Friday, May 06, 2011

U.S. Employment Surveys Diverge More And More

Today's U.S. employment report was quite weak -or relatively strong depending on whether you focus on the household survey or payroll survey.

The household survey showed a net job loss of 190,000, causing the employment rate to drop from 58.5% to 58.4%. It is now in fact lower than the 58.7% in April 2010 as the 0.2% employment gain is a lot lower than the population increase. Meanwhile, the unemployment rate rose from 8.8% to 9% during the latest month. And while that is lower than the year before this is entirely and more due to a drop in the labor force participation rate.

The payroll survey by contrast showed a relatively strong gain of 244,000 for the month and a 1% gain during the latest year, with the numbers looking even stronger if you only look at private sector employment. The only thing that was weak was that the gain in nominal wages was so small that real wages likely declined.

There has always been some divergence between the two surveys, but it is unusual that the payroll survey shows as much as 0.8% more job growth during the latest year.

Since the two surveys are supposed to describe the same thing and since there can only be one truth about that thing, this means that at least one of them is wrong. The payroll survey is regarded by most economists, including me, as more reliable when it comes to describing monthly changes, so the truth is probably a lot closer to the payroll number about the latest month. However, that the household survey was so weak indicates that it is more likely that the payroll survey overestimated than underestimated the increase. Furthermore, when it comes to yearly changes the household survey is more reliable than it is for monthly changes, making it even likelier that the payroll survey during the latest year greatly overestimate job growth