Explaining The Government's 1.8 Million Job Overestimation In Pictures
Last October the BLS announced it would revise historical payrolls lower by 824,000 on February 5 (this Friday's NFP release). While this number will not impact the actual January NFP report (a loss of nearly one million jobs in a month would probably even take out the persistent SPY algo that has been hugging the bid for the past 10 months), it will be prorated across all months in the 2008-2009 reporting period. The reason for this adjustment has to do with a huge glitch in the birth-death model, which is exactly the same problem that the rating agencies faced when housing prices plummeted: the birth/death model assumes, in the long-run, jobs are created, not destroyed. Any period of excess volatility in the stock market therefore translates into major prior downward revisions to already disclosed payrolls. And while we know what the current revision will be, the scarier prospect is that the next historical adjustment, due out in early 2011, will be even larger, at least 990,000. This means that the government has overrepresented running payroll data by over 1.8 million jobs over the past 20 months.
Bloomberg has prepared a wonderful interactive presentation explaining just how the BLS is full of itself, absent the L.
First, here is what we know will be the BLS adjustment on Friday.
Second, the reason for the adjustment has to do with the great recession, which having run for over 2 years now is still in no way abating (contrary to what you may be hearing in other still GE-contolled media outlets).
Lastly, and most notably, the number that will have to be whacked from the payroll report in 2011 retroactively is even larger: according to Bloomberg it will be at least 990,000. And this is only for 9 months in the current period.
In essence, a Moody's-like glitch has misrepresented the true payroll picture due to modelling error to the tune of over 1.8 million jobs. How that will impact the president's "jobs saved or created" calculation has yet to be determined. Unless of course all those jobs appear merely on the same excel file that the BLS uses for all it other erroneous calculations.