Friday, October 03, 2008

Unemployment Report - The Devil's in the Details...

So the consensus estimate was for a decline in September payrolls of 100,000 jobs and the actual job losses were 159,000 (almost 60% worse than expectations and the worst jobs number in 5 years), but a couple of statistical anomalies allowed the jobless rate to stay at a headline number of 6.1% so everyone seems to be saying "Well, it could have been worse!!".

1) I'll start with some second grade math. The unemployment rate in August was 6.05% (rounded to 6.1%) and the unemployment rate in September was 6.12% (rounded to 6.1%). So while, it's true that the rate was unchanged it's clear that even using their flawed data the unemployment rate went up last month.

2) Birth/Death Adjustment - In brief, the birth/death adjustment estimates how many new jobs were created by companies too new and/or too small to participate in the employment survey. In essence, this is a standard statistical tool. However, at the end of the growth cycle (probably around 2006) the birth/death adjustment imagines substantially more job creation than exists (in fact, lately, it's been all birth and no death).

Without any statistical background, if I told you that the government statisticians estimated that the financial industry ADDED 8,000 jobs that weren't measured, what would you think? How about construction? They estimate 12,000 new jobs added that weren't picked up the survey. That's just plain silly. In fact, I think that you could argue that these numbers should have been negative.

Given the highly politicized nature of all Government agencies, I think it is very telling that this is the last employment report that will be issued before the election. We'll see if the Birth/Death number reverts to reality after the election. Please remove your tinfoil hat now.

No comments: