September Jobs Report

There's some unusual news in the area of unemployment statistics. First, in late September, the BLS issued a preliminary revision to their March 2011 – March 2012 numbers; they underestimated job growth by about 20%. In a way this is news only a statistician could love, the effect of these job gains won't change because the measurement is more accurate. And it does make the apparent slowing down of the economy look more pronounced. But it's still good news that we weren't doing as poorly as thought. The figure below shows the gains since March starting from where the revision put the total increase in jobs ending March 2012. 

The news for September in particular is complicated. The increase in total non-farm jobs was reported at 114,000. A lame number, not terrible but not any good either. The numbers would be terrible if the government was still shedding jobs, instead government employment is creeping up. On the good side the numbers for July and August were revised upwards by a total of 86,000 jobs. 

Further complicating things is the fact that the unemployment rate dropped, from 8.1% to 7.8%, and not from people leaving the labor force. It is the first time the unemployment was below 8% since January 2009. The unemployment rate is calculated using a different survey, of households rather than businesses. The household survey shows an astonishing 873,000 more jobs than in August. The household survey numbers are volatile, and the two numbers often diverge, but over time are similar in aggregate. As Floyd Norris (Chief Financial Correspondent for the New York Times) points out,
“A year ago, the the establishment numbers were looking better than the household numbers. Now the reverse is true...Over the last 24 months, however, the two reports are virtually identical, showing an additional 3.6 million jobs and workers." 

I really don't have the expertise to shed any more light on the confusing batch of statistics today. But the take away message seems to be that the economy is and was doing better than we thought, but it's still slowing down.