Anomaly In the Household Survey

There’s an anomaly in last month’s labor market data. I don’t know what to make of it; but some conservatives see some sneaky stuff being done to the data. I recommend we keep an eye on these numbers for a few months. Here’s what’s going on.

The Bureau of Labor Statistics determines the unemployment rate from two different surveys. Both survey physical locations.

One surveys households, the other survey “establishments” (basically, employers).

Similar questions are asked of both, but what people in a household consider employment can differ considerably from what a business would. So, the establishment survey is usually thought to be more definitive. On the other hand, it misses a lot of people doing unconventional jobs, particularly early in recoveries: starting businesses, working for new businesses that aren’t in the establishment survey yet, working off the books, and so on.

Normally, population rises every month, and you’d expect employment, unemployment and people classified as not-in-the-labor-force to go up too. Business cycle effects swamp the slow movement of population though. So, right now, you wouldn’t be surprised to find employment rising faster than population, and unemployment dropping. But what of not-in-the-labor-force?

Well … last month, not-in-the-labor-force went up by an extraordinary amount. Why is that so?

It could be that people dropped out of being employed, and decided to retire or otherwise take some time off. Except that unemployment went up, so there’s need to be an even larger group of new quasi-retirees to make the numbers work.

Or it could be that people are getting discouraged with being unemployed and have stopped looking. Except that the change is so large that if we counted all those newly not-in-the-labor-force people as unemployed, the unemployment rate would have gone up to 9.0% last month. The labor market may be soft, but that magnitude of change dwarfs what happened in the fall of 2008 when the economy was in a nosedive. No one is in a panic, so this story can’t work.

Or, it could be that all of a sudden, a huge number of teenagers came of age all in one month, or the armed forces discharged a million soldiers … or … or … a bunch of alien abductees were returned. That last one is no sillier than the first two.

More than likely this is all because of a correction for some past mismeasurement of the numbers. Fair enough: but then how is this revision going to filter into the other numbers? I have no idea: Table B in the technical notes to the news release indicate that about 75% of the increase in not-in-the-labor-force was due to better counting of women who aren’t working.