Friday, April 04, 2008
Even Worse Than It Looks?
by Tom Bozzo
The U.S. employment report for March, as most of you probably know, stank. But it would have stank more but for this model, which estimated that there were an additional 28,000 jobs in the construction sector, 7,000 in manufacturing, and 6,000 in financial activities jobs. I don't think so.
One of the things that keeps applied economists busy is figuring out whether Sophistimacated Methods are better than simpler ones. In this case the model seems to perform worst exactly at the turns in the business cycle when you'd think policy-makers would want the most accurate reading on employment changes, which is not a great recommendation.
If someone wanted to make the argument to me that the CES Net Birth/Death Model is more trouble than it's worth, today would be a good day to do so.
The U.S. employment report for March, as most of you probably know, stank. But it would have stank more but for this model, which estimated that there were an additional 28,000 jobs in the construction sector, 7,000 in manufacturing, and 6,000 in financial activities jobs. I don't think so.
One of the things that keeps applied economists busy is figuring out whether Sophistimacated Methods are better than simpler ones. In this case the model seems to perform worst exactly at the turns in the business cycle when you'd think policy-makers would want the most accurate reading on employment changes, which is not a great recommendation.
Comments:
<< Home
That's a feature, not a bug. (Think CPI without food and energy, and without a CPI that considers them on a rolling- or SA basis.)
Post a Comment
<< Home
