This tedious debate is raised every month by administration officials and their apologists, but they keep switching sides -- flip-flopping, if you will -- depending on which set of numbers suits their purposes.
For months, the BLS' establishment survey, which produces the monthly payroll numbers, was showing a dismal labor market, mired in its worst slump since the Great Depression.
But the BLS' household survey, which produces the monthly unemployment numbers, was showing great galloping job gains. Administration officials and their apologists encouraged everyone to pay more attention to this survey, saying it was capturing job growth in new businesses, which were overlooked (so they said) by the payroll survey.
This was widely accepted by many Wall Street economists (many of whom had agendas of their own, to be sure), and even by some in the financial news media.
But then, lo and behold, in the spring of this year, the payroll survey started showing walloping gains, while the household survey showed something of a labor-market slump.
Suddenly, administration officials and their apologists were pointing to the payroll survey as proof of the health of the job market. The once-favored household survey was kicked to the curb, where it joined Iraq's WMD, Bush's plan to go to Mars and other discarded GOP talking points.
But faster than you can say "wishful thinking," the tables turned again, and suddenly the payroll numbers were weak, while the household numbers were strong.
Once again, that old chestnut about the household survey being better than the establishment survey was trotted out, with GOPers pretending it was a pretty new pony. The problem was, that pony had died long, long ago, and the stench of it was getting pretty bad.
The media, to their credit, didn't really buy it, choosing to focus on the weak payroll numbers. More importantly, though, a couple of recent studies have pointed out just how stinky that old horse really is.
First, there was a Fed-related study, the precise origin of which has escaped by addled brain. I'm still trying to find it, and I'll update this post when I do.
Then, today, in the Stock Traders Almanac's latest monthly newsletter (available to subscribers only), Barry Ritholtz, chief market strategist for Maxim Group, helps to further debunk this myth.
He cites the testimony of Alan Greenspan, who said the payroll survey was more accurate. Just because Alan Greenspan says something doesn't make it so, of course.
But he also cites the BLS itself, which recently pointed out [warning: PDF file] that the two surveys are as different as apples and oranges and thus can be consistent, even as they tell of different job-creation totals.
First, the payroll survey is much, much larger, getting results from some 400,000 employers. The household survey only talks to about 60,000 households.
Second, the household survey throws all sorts of jobs into the mix, including farm workers, self-employed and household workers and people on temporary layoffs. The payroll survey doesn't include any of these people.
What's more, both surveys are subject to a host of data goblins, including benchmark revisions to the payroll survey, population controls in the household survey, job-switching by workers, sampling errors, and a lagged accounting for job births in the payroll survey (though there are guesses made about that).
The BLS adjusted the household data by removing those extra folks, and the results are striking -- you get a line that looks almost exactly like the line created by the payroll data.
It's a graphic I'll be saving and trotting out if I ever get in this argument with anybody.
"The argument that the Household Survey more accurately reflects Job creation has been, in our opinion, thoroughly discredited," Ritholtz wrote. "The Household Survey 'excuse' has become the last refuge of economic scoundrels. Employ it at your own risk."
No comments:
Post a Comment