Why U.S. Employment Numbers Can Be Hard To Read

When the latest employment report indicated that U.S. payroll job growth in August had come in at just 179,000--far below the average 312,000-per-month pace of the previous six months--the financial markets heaved a collective sigh of relief. Fears that tightening labor markets were about to put upward pressure on wages evaporated in the face of this fresh evidence that the Federal Reserve's efforts to put the economy on a slower, more sustainable growth path were succeeding.

Not everyone was surprised by the deceleration in the payroll numbers, however. Anticipating a sharp slowdown in economic growth to a 2%-or-lower pace in the months ahead, economist Robert A. Brusca of Nikko Securities Co. International believes the payroll data have been painting an overly buoyant picture of job growth for some time.

What intrigues Brusca is a stark divergence between the payroll data, based on a survey of business establishments, and the Labor Dept.'s other employment measure, its survey of households. Between January (when the household survey was revamped) and July, the household data showed an employment increase of just 501,000, compared with a nearly 1.9 million increase in payroll jobs. The gap narrowed in August, but remained wide (chart).

Such gaps have occurred in the past, though they tend to disappear over time. Because the payroll survey is based on actual business records and is less volatile on a month-to-month basis, however, most experts regard it as more reliable than the household survey, which is based on interviews.

On the other hand, the payroll measure counts jobs and not people as the household measure does, so it can overestimate employment growth when people take multiple jobs. And Labor Dept. statisticians massage it with a "plug" factor, an upward adjustment to account for jobs created by business startups that are missed by the survey.

This adjustment is an educated guess, at best. It can be far too low when the economy is accelerating and too high when it is slowing down. With the plug factor now set at 158,000 jobs assumed to exist before the counting starts, Brusca thinks the payroll data are currently out of whack on the upside. "The survey itself picked up only 21,000 new jobs last month," he says. "Welcome to the ersatz job market."

What of the leap in the household employment measure last month? Brusca notes that seasonal adjustment weighs heavily in August, and lacking historical data, the Labor Dept. is using seasonals from the household survey's previous incarnation. In August, that "iffy" procedure, he says, converted an actual dip in employment of 10,000 into a seasonally adjusted gain of 714,000.

Before it's here, it's on the Bloomberg Terminal.