Press Celebrates Unemployment Claims Drop Below 400K, Ignores Track Record of Upward Revisions

July 28th, 2011 4:16 PM

Two "alert" emails hit my inbox this morning concerning the Department of Labor's just-released unemployment claims report.

The one I expected came from CNNMoney.com, which read: "Initial unemployment claims fall below 400,000 for the first time in more than 3 months, dropping 24,000 to 398,000 in latest week." The other one came from USAToday.com, which does not ordinarily issue alerts when this report appears, took the opportunity to relay the same message, followed by an assertion that today's report is "a sign the job market may be healing after a recent slump."

Over at the Associated Press, Christopher Rugaber joined in the premature e-celebration (possibly more permanent link here):

The number of people seeking unemployment benefits dropped last week to the lowest level since early April, a sign the job market may be healing after a recent slump.

The Labor Department said Thursday that weekly applications fell 24,000 to a seasonally adjusted 398,000. That's the first time applications have fallen below 400,000 in 16 weeks.

The four-week average, a less volatile measure, dropped to 413,750, the lowest since the week of April 23.

Stocks rose after the report was released.

Economists cautioned that the lower level only reflects one week of data and that doesn't necessarily signal a trend.

The two emails, Rugaber's dispatch, and Annalyn Censky's CNNMoney's report (USAT carried Rugaber's work) all engaged in premature e-celebration because they failed to inform readers about the report's history of subsequent-week revisions. In 18 of the past 20 weeks, upward revisions have been 2,000 or more. An increase of that magnitude or greater would of course restore the supposedly broken streak if it takes place next week. The average upward revision for the past 20 weeks from initial announcement to final result has been 4,500:

InitialClaims030511to072311

The folks at Zero Hedge, who have been looking at this trend much longer than yours truly and observing similar results, today found it all "quite amusing as next week's upward revision will mean the 400k+ streak will continue."

Well, not for certain. But just as it would be irresponsible for a sports announcer to report a halftime score as a final, it's just as irresponsible to claim that a streak has ended without identifying its clearly tentative nature -- especially given what's happened during the rest of the game for the past several months.

Cross-posted at BizzyBlog.com.