Tuesday, October 12, 2010

Another Thing About Employment Report

I noted in my analysis of the U.S. employment report that the household survey was stronger than the payroll survey in terms of the number of people employed.

What I missed was however that the household survey was weaker than the payroll survey in terms of how much time the employed worked.

The payroll survey claimed that the average work week for those employed was unchanged. Yet the household survey claimed that part-time unemployment (or underemployment) increased dramatically, from 5.7% to 6.1%.

While it is possible that this discrepancy can be accounted for by a decrease in voluntary underemployment, the most likely explanation is that there is a statistical discrepancy between the two surveys.

As I've explained before, the household survey and the payroll survey should give the same result since they're supposed to describe the same thing. But because they use different methods in data collection, they often get different results.

The payroll survey is usually more reliable in terms of monthly fluctuations. But the likelihood that the real numbers differ in a certain direction increases when the household survey gives a certain result. And in this month, the household survey suggested that the change in the number of employed may have been stronger than what the payroll survey suggested, but that there might have been a decrease in the average work week (and therefore also in average weekly earnings) instead of the flat number suggested by the payroll survey.