Purpose

To determine whether only randomized eligible participants are selected for the Pay for Success employment training and job placement program, and whether the employment and recidivism rates are accurately monitored for both the treatment and control groups. We audited selected aspects of Phase 1 of the program for the period October 1, 2013 through September 30, 2016.

Background

To increase employment opportunities and decrease the risk of recidivism, the Department of Corrections and Community Supervision (DOCCS) and the New York State Department of Labor (DOL), in collaboration with Harvard University, developed the Pay for Success (PFS) employment training and job placement program (Program) to target high-risk parolees. Program contracts are based on the social impact bond concept, where private investors provide initial funding for the social services designed to achieve a desired outcome. The investors are reimbursed, and thereby receive a return on their investment, only if performance outcomes are achieved.

For the purposes of this test Program, participation is limited to parolees at high risk of recidivism (high-risk parolees) released to either the New York City or Rochester areas. Highrisk parolees are identified based on a clinical assessment instrument which helps determine an offender’s particular risk factors and behavioral needs, and thereby helps staff design an effective supervision program for that offender. Through the assessment instrument, staff determine the offender’s “Supervision Status Level,” which affects the intensity of the offender’s supervision and management – factors that, if inadequately addressed, could contribute to further anti-social behavior.

According to the design of the Program, its social impact is to be evaluated using a randomized, controlled trial design, whereby eligible inmates are randomly selected to be referred by parole officers to the employment services provided by the Center for Employment Opportunities (CEO) upon release (treatment group). Their outcomes are then compared with an equal number of inmates randomly selected and not referred by parole staff to the employment services provided by CEO upon release (control group). Although not specifically referred for employment services, individuals in the control group can request and receive such services. Success is measured based upon achievement differences in three factors: employment, transitional jobs, and recidivism, with payment contingent on the fulfillment of certain thresholds as specified in the contract.

In Phase 1 of the Program, 2,357 individuals were randomly assigned to the Program between December 9, 2013 and September 29, 2015: 1,502 to the treatment group and the remaining 855 to the control group. Of the 2,357 inmates in the Program, 1,875 were scheduled for release in New York City and the remaining 482 in Rochester.

Key Findings

We found that, for both the treatment and control groups, only eligible individuals were selected for Program participation and the selection was properly randomized. We also found that the Program has procedures in place to accurately monitor employment of the group members. Although several conditions complicate the measurement of employment, we found that they did not have a material effect on the results and should not impact measurement of achievement differences between the treatment and control groups. We also determined that DOCCS is properly tracking individual days of re-incarceration to accurately account for recidivism.

Other Related Audit/Report of Interest

Department of Corrections and Community Supervision: Oversight of Sex Offenders Subject to Strict and Intensive Supervision and Treatment (2014-S-50)