Buried in last week’s payroll data was a huge revision upward by the BLS in job creation over the entire year from March 2005 to March 2006. From Barry Ritholtz at The Big Picture (see here and here):
Yes, we know that 51k new jobs stink; No, it was not a “perfect number” (Attention Mark Zandi: please lay off the Psilocybin before airtime).
Notable beneath the awful headline was the even more astounding adjustment: Payrolls for the 12 months ended in March 2006 will be revised higher by 810,000
Thanks to this adjustment, the BLS now claims that job growth during the 12 months ended in March 2006 was 45 percent higher than previously reported. The revision magically adds payroll employment growth between March 2005 and March 2006 up by a 67,500 per month.
This was the biggest revision since the Labor Department started benchmarking in 1991. To make a comparison, “the aggregate benchmark revisions dating back to 1996 added a whopping 1,555K jobs to the economy, 810K of which (52%) were added during April 2005-March 2006!”
In a follow-up post, Barry quotes the BLS:
“Each year, the Current Employment Statistics (CES) survey data are benchmarked to comprehensive counts of employment for the month of March derived from state unemployment insurance (UI) tax records that nearly all employers are required to file. For national CES series, the annual benchmark revisions over the last 10 years have averaged plus or minus two-tenths of one percent. The preliminary estimate of the benchmark revision for March 2006 is +810,000 (0.6 percent).
BLS currently is researching possible sources for this larger-than-normal expected benchmark revision. On initial review, the difference between the CES sample-based estimates and the UI employment counts does not appear to be concentrated in any one industry or geographic region.”
So, the BLS just discovered that nearly a million jobs were created, between 6 and 18 months ago, that they didn’t know about at the time:
“Researching possible sources” — let’s call that the bureaucratic understatement of the year. The BLS has a benchmark restatement that is 300% of its 10 year average, and they announce: “We are looking into this.”
As Barry says, something just isn’t right:
I suspect that this revision may accidentally reveal a few economic and market truths:
- Is the Fed really data-driven, or do they merely say that because it is more palatable than admitting this is all a seat of the pants exercise?
- Has the equity market rally truly been based on a Goldilocks scenario, or has it been correctly anticipating a bold and robust economy?
- Was the bond market anticipating a slow down? Given all this hiring and economic strength, we have to wonder how real their fears and concerns actually are.
Something strange is afoot at the Circle K: Either the economy is much stronger than we previously believed, ending the Goldilocks scenario — or the data driven Fed is reliant on what we now know to be bad data.
Neither case presents an attractive outcome.
How much you wanna bet that we never see the government come through with a similar ‘absurdly large’ revision to the CPI data?