Lost Jobs in Recessions

The WSJ has a nice article showing just how hard it has been for many people who lost jobs in the recession to get back to work. Their profile is typical of what I have read and not the typical picture of unemployment: Middle age middle managers. The paper by Steve Davis and Till von Wachter is here. They present the fact largely as a puzzle, which it is:  "losses in the model vary little with aggregate conditions at the time of displacement, unlike the pattern in the data."

As the story makes clear, the problem is really not unemployment. There are lots of jobs available. The jobs just don't pay much, and don't use the specialized skills that the workers have to offer. The problem is wages at the jobs they can get.

This is a very interesting fact, with many less than obvious interpretations. It strikes me as a good teaching moment for economics classes.

The natural interpretation of all correlations is causal: There are  two identical workers in two identical jobs at two identical companies. One worker happened to lose his or her job in a recession, and so faces a harder climb back. We learn about the difference in job markets over time.

Maybe, but the job of being an economist is to recognize lots of other possibilities for a correlation. So the proposed discussion question: what else might this mean? How does taking averages reflect selection rather than cause?

Perhaps not all workers are the same. The conventional view of recessions is that companies fire people from lack of "aggregate demand," or shocks external to the firm.  In good times, companies fire people when those people aren't very good. Then, you would think, being laid off in a recession is better than being laid off in good times. If you're laid off in good times that is a signal you're not a great worker. In a recession, everybody got laid off, so there is not any particular stigma in it.  Well, so much for that story.

A contrary story is that it's easier to get rid of people in a recession. The head of a large business once told me how useful the last recession was, as he could plead financial problems and finally get rid of the army of unionized workers that were playing solitaire all day. Guido Menzio  and Mikhail Golosov have a model that (I think!) formalizes this story. (Menzio was recently in the news, as an idiot fellow passenger thought he was a terrorist because he was doing algebra on a plane, a different sad commentary on contemporary America.)

Perhaps not all businesses are the same. Businesses and occupations that get hit in recessions are different from those that get hit in booms...

Perhaps times are not the same. Recessions are pretty much by definition a time when different sorts of shocks hit the economy. If recession shocks require bigger changes in specialized human capital than normal-times (more idosyncratic shocks), or people to move industries and cities more, then you'll see this pattern.

And so on. Interesting facts, not so obvious interpretations, averages that don't always mean what you think they mean, that's why economics is so fun.

Update:  Steve Davis writes to explain that job losses in recessions are concentrated in specific industries:

You write: "...If recession shocks require bigger changes in specialized human capital than normal-times (more idiosyncratic shocks), or people to move industries and cities more, then you'll see this pattern.” 

Here’s a modified version of this story that has more promise in my view.  First, an under appreciated empirical observation: The cross-industry (cross-firm, cross-establishment) distribution of employment growth rates becomes more negatively skewed in recessionary periods.  Job loss is also concentrated in industries (firms, establishments) that experience relatively large net and gross job destruction rates.  Taken together, these two observations tell us that, in recessions, a larger share of job losers hail from industries (firms, establishments) that get hit by especially large negative shocks (even compared to the average), reducing the value of skills utilized by workers in those industries (firms, establishments).  I conjecture that negative skewness in the cross-occupation distribution of employment growth rates is also counter cyclical, but I don’t recall any direct and convincing evidence on that score. 

Restating, the setting in which job loss occurs worsens for the average job loser in recessions, because (1) overall economic conditions worsen in recessions, AND (2) conditions worsen especially for industries (occupations, etc.) with a disproportionate share of job loss. Many models consider the effects of (1), but there is little work on (2).  Testing hypotheses and building theories related to (2) requires good measures of the individual-specific “setting” in which individual job losses occur.  One of my PhD students, Claudia Macaluso, is making good progress on that front in her dissertation.

William Carrington and Bruce Fallick have a review paper on why earnings fall with job displacement.