John Williams (Shadow Government Statistics) raises an interesting issue I had not considered: He contends that the Household Survey is more reliable than the Establishment Survey, and I wonder if he isn’t on to something.
Some years ago, I would have argued against it. The more stay-at-home, self-described independent contractors you survey, the more inaccurate your results are going to be. I do not know a single person who lost thier job over the past 3 years who, when asked, did not describe themselves in that manner. It is a matter of pride.
But Williams argues the Household survey sampling is now fairly well established and, significantly, the unadjusted raw data are not revised.
Consider this: In the pre-Bush years, the Establishment Survey (CES) actually measured employment via real paychecks that were actual wages. The B/D adjustment was tiny. But since 2003, the CES is no longer a pure measure; rather, it is only part measurement, and part conjecture, working off of new state incorporation filings.
The Birth/Death Model is now the tail that wags the dog.
Here’s Williams via Abelson:
“John Williams dismisses the consensus outlook and hype from Wall Street and Washington about improving economic conditions as “irrationally optimistic.” And he feels that “with the constraints on broad systemic liquidity still tightening, unhappy surprises are likely in that area as well.”
It seems like forever that we’ve been delivering a monthly rant occasioned by the release of the data of the so-called birth/death model, launched during the Bush administration and extended by the current one, that is supposed to capture the employment additions of new firms and the subtractions of those that go belly up. As it happens, the model is a bust during recessions (as a modest example, it created 34,000 mythical jobs last month).
Comes now the preliminary revision of the March 2009 benchmark, and it turns out that, in fact, courtesy of the birth/death model, that month’s payrolls were overstated by a mere 824,000. As Philippa Dunne and Doug Henwood, proprietors of the Liscio Report note, since job losses in the first quarter of this year were already reported at 2.1 million, “adding the better part of a million to that suggests a truly savage bloodletting in early 2009.”
They also point out that September was the 21st month in a row of shrinking employment, the longest losing streak since the monthly numbers started being published back in 1939. It’s also the worst decline — even without the benchmark revision — since the post-World War II demobilization. The sorry consequence of the severe damage wrought by the recession, combined with the weakness of the 2002-07 expansion, is that private employment is now 2.6% below where it was at its December 2000 peak.
And (you can almost hear them sigh) as Philippa and Doug observe, “We’ve never seen anything remotely like that kind of long-term carnage” in poring over 70 years’ worth of monthly stats.
Yeah, yeah, we know, employment is a lagging indicator. Not this time, buddy.”
The problem with “lagging indicators” is every idiot in the world seems to use that as an excuse when making their early, incorrect call that the cycle has turned. Just because something is lagging — and is still poor — doesn’t mean the economic cycle has turned yet.
As Rosie likes to point out, this wasn’t your run of the mill manufacturing recession; it was a full blown credit collapse/housing bust/bank failure cycle. Comparisons to earlier manufacturing recessions will yield parallels that are utterly inappropriate . . .
Barron’s October 5, 2009