U.S. Private Sector Employment Contracting

As a further confirmation of the fact that the U.S. economy has slipped into a recession was today's employment numbers which showed falling private sector employment even according to the payroll survey, despite the continued use of the misleading "birth/death model".

What few commentators have noticed, however, is the fact that not only did private sector employment fall compared to November, there is even good reason to believe that private sector employment fell compared (albeit only marginally) to December 2006!

The household employment survey showed a dramatic drop (-436,000) in employment and an increase in the unemployment rate from 4.7% to 5.0%. As I've stated previously, the monthly change in the household survey is very volatile and erratic, so one shouldn't be concerned about the monthly fluctuations. However, as an indicator of annual trends, the household survey is superior to the payroll survey because it isn't distorted by the "birth/death model". And if you look at the annual change, employment according to the household survey is up by just 262,000 compared to December 2006.

If you subtract from that the 274,000 increase in government employment, you end up with 12,000 fewer private sector jobs.

The payroll survey by contrast claims that employment rose by 1.328 million compared to December 2006. Which implies a gain of 1.054 million private sector jobs if you subtract the increase in government employment. However, more than 100% of that increase is simply assumed through the "birth/death model". That model added 1.13 million private sector jobs in 2007, meaning that actually reported private payroll employment fell by 76,000 compared to December 2006!

Thus, there is good reason to believe that the annual change of private sector employment is now negative.

And if you adjust the monthly change in December for the "birth death model" you end up with a monthly contraction in private sector employment of 107,000 and a total contraction in employment of 76,000 (As I explained last month, because the birth/death model monthly number is not seasonally adjusted, you cannot subtract that number from the seasonally adjusted change. Instead you should subtract the monthly average number of 94,000 (1.13 million/12)).

This means that had payroll employment been properly measured, we would have seen another of the NBER's recession indicators indicate that the recession has already started. Already we've seen real disposable income and industrial production fall, now payroll employment is falling too.