I've read good pieces in the past 24 hours from Felix Salmon, Paul Krugman, and Megan McArdle on the plight of the long-term unemployed all inspired by Rand Ghayad's excellent field experiment showing that employers discriminate against the resumes of the long-term jobless even when they're otherwise identical. I've seen a bunch of people with varying political perspectives engage in various forms of theoretical or rhetorical overreach in order to explain this, but I think it's a pretty basic case of statistical discrimination.
Suppose you had to hire one of two candidates for a job, and you had to base the decision solely on a resume. No interview allowed. The resumes are identical, but one person lost her job in a mass layoff event last week, while the other lost her job in a mass layoff event a year ago. Who are you going to hire? If you're smart, you hire the woman who lost her job last week. You're being asked to make a decision based on very little information. By discriminating against the long-term unemployed candidate, you can in effect "outsource" your decision-making. Most likely this woman has interviewed for several jobs since being laid off. If she's still unemployed, there's probably something wrong with her. What? You don't know. You don't have any evidence. But faced with the need to decide under conditions of severe uncertainty it's a sound heuristic.
The real world is more complicated than that, but not all that complicated. Your time as a manager is finite and valuable. You don't want to call back every resume that comes in over the transom. You set yourself a target quantity of "good" resumes you want to identify to call. Then you start your search by assuming that every single person who's been jobless for over a year doesn't withstand scrutiny upon interviewing, and just search for good resumes among the short-term jobless. If you hit your target, then you call those people. It's only if you don't hit your target that you start looking at the resumes of the long-term unemployed.
This is wildly unfair, but it makes perfect sense and it seems unlikely that any kind of rule to ban it would actually halt the use of discriminatory heuristics. In fact, it's not clear to me at all what can be done about this problem. The historical example of a solution that worked was World War II. The urgent need to mobilize all able-bodied individuals as either soldiers or war producers of war materiel essentially "reset" the clock and meant that everyone entered the post-war labor market with their Depression-era situation wiped away. But while World War II was a really effective jobs program across a whole range of dimensions, it's not obviously replicable. A WPA-style direct government employment program might help, but it would still mark its beneficiaries as having been rejected by the private sector labor market. The genius of an enormous war is that it swept everyone up.
One takeaway I have is that this underscores how scandalous it is that America still hasn't done anything to prevent the next occurrence of a "zero bound" episode. What's becoming clear is that a three-year span of weakness in the labor market is much much much worse for society than a one-year span of weakness. A normal recession disrupts people's lives, but a long recession destroys them. You lose output, prosperity, family stability, self-esteem, and many other qualities on what looks to be a semi-permanent basis. But instead of recognizing an urgent need to develop a politically tractable strategy for the next time, policymakers seem largely focused on congratulating themselves for having avoided a situation as severe as the Great Depression.