Have employment trends this year been even weaker than the monthly payroll job numbers indicate? The question is important for a variety of reasons. If the modest job gains reported earlier this year never took place, for example, the current malaise expressed in consumer sentiment surveys would be more understandable. Moreover, the payroll numbers undergo annual benchmark revisions, and such revisions affect final estimates of income and output. Thus, they have the potential of altering the official measure of the length and severity of the recession.
That's exactly what happened earlier this year, when the 1991 job count was revised based on comprehensive employment data for March, 1991. This benchmark revision lowered the national payroll-employment level by some 650,000 jobs. And that had the effect of converting a modest rise in gross domestic product in the third quarter of 1990 into a decline, and of deepening the drop registered in the first quarter of 1991. In short, it changed what appeared to be a very short, mild recession into a more serious contraction.
According to economist Mark Zandi of Regional Financial Associates, it could well happen again. Zandi points out that in constructing the monthly payroll-employment estimates from its nationwide sampling of 350,000 companies, the Bureau of Labor Statistics adjusts the count to correct for underlying trends it feels are not picked up by the survey. At the same time, each state bls office makes similar adjustments for factors it feels are relevant to its own count. Because the state and national adjustments are different, the national count and the sum of all the state tallies often diverge, and this year the deviation has been especially large (chart).
Since January, notes Zandi, payroll jobs as measured by the state tallies have been falling steadily, while the national estimate has only recently turned down. "Between March and July," he says, "the states' job count fell by 377,000, while the national estimate shows an increase of 400,000--a disparity of 777,000 jobs in just four months."
Zandi doubts benchmark revisions next year will confirm the job growth portrayed by the national count. Record business failures, slowing new-business formation, and sagging consumer confidence, he says, suggest that the weaker state tallies are on target. And if that's the case, he adds, next summer's revisions in gdp could well turn the economy's reported 1.5% gain in the second quarter of this year into a 1% decline.