September jobs data like all the other months gets adjusted for what have been seasonal variations. Some of these adjustments are based on historical patterns and while they made sense when initiated, sometimes don't make sense in unusual economic times.
A lot of adjustments related to what has been historical patterns in the auto industry are simply wrong now. How much these distort the figures is open to debate, but the pattern that used to exist where many auto workers were laid off for a period while the plants retooled for the new model year has been disrupted by all the current chaos in the industry.
The question then is point is served by the use of these seasonally adjusted numbers? At one time, it was felt that the raw variations could provide a false impression as to the health of the economy. If you could predict that x thousands of auto workers let go in July would be hired back in September, the spike in unemployment they represented wasn't considered real. It was also of course politically problematic to have unemployment rates shoot up every summer and then drop again.
Statistically, if you were trying to determine the real structural trend, you had to eliminate these cyclical impacts. However, they require constant adjustment and can distort when times become unusual.
No comments:
Post a Comment