There seems to be no more fertile area for statistical misinterpretation than with unemployment numbers. We saw this last week with the monthly jobless report, in which a 0.4% drop in the topline unemployment rate occurred even though the economy only added 120,000 jobs, which would only cover the average population growth, while almost three times that number dropped out of the workforce. Even the media didn’t really buy that as an improvement, and neither did voters, if one uses the Gallup tracking poll as a measure, and for good reason.
Today, we’re seeing a resurgence of a long-debunked myth with the latest report from the Department of Labor on weekly initial jobless claims. The DoL reported a mild improvement in the level:
In the week ending December 3, the advance figure for seasonally adjusted initial claims was 381,000, a decrease of 23,000 from the previous week’s revised figure of 404,000. The 4-week moving average was 393,250, a decrease of 3,000 from the previous week’s revised average of 396,250.
The advance seasonally adjusted insured unemployment rate was 2.8 percent for the week ending November 26, a decrease of 0.2 percentage point from the prior week’s unrevised rate.
The advance number for seasonally adjusted insured unemployment during the week ending November 26 was 3,583,000, a decrease of 174,000 from the preceding week’s revised level of 3,757,000. The 4-week moving average was 3,667,250, a decrease of 20,500 from the preceding week’s revised average of 3,687,750.
We have been bouncing between 380K and 410K for the last several weeks, which is an improvement over the level we saw this summer of 420K or so. However, this essentially brings us back to the level seen in Q1 before the impact that rising gas prices and the tsunamis in Japan had on the economy. Not coincidentally, the overall economy has returned to about the same level of health, with a stagnation-level 2.0% GDP rate in Q3. It’s an improvement, but only a marginal one, especially in the context of the continued exodus of working-age adults from the workforce and the lowest level of civilian participation in 30 years.
However, that still hasn’t kept media outlets from heralding the 400K level as an indicator of job growth, which is — as I have repeatedly shown — a myth. Reuters offers a mild version of it in its report today, linking it to the jobless report from last week as a sign of momentum:
New claims for unemployment benefitsdropped to a nine-month low last week, a government report showed on Thursday, suggesting the labor market recovery was gaining momentum.
Initial claims for state unemployment benefits fell 23,000 to a seasonally adjusted 381,000, the Labor Department said, the lowest since late February. That was below economists’ expectations for a fall to 395,000.
The report, coming on the heels of data last week showing a rise in hiring and a sharp drop in the unemployment rate to 8.6 percent in November, pointed to some healing in a sector that has been the Achilles heel of the economy’s recovery.
Nowhere in this report does Reuters mention the decline in the labor force, which puts the context of the “drop” in unemployment in a completely different context. National Journal goes a little farther, claiming that going below the 400K level brings the job market into a place where it can “dent” unemployment:
Economists had expected claims to drop from 402,000 to 395,000, according to Reuters. Instead, the previous week was revised up to 404,000 and claims fell by 23,000. The substantial drop puts them safely under the 400,000 level economists say they need to drop below to dent unemployment. The four-week moving average, a less volatile measure of claims, was also below that level at 393,250.
That is simply not so, at least not directly correlated to the 400K level. If anyone needed a check on that, all we have to do is look at the unemployment rates from Q1 when we spent several weeks at the 380K level and the jobless rate never dropped below 9%. Otherwise, we could take a look at the historical correlation between weekly jobless claims and job-growth levels, which show that we have to go much lower to indicate real job creation:
Take a look at the historical series of weekly claims between December 2005 and December 2007, the last time we really had “stability” in the labor force. The highest number in that period was 355,000 in a week, and that was in December 2007 when the economy slid into recession. In fact, between January 2004 and January 2008, we had only two weeksof 400K-level weekly claims, both in September 2005, and they were very much the exception. The average for that four-year span is 326,735, and the median number is 324,000 — which is why I usually use the 325K number in my analyses. We actually didn’t get to the 400K level until July 2008, at which point no one considered the labor market “stable.” …
Think I’m fudging those baseline expectatations by using the supposedly “overheated” Bush economic expansion? Well, take a look at the same series for the four years between 1996 and 1999. The average number of initial jobless claims per week in that period was 321,986, and the median was 317,500. There was exactly one week of 400,000 or more claims in a week, and that took place in January 1996.
The 380K level has never been associated with positive “stability” in the job market or significant job growth, something one would think that the media would have figured out by now.
Join the conversation as a VIP Member