Recent speeches by Federal Reserve officials suggest that the Federal Open Market Committee (FOMC) continues to closely monitor the health of labor market conditions.1 The reason, in short, is that labor market conditions are useful for identifying emerging strengths or weaknesses in the pace of aggregate economic activity.
For most economists, a key barometer of the health of labor market conditions is the monthly change in the number of jobs created each month. This series, usually referred to as payroll employment, is released by the Bureau of Labor Statistics (BLS) in its monthly employment report. An often underappreciated aspect of the monthly payroll estimate is that it is subject to several revisions over time. For example, the current estimate of February's job growth (197,000) is quite different from the initial BLS estimate of 175,000. These revisions are par for the course. Although often challenging for forecasters and policymakers, they nonetheless improve the accuracy of the estimates because they incorporate new information that was not available at the time the initial estimates were reported.2
The monthly employment reports are closely scrutinized by the FOMC as it seeks to ascertain the strength of the economy. Likewise, private forecasters and financial market participants use the same approach, not only to gauge the economy's strength but also to help assess the future direction of monetary policy. But how much weight have policymakers, forecasters, and financial market participants attached to this initial number? Maybe too much, if recent history is any guide. The charts plot the average monthly change in total and private (total less government) payroll employment since the January 2012 report (released on February 3, 2012).
The leftmost bar in the first chart plots the average monthly gain over this period as recorded by the initial estimate. These are the numbers that garner the most attention by the FOMC, private forecasters, financial market participants, and the media. Since January 2012, the initial estimates averaged about 153,900 jobs per month. However, the chart also shows that the initial estimate over this period appears to be downward-biased by a substantial amount. For example, the average first revision was about 160,800, a 4.5 percent increase from the initial estimate. The average second revision was even higher, showing a monthly gain of about 174,300 jobs, 13.3 percent more than the average initial estimates. Perhaps even more remarkable, the current estimate—which accounts for all previous revisions—shows that total payroll employment averaged nearly 188,900 per month over this two-plus year period—35,000 more than the initial estimate.
The second chart depicts the same exercise for private nonfarm payroll employment. The pattern is the same: Over this period, the average initial monthly change was about 158,400.3 Compared with the initial estimates, the average first revision was about 165,600, or 4.5 percent larger; and the average second revision was about 178,600, or 12.8 percent larger. Using current-vintage data, private nonfarm payroll gains averaged about 192,800 per month over this period, roughly 34,300 (or 21.7 percent) more than the initial estimate.
The pattern of the revisions shown in the charts is consistent with the evidence, which shows that revisions tend to be pro-cyclical. That is, during an expansion, revisions to payroll employment tend to be positive; and, during periods of slow economic growth and recessions, the revisions tend to be negative.4 From this standpoint, then, the outlook for the economy remains positive, as suggested by the most recent Summary of Economic Projections released by the FOMC.5 Still, while the initial monthly estimates of nonfarm payroll employment are important for policymakers and financial market participants, they would be wise to remember that these estimates can be (and often are) revised—by a lot.