The national employment data are telling two stories — one not so bad, the other unequivocally bad. Which one is right?
Nonfarm payroll jobs, as measured by employer survey, rose a respectable 166,000 last month, defying credit-crunch pessimists, with gains in services more than offsetting losses in manufacturing. Total employment, as measured by household survey, however, showed no improvement. The unemployment rate was unchanged at 4.7 percent.
But one month does not a trend make. Looking at the year to date, the divergence in the two employment series is more dramatic. Payroll jobs are up more than 1.2 million since last December, an average gain of 125,000 a month, whereas total employment has been essentially unchanged since the end of last year.
Total employment, unlike payroll jobs, is a person count and includes agricultural workers and the self-employed, among others, while nonfarm payrolls include multiple jobs held by individual workers. Some of the discrepancy between the two employment series results from this difference in coverage, notably from a decline this year in self-employment.
When total employment is adjusted to the payroll definition and smoothed to account for the annual updating of population controls in the household survey, employment this year picks up somewhat but still falls short of the gain in payroll jobs. After adjustment, the discrepancy in growth between the two employment series since the end of last year is reduced to about 700,000 — still a large number that needs explaining.
Because it is based on a smaller sample, the total employment number jumps around more than the payroll count from month to month. However, over several months that problem diminishes, and the signals from both series have to be taken seriously. Doubters are referred to the study by George L. Perry, senior fellow at the Brookings Institution: "Gauging Employment: Is the Professional Wisdom Wrong?" Another way to assess employment is relative to the growth in the working-age population. The employment-population rate, reported monthly, is a measure of labor utilization. So far this year payroll employment has about kept up with population growth, but total employment has not.
The total employment-population rate, estimated from the government's household survey, reached a post-recession peak of 63.4 percent last December and has since slipped to 62.7 percent last month, the lowest rate since mid-2005. This fall-off represents a sizeable employment loss, about 1.5 million.
Potential workers responded to the weakening employment rate, as often happens, and the labor force participation rate fell from 66.4 percent at the end of 2006 to 65.9 percent last month. That decline represents a labor force loss of 1.2 million people, which some economists refer to as a rise in "hidden" unemployment. If this year's dropouts had met the official job search definition and been counted as unemployed, the unemployment rate today would be 5.5 percent instead of 4.7 percent.
Total employment and the unemployment rate both come from the government's household survey. When employment is flat, unemployment normally rises, if for no other reason than growth in the working-age population. With total employment stagnant this year, the official unemployment rate should have increased significantly, even allowing for some labor force withdrawal. But the latest 4.7 percent jobless rate is only slightly above the 4.5 percent rate of last December. If you believe the unemployment data, the implication is that total employment growth is underestimated in the household survey. The moderate growth in payroll jobs better explains the behavior of unemployment than does total employment.
Which employment story is right? The recent strength in output seems to side with the more optimistic payroll data. On the other hand, some economists believe the statistical model used by the Bureau of Labor Statistics (BLS) to estimate business births and deaths is causing an overstatement of payroll job growth. There is also a suspicion among data users that the household survey is doing a better job measuring employment losses among illegal immigrants, particularly in the hard-hit construction sector. Solid information is lacking.
The BLS is responsible for both employment series and has capable economists and statisticians closer to the data than anyone else — career professionals who research, analyze and interpret employment trends. The public and policymakers need to know what is happening to employment, and the BLS needs to keep us up to date on what it learns. Time is important, and a page in the Bureau's employment release summarizing its latest findings would suffice, at least for the moment. If the BLS has no new insights as yet, we need to know that, too.
Alfred Tella is former Georgetown University research professor of economics.
Email This Page!