Western notes that “Research also shows that people are deterred by the stigma of a criminal record. When severe punishment becomes commonplace, stigma, and its deterrent effect, are diluted.” However, Western doesn’t seem to understand how these penalties work. The problem is that reputational penalties are much less important for those who have less to lose — those on the lowest rung of the ladder can’t be punished much this way. Unfortunately, that is why locking up people in prison might be a very important way of deterring those who have nothing else to lose.
For the same crime, white collar, higher-income criminals face much, much larger reputational penalties from the legal system than lower-income criminals. Of course, they also face larger fines and restitution, but the reputational penalties constitute the largest part of the total penalty. I provided the details in my first posting. There are two other findings to add: putting people in prison produces a bigger reputational penalty than arrest, and the longer the prison term, the larger the reputational penalty. Evidence on the reputational impact from arrest is hard to find, but the only reputational effects that I know of from arrest occur, for the same reasons, for high income criminals.
This creates an obvious problem for Western’s hope of relying on punishment short of imprisonment to deter poor criminals. While reputational penalties represent most of the penalty for higher-income individuals, for non-drug offenses, prison represents most of the penalty for the lowest-income criminals.
Let me provide some perspective to Western raising the issue of certainty versus severity. Certainty involves the probabilities of arrest and conviction. My first post cited my survey of the evidence that pointed to a similar pattern: “Changes in arrest rates account for up to about 18 percent of the variation in murder rates. Conviction rates explain up to another 12 percent. Prison sentences another 10 percent and the death penalty another 10 percent (see Lott, 2007, Chapter 4).”
Both certainty and severity matter, but relying on changing levels of certainty wouldn’t make much difference if there wasn’t a penalty to back it up, and the reverse is also true. And the cost-benefit discussion that I also provided in an earlier post was based on that same percentage for the impact of prison sentences.
No one has to rely on unjustified retribution to explain a preference for prison over a jobs program. The very thing about prison that upsets Loury and Western — the fact that it is punitive — is the precise reason why we want to have prison sentences: to deter crime, we need punishment. I think that it is that very intuitive point that would be behind people’s rejection of proposals that replaced prison with a jobs program. As I understand Western’s Brookings Institution article, the jobs program works in the shadow of the threat of prison. So it is impossible to evaluate the jobs program as a substitute for prison.
There is also one very serious cost of the jobs program that Western’s cost-benefit analysis for Brookings ignores. If the jobs program takes some of the long-term penalty sting out of committing a crime, a permanent program where people thinking of committing the first time knew about it, the jobs program could increase crime. Because of this it is likely that Western’s proposal has overestimated the benefits he claims exist.
Wilson’s arguments aren’t based on “disproportionate and prejudiced” anger, but on logical arguments about deterrence. Interestingly, Western keeps pointing to evidence such as the so-called unexplained arrest-prison gap, or now reputational penalties, that shows that high income and white criminals are the ones that get punished more severely by the legal system.
 You just can’t compare a one percent change in arrests or conviction rates and a one-month increase in prison, because the two are in different units. How much these variables explain of the change in crime rates depends on how much they change and the impact that they have. Prison sentence lengths tend only to change relatively gradually, especially when compared to arrest rates. I could also get into confusions over marginal versus average deterrence, but those discussions are beyond the scope of the current debate.