarm payrolls would increase by 175,000 in July. Civilian employment rose by even less in July to 67,000.

It is rarely wise to attach too much significance to one particular data point. That said, the trend in recent gains in nonfarm payrolls does not conform to the historical average shown in the first table: There has been a steady slowing in job growth this year through July. However, the recent trend in civilian employment growth does conform to the historical average noted in the second table: Relatively stronger growth in the prior four-to-six-month period, weaker growth in the prior one-to-three-month period, and then faster growth in the current month. But as I argued above, the CPS data are too volatile on a month-to-month basis to yield definite conclusions.

Overall, then, the big picture is that employment growth remains positive. There are few signs that would lead an economist to question the duration of the current economic expansion based solely on the payroll and civilian employment data.

Notes

1. See Kevin L. Kliesen, “How Well Does Employment Predict Output?”, Federal Reserve Bank of St. Louis Review, September/October 2007, Vol. 89, No. 5, pp. 433-46.
2. Kliesen’s 2007 Review article details the key differences between the CES and CPS.
3. Nonfarm payrolls declined by 3.31 million in March 2020 and then by an additional 22.19 million in April 2020. Nonfarm payrolls then increased by 3.95 million in May 2020 and by 5.21 million in June 2020.