>On Fri, 9 Jan 2004, Doug Henwood wrote:
>
>> This morning's U.S. employment report showed a gain of 1,000 jobs in
>> December, which sucks.
>
>I was wondering -- is there any chance this is because the seasonal
>adjustment figure needs re-adjustment? That retailers didn't hire as much
>seasonal help as past trends would indicate?
That happened last December, when seasonal hiring was lower than usual, so after adjustment there was a loss. And when January came, there were fewer layoffs than usual, which became a gain after adjustment. But starting in June, the BLS has been using a method called "concurrent seasonal adjustment," which generates the adjustment factors on the fly, instead of being projected forward from past experience. The new technique is supposed to reduce the effect of seasonal anomalies like that. A lot of Wall Street analysts haven't really figured that out yet, so they're still saying things left over from the old days. One example: some of them were saying that the late Thanksgiving postponed Xmas hiring into December, which is one reason expectations for December employment growth were so high. But, according to the BLS, had that happened (which it didn't) concurrent adjustment would have distributed some of the strength back to November (resulting in an upward revision to November's number). The point of seasonal adjustment is to smooth out month-to-month volatility to clarify the underlying trend. The BLS explanation of concurrent adjustment is available from this link: <http://www.bls.gov/ces/cesnaics.htm#concurrent>.
Doug