Friday, May 04, 2007

Taking the Over Bet on the Jobs Report

I might be deleting this post if I'm wrong.

The monthly jobs report is due out in about a half hour. Consensus is calling for a gain of about 100,000 new jobs created in the month of April. Atrios takes over/under bets shortly beforehand just for fun. This month he predicts the report will come in below consensus predictions. Usually that's a safe position, but this month I think I'll take the over bet. Why? Because of an artificial statistical technique the Bureau of Labor Statistics uses called the Birth Death Model.

This model attempts to capture the jobs created by new businesses that haven't filed with their respective state employment agencies yet. I've written about this model and the distortions it creates in our jobs statistics here and here. Long story short, the model assumes for each business that goes kaput, another business in the same area and industry rises from the ashes. From that assumption comes the next assumption that jobs lost by the old business are replaced by the new. Therefore, measuring jobs lost allows you to "impute" (that is "make up" or "imagine") the same number of new jobs in the same areas. The way the BLS describes the math and application of this model suggests that the vast majority of job growth reported under the Bush administration may be totally imaginary.

There's a seasonal quality to this model that's consistent from year to year. For last month's report, an average of 152,000 "imputed" jobs were added to the actual measured statistics in the last three years. For the April report, the model added an average of 234,000 "imputed" jobs in the last 3 years.

This suggests that if the jobs report comes in at 100,000 new jobs as predicted, the actual statistics are pointing to a net loss of some 100-150,000 jobs, papered over by the substantial addition of "imputed" jobs by the Birth-Death model. I don't think we're in that bad a shape, so I'm going to predict that the jobs report will show some number well above 100,000 jobs. This will result in an orgy of optimism that we've "turned the corner" in the anticipated economic slowdown and everyone will imagine nothing but sugar plums and ponies for the US economy from now into the long term. Which, on Wall Street, means sometime after lunch.

Beware irrational exuberance and keep your hands on your wallet at all times.

UPDATE: The jobs report just came out. Atrios wins. This is not good. Not because Atrios won and I lost, but the report shows a net gain of only 88,000 jobs. Not only is that only about half what the economy needs each month to keep up with population growth, but it looks to be much worse than that. The BLS' Birth Death model page shows that 317,000 "imputed" jobs were added by the model, suggesting that the real data measurement of jobs shows a net LOSS of over 220,000 jobs in April. Not good. I'm rarely accused of being too optimistic, but it looks like that's where it is today.

UPDATE 2: The revision to the May jobs report just came out. I win after all. The net gain for last month was revised upwards to 109,000. Still pretty dismal. Just slightly less so.