Saturday, December 08, 2007

Job Report Much Weaker Than It Seems

The employment report supposedly showed according to some commentators that the expansion is continuing, albeit at a more moderate pace, with the payroll survey showing a 94,000 increase in jobs. The household survey reported extremely fast job growth, but this survey is extremely erratic and unreliable with regard to monthly changes. The more reliable year over year change also picked up, but it is still a lot lower than the payroll survey annual increase and it seems likely to drop sharply next month. The truth however, is that the 94,000 number greatly overstate employment growth, particularly private sector employment growth.

Of the 94,000 new jobs, some 30,000 was government jobs, meaning that private sector employment growth was a weak 64,000. However, even this assumes that a lot more new jobs were created in new businesses than destroyed in bankruptcies. These assumptions of the so-called "birth/death model" were made based on conditions in the height of the housing bubble and no adjustments have been made for the fact that the economy has turned down. During the latest year, some 1,128 million new jobs have simply been assumed and not derived from the data. That is a full 92 of private sector employment growth the latest 12 months (As government isn't started or ended, the model only applies to the private sector).
Thus while actually reported private sector employment growth have nearly disappeared and fallen to a mere 98,000 for the latest 12 months, job creation through the start and death of new businesses are assumed to be constant.

Even in the badly hit construction sector, some 140,000 new jobs are simply assumed. As it would be more natural to assume that more businesses have been knocked out than started in the construction sector, this illustrates the misleading nature of the "birth/death model" during cyclical downturns such as the current one.

Many who criticize the "birth/death model" make the mistake of comparing the monthly number to the monthly payroll number. But that is misleading as the monthly "birth/death" is not seasonally adjusted, while the payroll number is. It is more fair to instead derive the seasonally adjusted "birth/death" number by taking the average adjustment of the latest 12 months, which is to say 94,000 (1,128,000/12).

If you then make the reasonable assumption that net job creation through new and ended businesses has developed the same way as actually reported job creation, this means that we can basically zero job creation through that process. Subtract then from the 64,000 number the 94,000 simply assumed jobs, and we get a private sector job loss of 30,000. Even counting in government jobs, and job creation was zero.

A stagnant to negative number is more consistent with the various other indicators of the job market, including the ISM employment indicator, the jobless claim numbers and most importantly, the sharp downturn in tax revenue growth.


Post a Comment

<< Home