The Vera Institute of Justice served as the independent evaluator for the Adolescent Behavioral Learning  Experience (ABLE) program, an intervention at Rikers Island designed to reduce recidivism for youth passing through the jail by improving decision making skills. The program was financed by the nation’s first Social Impact Bond (SIB)—an innovative pay-for-success contracting mechanism that leverages private funding to finance public services—financial support for the program was provided by Goldman Sachs. After releasing a summary of findings last year which concluded that the program did not reduce recidivism, today Vera released a technical report which describes the evaluation’s design and findings in detail.

To determine whether the program achieved its goal of reducing recidivism by 10 percent or more—the pre-defined threshold of success at which the city would have to pay Goldman Sachs back for its investment—Vera employed a rigorous quasi-experimental design,  a technique commonly used in program evaluations when randomization is not possible. The evaluation compared young people—ages 16 to 18—who entered the jail in 2013 after the program was established with a matched group of young people entering the jail from 2006 to 2010. Vera tracked recidivism for one year by comparing the number of days that both groups spent in the jail—known as recidivism bed days (RBDs)—as a measure of the program’s effectiveness. To control for external factors that may have accounted for differences in recidivism rates between adolescents arrested from 2006 to 2010 and 2013, such as citywide changes in crime rates or changes in policing practice, researchers also tracked RBDs for 19-year-olds who were not eligible to receive ABLE services over the same period and then adjusted the results of the analysis accordingly.

The change in recidivism for the eligible 16- to 18-year-olds, adjusted for external factors, was not statistically significant when compared to the matched historical comparison group. Furthermore, the 19-year-olds and the study group (16- to 18-year-olds) displayed similar trends in rates of recidivism over time, indicating that any shifts were the result of factors other than the ABLE program. The program did not reduce recidivism and therefore did not meet the pre-defined threshold of success of a 10 percent reduction in recidivism bed days, and was discontinued on August 31, 2015.

While the evaluation found that the ABLE program as implemented was not effective at addressing the problem of recidivism, the social impact bond funding mechanism operated as intended. Because the program was quickly taken to scale, it also demonstrated the feasibility of large-scale programming within correctional facilities.

The technical report and more information on the evaluation can be found on Vera’s website at https://www.vera.org/rikers-able-evaluation.