Weekly jobless claims spike up 24K to 399K

That’s two big spikes in a row for weekly initial jobless claims, although this one has a twist.  Last week, the announced number of initial jobless claims was 381K.  Today’s report shows a further bounce upward to 399K, but the gap is wider thanks to a rare downward revision in the previous week’s figures (see update below):

In the week ending January 7, the advance figure for seasonally adjusted initial claims was 399,000, an increase of 24,000 from the previous week’s revised figure of 375,000. The 4-week moving average was 381,750, an increase of 7,750 from the previous week’s revised average of 374,000.

The advance seasonally adjusted insured unemployment rate was 2.9 percent for the week ending December 31, unchanged from the prior week’s revised rate.

The advance number for seasonally adjusted insured unemployment during the week ending December 31, was 3,628,000, an increase of 19,000 from the preceding week’s revised level of 3,609,000. The 4-week moving average was 3,605,000, unchanged from the preceding week’s revised average.

This year, there have been some rumblings from economic analysts that the Department of Labor’s seasonal adjustments are overly aggressive.  Jay Cost wrote about this on Tuesday, and produced a nifty chart detailing the problem, which appears specific to December:

As I noted last Friday, there was a quirky gain of 40,000 jobs in delivery services like UPS and FedEx. This will be “given back” next month, in all likelihood.

More broadly, per Cardiff Garcia of the Financial Times and BizzyBlog, there has been a strange development in the seasonal adjustments to holiday economic numbers since the start of the current downturn.

The economic data that the media reports is typically adjusted for seasonal factors to give us a sense of the underlying trends in the economy. However, the economic collapse in the fourth quarter of 2008 might have been partially interpreted by the statistical models that produce the adjustments as a change in seasonal patterns. So, the models may now be adding a larger seasonal correction than they should be. As we can see here, it makes a difference in the numbers that the media reports. …

The formula that calculates the seasonally adjusted number started adding an extra 100,000 or so jobs to the December adjustment factor in 2008. If we take that out, we would have an actual print closer to 140,000 jobs, which (again) is right in line with the annual average.

That could be leading to a larger-than-expected seasonal impact on this weekly metric on initial claims, too.

Thanks to this statistical tweaking, even Reuters can’t ignore the significant shift in jobless claims:

The number of Americans applying for first-time jobless benefits rose on Thursday, reversing a recent decline and suggesting the labor market remains brittle.

Unemployment claims jumped to 399,000 in the first week of 2012, the highest in six weeks, from an upwardly revised 375,000 in the prior week. The four-week average of claims also marched higher to 381,750 from 374,000. …

The unemployment rate has fallen sharply in recent months and was 8.5 percent December, but some economists worry the drop has been due in part to discouraged workers dropping out of the labor force.

Not even in part.  If Jay Cost’s calculations are accurate, then we only added just enough jobs in December to barely stay ahead of population growth, which means that the entirety of the drop would be due to a decline in the work force.

Update: If you followed the link, you’ve probably already realized my mistake.  The 381K figure was from two weeks ago, not last week.  The revision from last week was upward by 3K.  Thanks to Steve Eggleston for the correction.

Blowback