Can We Believe Last Week’s Big Drop In Jobless Claims?

Maybe not, some analysts warn. As reported earlier today, new filings for unemployment benefits declined last week to the lowest level since January 2008. On its face, the data suggests a big improvement in the dynamics of the labor market. But in the hours after the numbers were released, questions about the validity of the report have been flying far and wide.


Stephen Stanley of Pierpont Securities, for example, thinks the report is hogwash, according to Fox Business News. Writing in a research note, Stanley charged that “this reading [on jobless claims] is worthless in terms of informing on the general economy.” The Fox Business News story also reported that a Labor Department spokesman

said the numbers were skewed by one large state that underreported its data. The spokesman declined to identify the state, but economists believe California is the only state large enough to have such a significant impact on the overall numbers.

According to the spokesman, the reason that state’s claims numbers fell short was because the state left out a pile of unprocessed claims related to seasonal factors around the beginning of the fourth quarter, which began Oct. 1.

If today’s reported is skewed, we’ll soon see a revision that corrects the distortion. But it would take one hell of an upward revision to reverse the generally favorable trend that’s evident in the recent numbers on a year-over-year basis before seasonal adjustment. This is the measurement that matters if we’re looking for relatively reliable signs from this data series in the context of business cycle analysis. On that score, it’s hard to imagine that we’ll have a complete reversal of fortunes next week, much less a change that turns today’s good news into dark news.
Consider that the unadjusted claims total was 301,073 for the week through September 29—a decline of 9.4% from a year earlier. The following week (via today’s reportedly skewed number in unadjusted terms for the week through Oct 6) reveals a rise from the previous reading to 327,063. That works out to a decline of 19% vs. 12 months previous. Even if we learn next week that today’s number is revised upward by an unusually high 50,000, the year-over-year change would still register a decline to the tune of nearly 8%. Even then, the numbers would still be telling us that the labor market is healing, which implies that recession risk is still low. Indeed, jobless claims have been declining by roughly 10% a year for much of the past year. When that trend reverses, and new claims start rising vs. year-earlier levels, we’ll have a strong signal that the economy is caught up in a cyclical swoon.
Meantime, it’s best to remain cautious about any one data point and instead focus on the year-over-year trend for this series, which has a history of high volatility in the short run. By that standard, the case for expecting more of the same—a modest but steady decline in new claims—is still a reasonable view.
For now, however, we have this chart to consider. It’s subject to revision, of course, but it requires one heck of a pessimistic outlook to expect that we’ll soon see an increase in year-over-year claims data.

None of this would mean much if the rest of the key indicators were sending dark signals about the business cycle, but that’s not the case either (for the details, see my post here). That doesn’t stop anyone from believing what they want to believe. But if you’re interested in what the numbers are telling us, writ large, it’s still early to start planning for the economy’s funeral.

2 Responses to Can We Believe Last Week’s Big Drop In Jobless Claims?

  1. david j michel jr says:

    yes it is early to start planning for the economy’s funeral,obama has not been reelected.

  2. david j michel jr says:

    is this brad delongs post.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>