Saturday, December 4, 2010

On Those Negative Unemployment Numbers

The Labor Department's initial estimate of how many jobs in November was just 39,000 versus 172,000 in October.

Bernanke will use this weak number as justification for QE2, but one has to realize that the jobs numbers are very misleading. The BLS uses so many adjustment factors that as opposed to real data that the numbers can be off significantly, even assuming the counting the actual counting they are doing is accurate.

For example, according to the BLS, November's retail employment fell by 28,000 jobs, and unemployment thus climbed to 9.8% from 9.6%. The decline in retail jobs accounted for most of the uptick in unemployment. But, this was all because of the seasonal adjustment. There is no way  retailers were laying people off heading into the Christmas season. The indicated decline was completely a function of the Labor Department's seasonal adjustments. Without the seasonal adjustment, retailers added roughly 300,000 jobs in November.

Now get this, according to WSJ, the Labor Department does its count of payrolls in the week that includes the twelfth day of the month. In November, that week came early, so the count might have missed the hiring that occurred ahead of Thanksgiving. As a result, some of November's hiring could be counted in December, when the seasonal adjustment won't be calibrated to cancel it out.

Also, the infamous birth/death index fudge factor that the BlS uses, and which tends to add about 100,000 to the employment number in most months, showed a decline of 8,000 in November.

Bottom line, the raw unemployment numbers are only very rough approximations in the first place. With the BLS massaging the numbers, they start to look like a concoction put together by Dali. In most months, the numbers tend to distort things to the upside, November is one of  the months that often distorts things to the downside.

No comments:

Post a Comment