Thursday, February 04, 2010

Births and deaths 

A story bubbling around the econoblogosphere (you like that? I think we should coin it) is an expected change in reporting the number of jobs lost since April 2008, due to a bad forecast in the number of new firms that adjusts the payroll employment survey figure reported by the Bureau of Labor Statistics. You can get an illustration from Bloomberg to see the issue. CNN is also reporting:

Job losses during the recession may have been underestimated by close to a million jobs. So instead of employers cutting just over 7 million jobs from their payrolls since the economic downturn began in December 2007, it's expected that the Labor Department's new estimate will be a loss of 8 million jobs.

"It's an enormous understatement of the severity of the crisis," said Heidi Shierholz, labor economist with the Economic Policy Institute, a union-supported think tank. "It confirms that things were actually worse on the ground than what the reports suggested."

The new reading will come when the economists at the department's Bureau of Labor Statistics release their annual revision of U.S. payrolls from April 2008 through March of 2009 Friday, using data that wasn't available as the monthly readings were being estimated and reported.

...There is a concern that this problem didn't end in March of 2009. In fact, the adjustment added even more jobs -- 990,000 -- in the nine months reported since then.

So another big revision in the payroll numbers could be looming a year from now. That means this Friday's report should give pause to anyone who is depending on the official numbers to signal real improvement in the economy.

Mike Shedlock says the birth/death model is broken (if so, how would someone fix it? Hard to say.) Casey Mulligan thinks this brings the data from the two employment surveys (households and payroll) into closer alignment. He has argued for awhile that the household data may be better to use right now, which I think is a minority position among economists.

The important issue to remember is that if birth/death misses turning points we may see an upward revision in a few years that offsets this one. The data for this one covers most of 2008 and a few months of 2009, so watch and reprimand those who want to use this as an argument against Porkulus. This data revision ends about when Porkulus starts. But another revision in February 2011 will possibly increase the size of job losses experienced since its passage. For now, we just don't know.