John Quiggin makes a good point:
Incarceration as a labor market outcome, by John Quiggin: I wasn’t all that surprised that Bryan Caplan didn’t like my interpretation of our bet on EU and US unemployment rates, which was that the combined rates of unemployment and incarceration in the US would exceed those in the EU over the next ten years. I was, however, surprised by the vehemence with which libertarian-inclined* commenters here and at Crooked Timber objected to this interpretation.
A string of them echoed Caplan’s argument that
From a labor market perspective, though, Quiggin’s incarceration adjustment would only make sense if you thought that most or all of the people in jail would be unemployed if they were released.
Caplan has missed my main point. I’m not suggesting that incarceration is disguised unemployment (though obviously it reduces measured unemployment). Rather, I’m saying that, like unemployment, incarceration should be regarded as a (bad) labor market outcome. If you want to evaluate the performance of the labor market, you need to look at both.
There’s nothing radical or leftist about this viewpoint: it’s one that is at least implicit in all economic models of the labor market of which I’m aware, and is most particularly explicit in that of the Chicago School*. Most of the crimes for which people are imprisoned in the US can be understood as reflecting economic choices which in turn are determined primarily by the labor market in which those choices are made. This is obviously true of property crime and drug dealing, and it’s true, directly or indirectly, of lots of violent crime as well. As Gary Becker put it (quoting from memory here) “a burglar is a burglar for the same reasons as I am a professor”. ...
There’s plenty of statistical evidence from scholars like Glenn Loury to show that criminals ... are drawn disproportionately from groups with bad labor market prospects: poor, disproportionately black, facing low wages and high risk of unemployment. But well-done case studies are often more convincing, so I’ll point to the Venkatesh study of Chicago drug dealers reported in Steve Levitt’s Freakonomics. Venkatesh found that most street dealers were making less than minimum wages, and were motivated by the very low probability of surviving to attain the only high-paying job realistically available to them, that of the local kingpin. Even more striking was the observation that, when gang members learned Venkatesh was a university professor, they approached him in the hope that he would be able to wangle them jobs as janitors - otherwise an ambitious, and probably unattainable aspiration.
The Chicago theory on which the case for flexible labor markets is based predicts that the lower is the return associated with the “outside options” of employment or reliance on social insurance, the higher will be the incentive to engage in crime as a way of making a living. ... That is, other things equal, low wages and weak or non-existent unemployment benefit systems can be expected to lead to higher crime rates, higher rates of imprisonment or both. ...
So, I would argue, my interpretation of my bet with Bryan Caplan is the more relevant one in terms of policy evaluation. The proportion of bad labor market outcomes is better measured by the sum of unemployment** and incarceration (expressed as a proportion of the labour force) than by unemployment alone.
* Or maybe shmibertarian: as we saw during the Bush era lots of alleged libertarians are quite comfortable with extreme use of state power as long is doesn’t touch their bank balances. On the other side of the coin, I should note that the Cato Institute has done some very good work on this subject, including publishing this Glenn Loury piece. ...
Not all prisoners are the result of bad labor market outcomes, and an index of labor market outcomes may weight these two pieces of information differently, so I'm not sure a simple sum of unemployment and incarceration is the optimal measure of labor market performance. But it does seem adequate as a first approximation, or at least better than the unemployment rate alone, and the underlying point is valid.