- The Washington Times - Wednesday, June 6, 2007

Nonfarm payroll employment perked up in May, rising a respectable 157,000, about twice as much as in April. Since the end of 2006, payroll jobs, as measured by the government’s employer survey, are up 664,000. The unemployment rate was unchanged in May, and at 4.5 percent, was the same as last December.

In contrast, total employment as measured by household survey (a more comprehensive count that includes the self-employed, agricultural, private household and unpaid family workers, and workers on unpaid leave) has been stagnant so far this year — a puzzling inconsistency. In the three prior years, both employment series were on a generally rising trend.

Total employment as a percent of the working-age population (a measure of worker utilization) and the labor force participation rate (the proportion of the population working or actively looking for work) have both fallen by a sizable 0.4 percent since last December. The declines have been spread across most major demographic groups.

Unfortunately, the media feature the monthly payroll data and rarely mention the total employment numbers. Granted, a single month’s change in household employment needs to be interpreted with caution, since the data are based on a smaller sample than the payroll data. But over a several-month period, the household data take on increased reliability and should not be ignored.

If the employment message from the household survey should happen to be right, the labor market so far this year is in worse shape than we have been led to believe. It would not be surprising, then, that the labor force participation rate declined along with the employment-population ratio. If jobs became harder to find, more people could have given up and left the work force.

Not all of this year’s decline in labor force participation, however, can be attributed to possible weakness in the job market. In recent years the participation rate has been on a declining trend apart from cyclical factors, reflecting the aging of the population and structural influences. Assuming the 2007 household employment data are correct, I estimate that about one-fourth of the participation-rate decline since last December is a continuation of a trend and about three-fourths represents labor force withdrawal in response to insufficient demand.

This translates to a cyclical decline in the participation rate equal to a labor force disappearance of about 700,000. Some economists refer to such workers waiting on the sidelines for the job market to improve as the “hidden” unemployed. If the recent estimated increase in their numbers is added to the labor force and to the current 4.5 percent official jobless rate, the unemployment rate so adjusted jumps to 4.9 percent. One can only hope that the rising trend in the payroll data proves to be correct and that the household employment data are, for some reason, only temporarily out of line.

If this year’s total employment data are wrong and the rising trend in payroll jobs is right, then we can ask what the current unemployment rate would be if the household employment numbers had followed the same upward trend as the payroll data. I estimate that the jobless rate in May would still have been about 4.5 percent.

Ironically, support for the payroll data comes from the household survey itself. The household survey is designed to better measure changes in unemployment than employment. So if the recent and current unemployment picture is reasonably accurate, then the upward-trending payroll numbers in recent months are certainly more consistent with officially reported unemployment than household-measured employment. Over time, as the population grows, employment growth is necessary to prevent joblessness from rising.

Some, though not much, of the recent weakness in total employment is attributable to a decline in the number of agricultural workers, who are not included in the payroll data. Also, multiple jobholders have slightly increased their share of total employment, which increases the count of payroll jobs but not the number of people working as measured in the household survey.

But the greater part of this year’s discrepancy between the two employment series remains unexplained, and one can only speculate about the possible causes. For example, it may be that in recent months various attempts by the federal and some local governments to tighten up on illegal immigration have caused more illegal workers to be wary and avoid responding to the household survey. If so, that would have a negative effect on household-measured employment. Yet some of the illegal immigrants avoiding the household survey may be reported by businesses responding to the payroll survey.

There could also be a problem with the payroll data. The statistical model that the Labor Department uses to estimate the net change in payroll employment resulting from business births and deaths is subject to error, and revisions in the data are sometimes sizeable. New firms are included in the payroll survey only with a lag. If the model is currently overcounting — we’ll be told later — measured job growth is upward biased.

For now, we can’t be sure whether the employment picture is good or bad.

Alfred Tella is former Georgetown University research professor of economics.

Sign up for Daily Newsletters

Manage Newsletters

Copyright © 2020 The Washington Times, LLC. Click here for reprint permission.

Please read our comment policy before commenting.


Click to Read More and View Comments

Click to Hide