- The Washington Times - Tuesday, February 15, 2005

New employment data from the Bureau of Labor Statistics (BLS) revise recent economic history and lend validity to household-collected employment, which has risen faster in this economic recovery than employer-reported payroll jobs.

Earlier this month, the BLS revised nonfarm payroll employment back to January 2000. The revisions included a more complete annual benchmark count of jobs from unemployment insurance tax records for March 2004, which added to the previously reported monthly data from April 2003 forward. Some 203,000 jobs (before seasonal adjustment) were added to the March 2004 benchmark month, slightly less than BLS’ advance estimate of 236,000. The revised data also incorporated updated seasonal adjustment factors from January 2000 forward.

Compared with the previously reported payroll jobs data, the revised series shows a somewhat different picture for the current business cycle and for last year.

In April-June 2004, the revised data show jobs increased 65,000 more than previously stated. The previously reported increases in the first and last three months of the year were trimmed by 64,000 and 59,000, respectively. The July-September period was little affected by the revisions.

The revisions also showed the job recession that began in early 2001 was two months shorter than previously reported, and the job recovery three months longer. The cyclical peak in jobs shifted back one month, from March to February 2001, while the trough shifted back three months, from a previous low in August 2003 to a low in May 2003.

The payroll job recovery, previously thought to have started in September 2003, is now shown to have begun three months earlier, in June, though the 2003 summer rise in jobs was sluggish. The revisions showed a slightly steeper job recession and a somewhat stronger recovery.

This month the BLS also reported revisions to its more comprehensive household survey of total employment. (Household measured employment is about 7.6 million more than nonfarm payroll jobs.) In its January 2005 data, the Bureau introduced new annual population controls, mainly reflecting updated estimates of net immigration. This latest revision, modest compared with earlier years, lowered December 2004 employment by 45,000.

In past years, the January introduction of new annual population controls often seriously distorted comparisons of employment over time. More recently, the BLS has helped data users overcome this by publishing total employment data that spread the lumpy January control adjustment over the previous year. Though the Bureau calls the adjusted data a “research series,” it is far more useful than the official data.

For example, in just the month between December 2002 and January 2003, the official seasonally adjusted total employment count jumped up by over a million, or by nearly 600,000 more than the smoothed data. From December 2002 to December 2003, the official data show a 2 million increase in total employment, compared with a 1.3 million gain in the smoothed series.

The population control adjustments to the total employment data did not revise their recent cyclical turning points ” an employment peak in March 2001 and a trough in April 2002 (more than a year earlier than the trough in payroll employment). But it shows a slightly shallower job recession and a slightly weaker recovery, in contrast to the pattern indicated by revisions in the payroll data.

To its credit, the BLS has gone a step further and published the control-smoothed total employment data back to 1994, adjusted to be conceptually consistent with its payroll employment data. Now trends in the two series can be fairly compared over time. The adjustment involves subtracting from total employment agricultural and self-employed workers, unpaid and private household workers, and workers on unpaid leave, and adding in nonfarm wage and salary workers with more than one job, since payroll data count jobs, not people.

A comparison of the payroll and the concept-adjusted household employment series over the last decade reveals five distinct patterns:

From 1994 to 1998, the rising trend in the two series were quite similar. From early 1998 to the 2001 recession, payroll jobs rose significantly faster than adjusted household employment. Both series declined at about the same rate during the recession. From the end of the recession in late 2001 until August 2002, adjusted household employment rose while payroll employment continued its recessionary decline, which raised much controversy.

Since August 2002, however, the two series have come together again and moved forward on an identical rising trend. Payroll and conceptually adjusted household employment both numbered 130.3 million in August 2002, and both measured 132.6 million in January 2005.

Even though the payroll measure excludes important work-force components, most data users have preferred it to the household series mainly because of its basis on a larger survey sample. But the impressive consistency in trend between the concept-adjusted and payroll data for well more than two years lends renewed credibility to household survey employment (i.e., total employment not conceptually adjusted). This in turn lends greater credibility to the household data’s message of stronger economic growth since the end of the last recession. Since then, total employment growth has exceeded nonfarm payroll job growth by more than 2 million.

With the kind of structural changes now occurring in the economy, we need employment data that will help us see the total picture and spot new trends. We need to know if and by how much nonpayroll employment may be expanding faster than payroll jobs, data only the household survey can provide.

In this budget season, it might be a good idea to increase the government’s investment in the household survey data, which would strengthen the total employment numbers and bring them the wider acceptance they deserve.

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