Jobs Report Contains Some Unwelcome Twists

• Two annual revisions and new concepts in the January employment report revealed a deeper story.
• Revised payroll data indicate 1.4 million fewer jobs at the end of 2009.
• Popular perceptions notwithstanding, labor income is recovering less quickly for managers and executives.

The story from the basic data in Friday's employment report was the same: The labor market took a decisive step toward stabilization around the end of 2009 as layoffs slowed, but firms have yet to add to their payrolls. But this time, the monthly report contained more information than usual—and that information was not comforting.

Along with the usual revisions to the prior two months of data, Friday's report contained the new benchmark level as well as revisions to seasonal and birth/death factors. Back in October, government officials said that the payroll survey had significantly overestimated employment for the 12 months ending in March. But more recent data were also revised downward. Save for November, job growth was lowered for every month of 2009. The bottom line: December employment was a stunning 1.4 million lower than previously reported.

The hole the labor market must climb out of before returning to full employment now looks even deeper: 8.4 million jobs have now been lost since the recession began. To put that in perspective, we would need payroll gains of 200,000 per month every month for three and a half years just to get back to late 2007 levels, and even that calculation ignores the labor force growth over the intervening years. Given this backdrop, it's easy to see why utilization rates will be very low for some time and why the Federal Reserve has significant leeway to keep interest rates low.

There are reasons to be optimistic that some hiring will begin soon, but there are also reasons to be nervous about how strongly hiring picks up, particularly in light of the setback in financial markets and the stall in jobless claims. Greater improvement in the labor market is needed before a lasting expansion can take hold.

The benchmark revision was unusually large, and the big miss is being blamed on the shortcomings of the birth/death model and on the payroll data underestimating business closures in particular. The implication is that small firms, which are not as well captured in the survey as larger firms, were disproportionately damaged by the recession. The government has talked about overhauling its birth/death model, but the revised factors in the latest report come from the same model run with newer data. The shortcomings of this model should become less apparent once small-business growth resumes.

Household survey

The household survey results were generally much better than they have been in recent months, but new population controls complicate comparisons between the January and December data. Because the population controls do not affect the data in 2009, month-to-month changes in the broad aggregates such as employment, unemployment and the labor force are meaningless.

That said, while the annual adjustments result in a break in the broad aggregates, the ratios such as the unemployment rate mostly control for these changes. The jobless rate fell 0.3 of a percentage point to 9.7% in January, and the broadest measure of labor underutilization, the U-6 unemployment rate, fell 0.8 of a percentage point to 16.5%, thanks to a large drop in those involuntarily working part time. On the other hand, the number of discouraged workers not in the labor force rose above 1 million for the first time.

In addition, although the reported 541,000 increase in household employment is distorted by a break in the population estimates, the rise was actually larger when controlling for the change. Looking at month-to-month changes in the major series while controlling for the change in population estimates, household employment rose 784,000. Even without this adjustment, the January results bring employment in the household and payroll surveys back into line after a period in which the household survey was substantially weaker.

New concepts

In another change, the establishment survey began reporting several new concepts, mostly for earnings and weekly hours. Previously, the establishment survey reported employment for all employees, but reported only average hourly earnings and the average workweek for production and nonsupervisory workers. This category included production workers in goods-producing industries and nonsupervisory workers in service-producing industries. Nonproduction and supervisory workers account for about 20% of total employment and, presumably, an even greater share of total income given generally higher earnings.

The new data extend back to 2006 and generally show slower growth in earnings and hours. Thus, the payroll proxy of labor income has risen less for all employees than for production and nonsupervisory workers only. For production workers, the number of hours worked has risen at a sturdy 4.2% annual pace in the past three months, and labor income has risen at a 6.4% clip over that period. Meanwhile, hours worked for all employees are up at a 1.8% three-month annualized rate, and labor income is growing by 4.2%.

The upshot is that the income of executives and managers is generally recovering more slowly than for the rest of the workforce, a story that might not resonate politically but is nonetheless supported by the data. The new seasonally adjusted all-employees series will likely become the preferred measure in the employment report. Besides the benefit of having a more complete count, the new series will also feed into other economic data that are estimated using earnings and the workweek, namely personal income and productivity.