Sunday, August 09, 2009

Employment data revisited...

First -- sorry for the slow posting lately. I've been switching laptops and it's been more cumbersome than normal - thx MSFT:)

There was much hoopla over the weekend regarding the "falling" unemployment rate. While yes it is true that the absolute reported rate did fall from 9.5% to 9.4% there are two important factors that influenced this number. Logic indicates that if 250k more jobs were lost in July that the unemployment rate should have ticked up but it didn't. Here's why:

1) The rate of unemployment is calculated by taking the number of unemployed persons and dividing it by the number of people in the workforce. Using simple numbers if 2 people are unemployed in a country of 10 people, the unemployment rate is 2 divided by 10 or 20%. Now if one of the unemployed leaves the workforce (retires, stops looking for work, etc) watch the magic of math CUT our unemployment rate --- now 1 out of 9 people are unemployed and the rate of unemployment is now just 11%. Wow, that's a 9% decline!!! Green shoots for everyone!!!

Hmmm, but somethings not right with that. Both of our hypothetical employees are still unemployed we just changed the numerator and denominator in our equation but the reality hasn't changed for workers. This is what happened in the US last month - a little over 400,000 workers were removed from the survey - thus while the number of unemployed actually rose the rate of unemployment fell because the total population of workers fell as well.

2) The second factor at play here deals with historical seasonality and the auto industry. Historically, the auto industry suffers substantial layoffs in July as factories shutdown to retool for new model. The BLS has built a seasonal factor into its models to correct for this planned shutdown. However, in 2009, many of the auto layoffs occurred in May due to the automakers many financial issues. When the BLS adjusted for expected layoffs that didn't occur it created a false impression of job "creation" in July. I'd expect this to be fixed in the August report.

As I've noted, the unemployment report is perhaps the most flawed set of data the government publishes. It is fraught with estimates and assumptions that rarely meet up with reality. This is a positive data point (there were several other small positive indicators - hours worked, hourly earnings, etc) but it's too early to call this a turning point.

In another jobs related subject, the NY Times points out that after 10 years the US has created basically no new non-government jobs.

"For the first time since the Depression, the American economy has added virtually no jobs in the private sector over a 10-year period. The total number of jobs has grown a bit, but that is only because of government hiring.

For the decade, there was a net gain of 121,000 private sector jobs, according to the survey of employers conducted each month by the Bureau of Labor Statistics. In an economy with 109 million such jobs, that indicated an annual growth rate for the 10 years of 0.01 percent.

Health care jobs continued to grow, particularly jobs that involve caring for the elderly. Home health care employment rose at an annual rate of 5 percent, a rate that indicates a total gain of more than 60 percent. On an annual basis, that was twice the overall rate for health care of 2.4 percent a year.

There were also job gains in education and in a host of service industries, including lawyers (0.7 percent a year), accountants (0.9 percent) and computer systems designers (2.4 percent). The field of management and technical consulting leaped at an annual rate of 5 percent.”


No comments: