Throw Away the Key or Throw Away the Jail? The Effect of Punishment on Recidivism and Social Cost

Miguel F. P. de Figueiredo.

We jail too many people and it costs too much. Incarceration is not only expensive, it also is prone to “hardening” and negative peer learning effects that may increase recidivism. With local, state, and federal budgets at a breaking point, politicians and regulators are increasingly considering alternative approaches to preventing crime. Yet, they face a problem. Studies show that incapacitation is a successful way of reducing crime, yet most scholars and policymakers think that the only way to incapacitate is to incarcerate. This study demonstrates that this assumption is problematic, arguing that we should understand incapacitation along a continuum, with incarceration at one end. This understanding is important because it allows policy makers to think about new ways to avoid the significant social and fiscal costs of jail while at the same time reaping some of the benefits of incapacitation.

This article explores the relationship between incapacitation and incarceration in the context of drunk driving. Policy makers have adopted a variety of incarceration alternatives to curb drunk driving, and this creates a kind of natural experiment that allows for the rigorous testing of the effects of sanctions on future behavior and that derives policy implications for regulating crimes of addiction. This article is the first to examine the effectiveness of the sanctions in curbing recidivism and vehicle crashes with some 200,000 alcohol tests.

Four key results emerge from the study. First, it demonstrates that non- carceral sanctions can be effective. Second, the primary channel through which drunk driving sanctions are effective in reducing recidivism and crashes is incapacitation, rather than specific deterrence. Third, non- carceral sanctions have varied success based on what form they take and who they target. A law passed mandating victim panels, increasing the length of license suspensions, and stimulating the use of ignition interlock devices (IIDs)—which require the driver of a vehicle to take an alcohol test—reduced crashes during and after suspension of a driver’s license. The same law decreased recidivism during the suspension period, but these recidivism- reducing effects ended soon after the license suspension did. In addition, a license suspension enhancement targeting those with higher blood alcohol content levels neither reduced recidivism nor crashes. Fourth, the probability of recidivism and subsequent crashes for first-time offenders given at least 6 to 24 hours of jail, fines, and a license suspension had no statistically significant effect relative to those who had no sanctions. This suggests that drunk driving sanctions at the legal limit are ineffective. This article explains these results, discusses theoretical and legal reform implications, and also outlines a trajectory for improving causal inference in the study of criminal law. This article concludes by discussing the promise and limitations of generalizing from the results to other domains of crime and law.

Full Article