Important Study from the Open Philanthropy Project on the Effect of Incarceration on Crime
One of the most intractable questions in criminal justice is whether incapacitation of offenders reduces crime. The assumption that it does undergirds our explosive rise in incarceration rates over recent decades, but is it accurate? The answer is not purely an academic one. It is a critical prerequisite to current criminal justice reform efforts, and in particular, those focused on decarceration and alternatives to incarceration. It is also important to sentencing advocates, who will frequently cite to studies that conclude that long sentences do not deter and that prison actually exacerbates recidivism. Now, there is a an important meta-study from the Open Philanthropy Project, published in September 2017, The Impacts of Incarceration on Crime (see also here and here), in which David Roodman subjects three dozen empirical studies on the link between incarceration and crime to a rigorous review, including replication and reanalysis of eight of the studies. He concludes, in essence, that while incarceration may modestly lower crime rates temporarily, this decrease is cancelled out by the criminogenic impact of incarceration, which increases crime rates. As such, the “best impact of additional incarceration on crime in the United States today is zero.” His key findings, subject to several caveats, can be summarized as follows:
- Deterrence: Longer sentences do not clearly deter crime. Swift and certain punishment can deter, when practical, but perhaps works mainly when complemented with positive incentives and appropriate treatment.
- Incapacitation: Incapacitation is real — time inside prison does indeed reduce crime outside prison.
* but *
- Rehabilitation: Most relevant, credible studies suggest that incarceration aftereffects are harmful, and strong enough to offset the crime benefits of incapacitation.
Finally, Roodman focused on “high credibility” studies (those that could exploit randomized experiments), and in particular, eight studies where he could access the underlying data and code (i.e. the studies with the highest integrity). “Replication and subsequent reanalysis of these eight revealed significant econometric concerns in seven and led to major reinterpretations in four” leading him to conclude that “even the best studies on incarceration and crime are less reliable than they appear.” Moreover, Roodman’s review “focuses mainly on studies that compare incarceration to ordinary freedom or traditional supervised released (probation and parole), as distinct from alternatives such as in-patient drug treatment and restorative justice conferences. Those options may offer promise, and deserve more research and evidence reviews.”