How Communities Can Reduce Recidivism

San Francisco City Hall

Crime imposes direct and significant costs on society. Victims of crime are affected both economically (through lost or damaged property) and socially (through emotional and physical trauma). Indirectly, crime is costly to society because the administrative apparatus that prosecutes, incarcerates, and supervises offenders is very expensive. Because repeat offenders commit a disproportionate share of crime, any program that reduces the propensity of offenders to recidivate is likely to generate significant benefits for society.

Policymakers in the United States are aware of the enormous potential gains to be had from reducing recidivism. They also know that the status quo approach for handling offenders (arrest -> prosecution -> incarceration or supervision) has done a poor job of preventing re-offense. Over the last few decades there has been a growing movement to support and promote alternatives to what might be called the “Law & Order” model of criminal justice, particularly those that are successful at reducing recidivism. For example Kamala Harris, the California Attorney General, recently launched a new program—the Division of Recidivism Reduction and Re-entry—with precisely this goal in mind.

One of these alternatives is the community court model of adjudication. The goals of community courts are identical to those of traditional courts (promoting public safety, reducing crime, and ensuring due process rights). However there are important differences in the means community courts use to accomplish these goals. First, community courts seek to directly address factors (such as unemployment, homelessness, and addiction) that contribute to criminal behavior. Second, community courts are located in specific neighborhoods within cities, and emphasize the importance of “belonging” to that community, for both offenders, victims, and the court itself. At present, there are roughly 40 community courts operating in the United States, as well as a handful in other countries; this number as well as the number of participating offenders has grown in recent years.

Because recidivism reduction is frequently the stated objective of newly founded community courts, a critically important open question for policymakers is how effective such courts actually are at accomplishing this goal. Answering this question was the goal of a report I recently authored with colleague Beau Kilmer of the RAND Corporation. Beau and I were specifically interested in the San Francisco Community Justice Center (CJC) – the community court in San Francisco’s Tenderloin neighborhood—and its effect on the probability that court participants would recidivate.

Using an administrative dataset consisting of all arrests occurring in San Francisco between 2008 and September 2012, we implemented a relatively strong research design to answer this question. This design (known as ‘differences-in-differences’ or DID) exploited geographic and temporal variation in eligibility for participation in the CJC to estimate the effect of the CJC on offender recidivism. Essentially, we compared re-arrest rates for four groups of offenders – the 2×2 partition of first, those arrested inside and outside of the neighborhood in which the CJC is situated and second, those arrested before and after the CJC became operational. (In general, offenders can only participate in the CJC if they were arrested inside a defined neighborhood area, and if they were arrested after the CJC opened.)

We found that outside of the CJC area, the rate of offender re-arrest grew between the pre- and post-periods, while inside the CJC area the re-arrest rate fell. This “difference” (between regions) in “differences” (between time periods) forms the basis for our estimate of the effect of the CJC on offender recidivism. Our findings support the hypothesis that the CJC reduces recidivism: eligible arrestees in the CJC catchment area in the post period were about ten percent less likely to be re-arrested within one year. This estimate was highly statistically significant and robust to a number of alternative specifications.

This is an encouraging research finding. Our study provides fairly strong evidence that the community court model can be successful at reducing future criminal activity among court participants. However important research questions remain.

First, our conclusions come from a study of the community court in San Francisco; for this reason they are of only limited value in answering the question of whether community courts in other jurisdictions are effective at lowering recidivism. Future researchers can and should examine the effectiveness of other courts, preferably using designs at least as strong as ours—if possible, using the “gold standard” randomized control trial design.

A second open question is that we still don’t know precisely why the San Francisco CJC appears to reduce recidivism. The CJC itself is really a collection of interventions: To begin with, offenders who come to the CJC are given access to a suite of services designed to address issues such as addiction and homelessness. At least in San Francisco, community court participants are also ordered to report to the court much sooner following initial arrest (about one week) than are offenders processed by the traditional court (a month or more). It isn’t clear at this point what the relative contributions of these different program components are, and a carefully designed study could be very informative in this area.

This blog post was written by Jesse Sussell, Senior Quantitative Associate at Social Policy Research Associates.