On Jan 12, 2009, at 6:37 PM, Patrick Bond wrote:
> Who's this guy John Williams?: "Noted statistician John Williams
> (shadowstats.com) reports that biases in measurement have
> understated the job loss over the last 12 months by 1,150,000 jobs.
> Williams reports the unemployment rate as it was measured prior to
> “reforms” designed to minimize the measured rate of unemployment.
> According to the methodology used in 1980, the US unemployment rate
> in December 2008 reached 17.5 percent."
Williams is totally unreliable. I really know a lot about how these employment numbers are put together - it's a major part of my day job. This quoted passage is just nonsense. Changes in methodology may have contributed a few tenths of a point, but no more. The burden of proof is on someone making such outlandish claims.
And the figure on job loss has little to do with the way unemployment is calculated. Figures on job loss usually come from the so-called establishment survey, a survey of some 300,000 employers. Unemployment comes from a survey of 60,000 households.
It's quite possible that the rate of job loss in the establishment survey has been understated. Tomorrow at 10 we'll get a sense of this when the Quarterly Census of Employment and Wages is released for the second quarter. The QCEW figures are based on the 99% coverage of the employment universe provided by the unemployment insurance system. There are estimates imputed into the monthly establishment survey to account for new establishments that can't be covered in the sample. Most of the time, these imputations improve the accuracy of the stats (when they're benchmarked against the unemployment insurance records every year). They're wrong at turning points, though, understating job growth in expansions and understating job loss in contractions. There could be a sharp downward revision of the second quarter data - we'll see tomorrow. But the source for this info is the same agency, the BLS, that's demonized by hacks like Williams.
Doug