Wires Ignore Bad News in Raw Unemployment Claims Data

Today's news from the Department of Labor on initial weekly unemployment claims was supposedly good -- as long as one doesn't scratch beneath the surface. Journalists used to do that. Today they didn't.

All one had to do is reach the third paragraph of DOL's release to realize that today's seasonally adjusted claims number of 343,000, touted as the lowest in two months in several news reports, was suspect. That paragraph told us that the 428,814 actual claims filed during the week ended December 8 were barely lower than the 435,863 claims seen in the week ended December 10, 2011, last year's comparable week; today's result only occurred because this year's seasonal adjustment factor was significantly different from last year's. I believe that this year-over-year drop of less than 2% in raw claims is the smallest weekly difference in a week not affect by storms or holidays this year. In other words, it really is news -- but not in the business press, which runs with the government's seasonally adjusted data and almost never looks any further. Examples follow the jump.

Here's Christopher Rugaber at the Associated Press:

WEEKLY US JOBLESS AID APPLICATIONS DROP TO 343K

The number of Americans seeking unemployment benefits fell sharply for a fourth straight week, a sign that the job market may be improving.

The Labor Department said Thursday that weekly applications for unemployment benefits fell 29,000 last week to a seasonally adjusted 343,000, the lowest in two months. It is the second-lowest total this year.

As shown above, today's tiny year-over-year dip in raw claims make a mockery of Rugaber's assessment. Year-over-year differences in five-day business weeks unaffected by storms have been running at about 7%-10% for most of the past year or so. So how can a difference of less than 2%, which will likely end up being below 1% after next week's virtually inevitable upward revision, be "a sign that the job market may be improving"? Answer: It can't.

Now here's Alex Kowalski at Bloomberg:

Jobless Claims in U.S. Decline to a Nine-Week Low

First-time claims for unemployment insurance payments declined more than forecast last week to the lowest level since early October, adding to evidence the labor market is improving.

Applications for jobless benefits fell by 29,000 to 343,000 in the week ended Dec. 8, the fewest since reaching a four-year low in the period ended Oct. 6, Labor Department figures showed today. Economists forecast 369,000 claims, according to the Bloomberg survey median.

Again, the narrowing of the year-over-year differences in raw claims contradicts the contention that "the labor market is improving." The miss in analyst's predictions demonstrates that most of them aren't even ambitious enough to look at the changes in year-over-year seasonal factors before making their estimates. That is really weak.

At Reuters, Jason Lange went even further:

Jobless claims fall sharply, recover from Sandy

The number of Americans filing new claims for unemployment benefits fell for a fourth straight week last week, pointing to steady healing in the labor market.

Last week's drop left new claims at their lowest since the week of October 6, and well below the levels just before superstorm Sandy, which slammed into the East Coast in late October and triggered several weeks of volatile claims numbers.

Just before superstorm Sandy, during the week ended September 29, raw claims of about 301,000 after upward revisions came in over 9% below the previous year's comparable week figure of almost 332,000. As already noted, this week's raw claims difference was less than 2% and will be probably be about 1% after next week's revision. That's not evidence of "steady healing in the labor market."

As I noted earlier today in a post at my home blog:

If last year’s comparable-week seasonal conversion factor of 117.4 had been used on last week’s raw number instead of the 124.9 actually used, claims would have come in 22,000 higher (428,814 divided by 1.174 is 365,000, rounded).

That 365,000 figure is only about 8,000 fewer seasonally adjusted claims than a year ago. It's amazing to think that the tone of today's coverage, with the same underlying data, would have been totally different if that had been the reported seasonally adjusted result. How ignorant, and absurd.

Cross-posted at BizzyBlog.com.