The United States has the highest incarceration rate and the most number of people in prison in the world, having more than quadrupled the incarceration rate over the past three decades. Many people believe the increase in America's prison population has been driven mostly by the War on Drugs, but in fact incarceration rates for violent, property and other crimes have increased dramatically as well.

Has this massive increase in imprisonment rates done any good? Proponents hope that prison will reduce crime by physically incapacitating inmates and preventing them from offending against the public, and by deterring offenders with the threat of punishment. A large number of criminologists and legal scholars are skeptical, however, noting that the U.S. experienced surges in crime rates during the late 1970s and 1980s even as incarceration rates increased sharply. But this comparison of aggregate crime rates and incarceration rates for the U.S. as a whole over time is far from definitive evidence about the effects of prison, since countless other factors have been changing over time as well - including the growth in crack cocaine use in the late 1980s.  Some previous research has tried to control for trends in confounding influences such as the crack epidemic by comparing trends in crime across states that experience different trends in imprisonment rates. But these studies may still fail to isolate the causal effects of incarceration, since whatever factors are causing some states to increase their imprisonment rates more rapidly might also independently influence the prevalence or intensity of criminal behavior - that is, crime rates might cause, as well as be caused by, imprisonment rates.

The best study on this question to date is by Crime Lab member Steve Levitt, and suggests increased imprisonment does seem to reduce crime. Levitt tries to overcome confounding from the possibility that higher crime might cause higher incarceration, and that both crime and prison might be determined by the same underlying, hard-to-measure social factors, by exploiting a "natural experiment" generated by the release of prisoners in some states and not others due to prison overcrowding litigation. It is plausible that overcrowding lawsuits affect crime in states only through their impact on the size of the state's prison population. Levitt finds that crime subsequently increases in states where overcrowding lawsuits are decided in favor of the plaintiffs relative to states where such lawsuits do not so decided.

But to say that increased imprisonment seems to reduce crime is definitely not to say that the current scale of imprisonment in the U.S. is optimal. It may well be the case that America has gone too far in the direction of mass incarceration. We expect the benefits to society from locking up each additional criminal to decline as the number of people imprisoned increases, assuming that the criminal justice system is at all good at its job and prioritizes the most criminally active or dangerous people for imprisonment first. On the other side, the costs of incarcerating over two million people are considerable. The government costs of physically housing someone in prison are the easiest to measure, and are thought to be on the order of $30,000 per year. But other social costs may also be important, particularly in minority communities. Harvard sociologist Bruce Western notes that for black male high school dropouts, the risk of either being dead or having spent time in prison by age 34 is over 60 percent . Unfortunately the magnitudes of these other social costs are currently not well understood. Prisoners who spend more time in prison do not seem to suffer much harm in the labor market compared to those who are incarcerated for somewhat shorter periods of time, but this does not mean that going to prison at all does not harm subsequent work outcomes because of the stigma of a prison record or some other reason. Crime Lab member Kerwin Charles has shown that mass incarceration could also harm children's outcomes by depressing marriage rates.

Given the large number of people coming out of prison, there has been growing interest in learning more about the ability of public policies to facilitate prisoner re-entry. Joan Petersilia studied several experiments around the country and found that intensive supervision probation/parole (ISP) for drug offenders does not seem to reduce offending rates or improve labor market outcomes. However a more recent "coerced abstinence" experiment ongoing in Hawaii, evaluated by Crime Lab member Angela Hawken, finds that increasing the frequency of drug testing (as with ISP programs), but making these drug tests random (rather than scheduled) and imposing immediate, certain, but mild sanctions (such as a day or two in jail) may significantly cut recidivism rates.

A related policy approach is to try to improve the labor market prospects of re-entering prisoners. An experiment carried out by the Manpower Demonstration Research Corporation (MDRC) found that providing re-entering prisoners with a transitional job sped up integration into the labor market slightly, and had some beneficial impact on recidivism rates - particularly for re-entering prisoners who participated in the program relatively soon after release. Crime Lab members Philip Cook and Mallory O'Brien are currently working in Milwaukee to evaluate a more intensive re-entry program for gang-affiliated prisoners, which provides social supports and mentoring in addition to employment assistance.

An alternative approach for reducing the costs of incarceration and the difficulties of prisoner re-entry are to come up with alternatives to imprisonment. "Restorative justice" is one increasingly common strategy that involves trying to repair the harm offenders have caused, for example through supported dialogue between offenders and victims. However, evidence from a variety of restorative justice experiments carried out by Lawrence Sherman and colleagues seems to be mixed to date. Intermediate sanctions, such as fines or community service requirements, provide another alternative to incarceration, although the evaluation evidence in this area is relatively limited.