Labor Market not So Anomalous After All
July’s employment report was welcome news, especially after the slowdowns in payroll growth that had occurred over the previous two months. The U.S. economy added 117,000 new jobs, according to the Bureau of Labor Statistics report. That is slightly better than the average monthly gain of the second quarter (about 105,000), but definitely worse than that of the first quarter (about 165,000). Manufacturing, trade, professional and business services and education and health posted significant gains in July, as in recent months. Government payrolls, on the other hand, kept declining. July’s decline was 37,000, most of it due to state and local governments (their payrolls declined −23,000 and −16,000, respectively). The temporary help services sector, which is thought of as a leading indicator for future payroll growth, was basically flat.
Separately, the household survey showed that the unemployment rate ticked down 0.1 percentage point to 9.1, partly due to a decline in the labor force of 193,000. However, among those who are unemployed, almost 45 percent have been unemployed more than six months, which is close to the all-time high reached in the midst of the last recession. So the news from the labor market is at best mixed: We do not see an all-out slowdown, but there are no robust improvements either.
Overall, the labor market has been adjusting very slowly during this recovery. Six months after the economy had started growing again, we were still losing jobs. In other words, employment, as measured through payroll survey, had experienced its largest decline (more than 6 percent) two years after the recession started. More troubling still is the slow pace at which employment is returning to normal. Three and a half years after the beginning of the recession, we are still 5 percent below the prerecession level of employment, almost 6.8 million jobs! This pattern of slow progress in the labor market was a key feature of the two recoveries that preceded the last one, and they were sometimes dubbed the “jobless recoveries” on account of it. The only difference between those two recoveries and the last one seems to be that this time around we suffered a much larger decline in employment.
The unemployment rate does not paint a better picture. It increased more during the past recession than in any previous recession, and moreover, since it peaked over the 23 months ago, it has come down only 1 percentage point. There is always some persistence in the unemployment rate; that is, the unemployment rate does not necessarily return to pre-recession levels even three or four years after the start of the recession. But the degree of persistence at these levels is a significant exception by historical standards.
However, there is a hint of good news. Some perspective about the type of a recession we just experienced might help to see it. When one thinks of the impact of the recession on the labor market on its own, one gets a pretty bleak picture. However, in light of the recent revisions of estimates to gross domestic product (GDP), the recession’s effects on the labor market don’t seem to be so far from what we would expect.
The Bureau of Economic Analysis (BEA) occasionally revises its estimate of GDP to reflect new data as well as methodological improvements. One such revision recently showed that U.S GDP, the broadest measure of aggregate economic activity, declined more than 5 percent between the fourth quarter of 2007 and the second quarter of 2009, making it the largest ever decline in postwar history. There is nothing good about this news per se, but the fact that the recession was actually much worse than initially thought puts things in a different context.
Economic theory suggests that bigger contractions in GDP will have bigger impacts on the labor market; deeper recessions imply large losses in employment and greater rises in the unemployment rate. Consider this relationship between the decline in payroll employment and the decline in measured GDP, from peak to trough. If we had done this calculation for the last recession sometime in mid-2009, we would have found that a 3.7 percent decline in GDP was associated with more than a 5 percent decline in payrolls, and that response would have been somewhat of an outlier. Similarly, a year later, after one set of downward revisions to GDP, the last recession might have still looked a bit puzzling—4.1 percent decline in GDP associated with a 5.9 percent in payroll employment (the cumulative decline by that time). Since then, both the BLS and BEA have revised their estimates (payroll employment and GDP estimates). The bad news is that the overall declines in payroll and GDP, from their respective peaks to their respective troughs, are larger. However, in some sense, this makes this last episode much less puzzling.
The unemployment rate’s behavior during this recovery also looks less mysterious after the GDP revisions. However, unlike the employment figures, the unemployment rate was not revised during this period. So every change we have seen over the last two years in the relationship between GDP and unemployment is due to changes in the GDP estimates. The 5.1 percent increase in the unemployment rate between December 2007 and October 2009 was exceptionally high relative to the first estimate of the GDP decline. However, as we got a better handle over time on how bad this recession really was, it became less of a puzzle.