Abelson's Employment Error

When I hit the road for a week of business travel, I was confident that market pundits would eventually sort out the mixed message from the September Employment Situation Report.  September was light, August was revised up dramatically, and there was a huge "benchmark" revision for job growth.  I noted that Alan Abelson's column the day after the report was (strangely) silent on this topic.  Apparently it took him a week to round up the usual suspects and come up with his usual bearish spin on the numbers.

After working on this for a week, he came up a truly awful analysis of the data.  Check it out here (subscription required).  Here is the key section:

"...maybe the problem is not with the boom's visibility, but the data that provide the basis for its belated discovery. Maybe, in short, the boom uncovered by the diligent men and women who man (and woman) the Bureau of Labor Statistics is itself simply a miscalculation.

That possibility begins to seem not quite so outlandish when you peer at the text of the bureau's September employment report. More specifically, run your orbs over that part of the "Explanatory Note" that deals with "reliability of the estimates" on page 8. (We're indebted to several keen-eyed Street scholars, including Barry Ritholtz and, inevitably, Stephanie Pomboy, for supplying chapter and verse on the somewhat startling material in their recent commentaries.) If, by some remote chance, you don't have the document handy, here's the relevant passage:

"Statistics based on the household and establishment surveys are subject to both sampling and nonsampling error...the confidence interval [range] for the monthly change in total employment is on the order of plus or minus 430,000. Suppose this estimate of total employment increased by 100,000 from one month to the next. The 90-percent confidence interval on the monthly change would range from -330,000 to 530,000 (100,000 +/- 430,000). The figures do not mean that the sample results are off by these magnitudes, but rather that there is about a 90-percent chance that the 'true' over-the-month change lies within this interval. Since this range includes values of less than zero, we could not say with confidence that employment has increased."

If Abelson would read Gene Epstein's work in his own publication, he would understand that he was making two serious errors.

1. The benchmark revisions are based upon an actual count -- a census of those reporting jobs for unemployment insurance as legally required.  This is as close as we get to a true count of jobs, but it only tells us what happened through March, 2006.  The benchmark revision is designed to address non-sampling error. 
2. He dredged up a footnote related to sampling error in the so-called household survey.  He then pulled this out of context and applied it to part of the methodology where it was not relevant at all.

The truth is that the benchmark revision is the best data we have.  The reason it was so large is that the BLS "birth/death" model under-estimated the number of new jobs created in the last year.  The birth/death adjustment is based upon the last five years of data, so the result suggests that the last year of the recovery was spawning new businesses faster than the prior five years.

Barry Ritholtz summarized a number of my fellow bloggers commenting on employment.  None of them, including Barry, distinguish between survey data (involve sampling error and non-sampling error) and the actual count from the benchmark revision.  Apparently he overlooked my analysis of the report, where I expressed confidence that he would clarify this for his readers.

Since hundreds of thousands read Abelson, and tens of thousands read Barry Ritholtz, the thousand or so who read "A Dash" can gain a significant contrarian advantage--just by understanding the data!  The facts are that the economy, as measured by employment, has been much stronger than the "bear cult" has portrayed it to be.  These data are consistent with the record corporate earnings, higher than expected tax receipts, and consumer spending.

Comments

[Mark T]

Alan Abelson should have read Gene Epstein properly (at all?) at any time over the last five years. He is the high priest of the bear cult and very selective in his data. Barry has been kidnapped by his comments section and suffered Stockholm syndrome.