Friday, July 8, 2011

The Birth/Death Adjustment Was Responsible For Over 50% Of The Payroll "Gains" In The Past Year....

Every time someone brings up the backing out of the birth death adjustment from the Non-Farm Payrolls number (which almost always is additive), economists get all defensive and say you can't really use that number because it is akin to comparing apples to oranges, it is only applicable for the Non-Seasonally Adjusted NFP data (while the numbers that make the headlines are the seasonally adjusted ones) which of course means that the conversion from the NSA to SA number is a product of a multivariate equation with two unknowns (the B/D adjustment and the actual seasonal adjustment). Of course, this is just how the BLS likes it: after all this gives them plausible deniability to blame either X or Y or a combination thereof, but never disclose just what is the culprit for any specific discrepancy. So what does the BLS tell us to do when attempting to eliminate the B/D benefit? Simple: add or subtract the B/D adjustment from the Non-Seasonally Adjusted number. Yet when doing it on a quantized, monthly basis this is impossible due to the layering of the seasonal adjustment.

Well, there is a very simple workaround: just look at the Year over Year change in the Non-Seasonally adjusted numbers. After all, that will eliminate all the intrayear seasonal adjustments, leaving just two clean numbers at the beginning and end of the full year sequence. This works like a charm when looking at June 2010 and June 2011 numbers, on both a Seasonal and Non-Seasonally Adjusted basis. The difference in the NSA series is 1,171K jobs, while the SA is 1,036K, almost a perfect match. And after all we have been hearing for so long how the administration has added 1 million jobs in the past year.

Luckily, now that we have a benchmark that does not need a seasonal adjustment, we can determine precisely what the Birth-Death contribution to the "jobs added" over the past year has been. The result: 606K, or 52% of the NSA jobs added (and 58% of the actual, seasonally adjusted jobs).

And as a reminder, the total Non Seasonally Adjusted jobs added over the same period, a number which encompasses a full calendar year period and thus needs no seasonal adjustment, includes in it this same Birth-Death adjustment.

And as noted above, the difference in the NSA jobs between June 2010 and 2010 is 1.1 million (June 2010: 130,908K, June 2011: 132,079K), and 1.0 million SA (June 2010: 129,981K, June 2011: 131,017K).

So we wonder, how long before someone, anyone, points out that well over half of the job "gains" in the past year have come from a purely statistic fudging of data, based on the same endless growth assumption that destroyed the credibility of the ratings agency excel models, and which has no actual mapping to real jobs?