Jobs Figures: Garbage In, Garbage Out?

The payroll jobs survey isn't just a raw survey. It's actually adjusted to a degree by seasonal/monthly modifications in an effort to "tune out" predictable cyclical noise. Those adjustments are, of course, based on assumptions. If the assumptions fail, the numbers won't be accurate.

This guy has been complaining about these adjustments for a while, and not just when they hurt Bush. I remember reading him some time ago arguing that Bush's big-growth numbers in the Spring were also partly phatasmal, based on sketchy adjustments which might or might not bear any relationship to reality. Hack that I am, I discounted him at the time; I didn't need him raining on my cowbell parade.

But he predicted yesterday that these adjustments would "make" July a bad job-growth month, at least on paper, just as they had last July. (January, too, is always adjusted into a poor-job-growth month.)

I'm not sure what to make of all this. Do we really have any good idea at all how many jobs are being created or lost at any given time? How much of this is all just back-of-the-envelope guesswork?

We do know, with nearly perfect accuracy, how many people are signing up for jobless benefits at any time. That's not a poll; that's just counting.

If the number of people signing up for unemployment benefits keeps dropping, doesn't that suggest that more people are finding work?

Hat tip to Kausfiles.

Bill Quick has a more rigorous analysis of this phenomenon.

posted by Ace at 04:21 PM

| Access Comments