Monday, October 05, 2009

An Important Detail On Unemployment Measures

Last week on CNBC there was a fascinating discussion by an economist on the mechanics of unemployment rate measurement.

I can't recall who was speaking, but his details called to mind similar comments by another economist last year on the network. The difference was, this time, the guest linked it to the allegation that rising unemployment lags an economic recovery.

The specific detail is this. When employment is surveyed, it's by telephone. The two key questions asked are:

1. Are you employed?
2. Are you looking for work?

The response to the first is pretty simple. If you answer "no," then, at that point in the survey, you are unemployed.

However, the second question is asked if the answer to the first was "no." If you answer "no" to this question, you suddenly become not unemployed, but simply not counted as either employed or unemployed. That's because if you say are not looking for work, you effectively reduce the denominator of people looking for work who are either working, or not.

From a purely arithmetic viewpoint, you can understand this logic. However, it has, as the CNBC guest noted, perverse consequences on the stated US unemployment rate.

For example, he said, during many months of no expansion whatsoever, people simply stop spending time and money looking for a job. Thus, by the survey's standards, they are not unemployed.

Then, one month, say, last month, the person hears of hiring activity somewhere in the economy, and decides to once again seek employment. The monthly phone survey now finds him "unemployed," because he's not working, but looking for work.

The guest pointed out that, in reality, nothing changed about the person's employment circumstance. What changed is the survey's treatment of this not-working person.

Because behavior like this tends to occur later in a recession, as some hiring begins to take place, the measured unemployment rate thus rises.

And that is why it appears that continued rising unemployment lags a recovery. It's an statistical measurement artifact, not a real phenomenon at the basic level of people who are, or are not employed.

Not only does this explain why unemployment appears to lag economic recoveries. It also sheds light into the probable underestimation of unemployment throughout a recession.

With the current unemployment rate standing at 9.8%, this suggests that you can bet the real rate of unemployment topped 10% a while ago. Perhaps in mid-summer. Nobody knows for sure.

Thanks to this pollution of the unemployment measure by removing those who once sought work, then did not, then did, again, what we have is consistent measure which we know is wrong.

Wouldn't it be better for the survey to be modified to track whether the respondent was looking for work, if not now, some months ago? Or ask how many months it has been since they last actively sought work?

Then several measurements of unemployment could be provided, with varying denominators, according to what one chose to measure.

As it is now, a supposedly "hard" number rests, in fact, on psychological influences upon the unemployed as to whether they will say they are actively seeking work.

No comments: