Sunday, 20 January 2008

John Maudlin on Jobless Claims

A few days ago I commented on the drop in US initial jobless claims and suggested that, if we see a few more weeks of low numbers we could be in for an upward revision to payroll numbers on Feb 1st. However as John Mauldin points out in his weekly newsletter initial claims are actually on the rise.

Every week I get an analysis of the claims data from long-time reader John Vogel. Mostly it is not very exciting, but he does a very thorough job of examining the actual data and comparing it to previous years. Stick with me here as I run through a few numbers.

Last week there were actually 521,280 initial claims. That number rose to 547,637 this week. So why didn't the seasonally adjusted number rise? Because in week three of previous years the number dropped, often considerably. In 2007 the number was essentially flat. But in 2006, the drop in the third week was 116,000 and in 2005 it was 226,000. There were also big drops in 2004 and 2003, 187,000 and 172,000 respectively.

So, when you smooth the number out by making seasonal adjustments, you expect a large drop in week three from week two. Except that we did not get that drop, we got a rise of 26,000, which is clearly not the trend for the last five years. That also squares with last week's employment survey which shows job weakness.

The bottom line being;

What it really means is that the BLS numbers should be taken with a huge grain of salt around periods when the economy is changing, as it is now. And using them to make a case that the economy is not weakening, as a number of pundits did, could be considered misleading.

I feel a little silly actually, falling for this rookies mistake of not looking closer at the seasonally adjusted numbers. However it does make me more comfortable with my view that the US labour market continues to deteriorate.