Here is another comment on Krugman. It is exactly the maximum length he allows (and three periods are followed by single spaces to get the counter to 0 from -3 — I might as well tweet).
His excellent post is here.
He links to an even better post by Matt O’Brien at wonkblog. It’s depressing, but you really should read it.
My comment is after the jump.
I think the error which you and O’Brien denounce might be part of the problem and not just because it is an argument against stimulus. Abandon the standard assumption that, when acting as economic agents, people are rational, even if they are irrational when acting as commentators and policy makers.
One problem for the long term unemployed is that potential employers might believe that there must be something ineffably wrong with the long term unemployed. All that is needed is for potential employers to think that previous potential employers intuited some ineffable problem which the current potential employer missed. The assumed employer intuition must be assumed (by employers) to be less than 100% correlated across employers.
This socially costly error can be the result of employers being accidental theories. Back in 2000, an employer noticed that the guy they hired who had been unemployed for 8 months lacked sticktoativeness. He crunched the number and concluded that the long term unemployed are bad hires. He assumed that the value of the information about unemployment duration is a constant so the long term unemployed are irrationally stigmatized vastly reducing average well being.
The unemployed might make the same mistake. A cost of unemployment is it makes them believe they are failures. If an unemployed person decides that he must have some characteristic which prevents him from being hired, he might be irrationally discouraged reducing re-employment hazard.