In his coverage of the Department of Labor's Unemployment Insurance Weekly Claims Report at the Associated Press this morning, economics writer Christopher Rugaber stubbornly referenced a supposedly predictive benchmark the wire service has been using which has consistently failed in recent months.
Rugaber also claimed that today's seasonally adjusted increase from the previous week, which will almost certainly become a bigger one after next week's revision, is "evidence that the job market's recovery remains modest and uneven." Uh, not exactly. Excerpts follow (bolds and numbered tags are mine):
APPLICATIONS FOR US UNEMPLOYMENT AID UP SLIGHTLY
The number of people seeking first-time unemployment benefits rose a slight 4,000 last week to a seasonally adjusted 372,000, evidence that the job market's recovery remains modest and uneven. 
The Labor Department said Thursday that the four-week average, a less volatile measure, increased 3,750 to 368,000.
Applications are a measure of the pace of layoffs. When they fall consistently below 375,000, it generally suggests hiring is strong enough to lower the unemployment rate. 
... The increases "suggest ... that job growth continued in August, but at a slower rate than July's pop,"  said Jennifer Lee, an economist at BMO Capital Markets. "So job growth there shall be, but not strong enough."
Weak hiring may prompt the Federal Reserve to take more action to stimulate the economy, Lee said.
 -- The unemployment rate was 8.1% in April, rose to 8.2% in May, stayed there in June, and rose to 8.3% in July. I have yet to find a planet on which this kind of increase in an already unacceptable unemployment rate can be described as being part of any kind of "recovery," modest or otherwise. That's before considering other elements of the job market, including a disturbing four-month trend away from full-time employment and disproportionate use of temps (over 800,000 temps added since the recession ended, making up 24% of the 3.4 million new private sector jobs during that period).
 -- This supposedly predictive 375,000 threshold for weekly claims, below which the unemployment rate is supposed to come down, isn't working, as seen in the following graph:
So in April, claims were largely above the 375K threshold; the unemployment rate didn't go up as would be predicted, but instead went down. Most of May, up to the probable last week DOL finished its data collection, was below the threshold; yet the unemployment rate went up. June's numbers are mostly above the threshold, and the unemployment rate stayed the same, representing the only example of the four months involved which could be construed to fit the supposely predictive benchmark. That's because July's numbers were below the threshold, but the unemployment rate went up.
I would suggest that the business press needs to find a better predictive benchmark -- or return to the one Rugaber was using three years ago, namely 325,000 weekly claims. (Given population growth, 350,000 might be acceptable. 375,000 clearly isn't.)
 -- Lord have mercy. July was not a "hiring pop," despite what Rugaber's quoted source says. It was an inconclusive popgun:
As seen, 1.204 million jobs were actually lost. That's fewer than the previous two years but more than July of 2004, 2005, and 2006. The seasonally adjusted conversion to 163,000 jobs gained isn't unreasonable in light of the conversions seen when the economy wasn't in recession. But the result wasn't a "pop" by any means. In fact, it's barely better than the 152,000-job average of the previous 21 months.
Rugaber's report was a disappointing performance all around. I don't think I'm out of line in suggesting that the public really, really wants and expects wire service reporters to stick to the facts (and as a result have a better shot of getting all of them right) without trying to tell us what it's all supposed to mean (often erroneously). But the folks at the Associated Press, aka the Administration's Press, just ... won't ... do it.
Cross-posted at BizzyBlog.com.