Saturday, October 16, 2004

Wrapping it up on Krugman

I have been getting more feedback and criticism in the comments on my last Krugman post. Some of it is on target, and some of it is not. I'm right about Krugman, but I'm wrong about Ryan. I'll deal with the Krugman part here and the Ryan part in the next post.

On the Krugman part, I think I addressed what the commenter is saying in the three paragraphs of the post beginning with "However, the employment-population ratio can fall ..." I'll try again, with as little ancillary discussion as I can. The commenter writes (with my emphasis in bold added):

You use one piece of anecdotal information to try and refute Krugman, but you misconstrue what he said. He did NOT say that every person who dropped out of the labor force did so because they stopped looking for work. What he said was the drop in the unemployment statistics was only due to a set of people dropping out of the statistics. This can be true even if your wife dropped out of the labor force as you described. The person who took her job may not have been unemployed. That person may have been reentering the labor force so even though the job did not "disappear" no net employment gain happened.
It is true that he did not say exactly that (but nor does my argument require that he said exactly that). The issue, in the commenter's language, is the particular set of people who are dropping out of the statistics. Let's take another look at what Krugman said:

... unemployment declined only because some of those without jobs stopped actively looking for work, and therefore dropped out of the unemployment statistics.
As I said, Krugman's use of the word "only" requires that there were no other causes for the reduction the unemployment rate other than the one he cites, "some of those without jobs stopped actively looking for work." The statement is falsified by any instance in which a person voluntarily (and not as a response to some diminution in his or her labor market opportunities) chooses to retire, raise a family, volunteer, or go back to school, provided that s/he is replaced on the job by an unemployed person who was actively looking for work. Like an unemployed person stopping an active search, a voluntary exit that allows an unemployed person to become employed causes the unemployment rate to decline, the labor force participation rate to fall, and the employment-population ratio to stay the same.

Krugman's use of the word "only" requires that there be no such instances. I think I am on much firmer ground asserting that there is at least one such person than Krugman is in asserting that there are no such people in a labor force of over 140 million people. Nothing else in any of the previous posts is necessary to demonstrate his error.

Thanks again for the comments.

3 comments:

Anonymous said...

If you're going to take Krugman to task for the word "only", I think you're going to have to hold yourself to the same standard, and your phrases "no other causes for the reduction", "falsified by any instance", and "at least one such person" do not meet the standard of accuracy. I think you need to change them to "no other statistically significant causes for the reduction", "falsified by any statistically significant instance" and "at least a statistically statistically significant number of persons".

That aside, I don't happen to believe that, particularly in our current economic circumstances, unemployment rates should be given as much attention as they get. A more valid statistic would consider under-employed individuals (employed but in a lower-earning position than previously), under-compensated individuals (employed in the same position, but with a pay or benefits cut), and self-employed individuals whose businesses are still afloat but are generating less income than before. A statistic on per-capita earned income, adjusted for inflation, would come closer to giving an accurate picture of the health of the employment economy -- although the distribution would also have to be taken into account to really be meaningful.

-richard schwartz
http://smokey.rhs.com/web/blog/rhs.nsf

Andrew said...

Thanks for your comment.

I agree with your first point about the need to acknowledge the statistical imprecision when evidence is presented. I don't think this helps Krugman--he didn't present any evidence, and so the criticism was really more about the logic of his argument.

Your second point is a frustration that I share as well. I may post more on it later.

Thanks again.

Anand Manikutty said...

People are usually able to get out of such issues by claiming "semantic context". For instance, if Krugman had said something like the following, he would be much better off :

Unemployment declined not because the employed fraction of the population increased, not because the economy is doing much better now, not because Bush's policies have encouraged people to get back into the workforce, but only because some of those without jobs stopped actively looking for work, and therefore dropped out of the unemployment statistics.
Here, I am using "only" to dismiss the other reasons. This is semantically permissible, although some people might disagree with me here. In the article you quoted, I don't think Krugman can claim "semantic context". He should have gone into greater detail regarding the reasons for his assertion.

I think what was significant about Krugman's article is that he was presenting some fairly simple, fairly fundamental data about the economy. Those fundamental data themselves, even after correction for bias, are pretty mind-blowing. This is not one of those subtle discussions where the question is whether the budget surplus was caused by over-valued stocks and an overheated economy versus whether it was caused by technology or productivity gains. I wish we were debating reasons why the economy is doing so well, as opposed to why it is so messed up. I wish it were 2000 again!

I think the best part about blogs is the possibility of discussion. I found a fairly subtle error in one of Hal Varian's articles on outsourcing in the New York Times. There, I talked about a scenario which violates Varian's statements on why outsourcing works.

http://delphicoracle.blogdrive.com/archive/cm-03_cy-2004_m-03_d-21_y-2004_o-0.html

I was surprised a few months later when a very similar scenario was played out in real life.

http://wetware.blogspot.com/2004/07/75-hour-work-week.html

I think the great thing about blogging is certainly the opportunity for discussion, analysis and correction. Blog on!