1. Enhanced Services for the Hard-to-Employ Project Transitional Jobs for Ex-Prisoners: Early Impacts from a Random Assignment Evaluation of the Center for Employment Opportunities (CEO) Prisoner Reentry Program John Jay College of Criminal Justice Prisoner Reentry Institute March 2008 Presentation by: Dan Bloom

2. draft - do not quote or cite2 Reentry and Employment Wide range of barriers to successful reentry, but most experts believe employment is crucial. Two types of labor market challenges: Characteristics that make ex-prisoners hard-to-employ (limited work and education, health problems, etc.) Additional impact of incarceration (legal barriers, employer discrimination). Few proven employment models for this group.

3. draft - do not quote or cite3 Enhanced Services for the Hard to Employ Project Funded by US DHHS (with US DOL) to test promising employment strategies for groups facing serious barriers to steady work. Led by MDRC, with Urban Institute and other partners. Four sites, each targeting a different type of hard-to-employ population; like four separate studies.

4. draft - do not quote or cite4 The Center for Employment Opportunities (CEO) One of the largest and best known employment programs for ex-prisoners. Started by the Vera Institute of Justice; independent since 1996. Serves approximately 2,000 parolees annually. Uses a transitional jobs (TJ) model; TJ widely seen as promising for ex-prisoners: Source of legitimate income in critical post-release period. Opportunity for staff to identify, correct workplace issues.

5. draft - do not quote or cite5 The CEO Model: Immediate Work Participants placed in CEO work crews (TJs) within one week after enrollment; supervised by CEO staff. Work 4 days/week; paid daily. Day 5: Meetings with job coach, job developer, fatherhood program, etc. When deemed “job ready,” receive help finding permanent job, then retention services after placement. Participation in CEO is voluntary for most, but meets a parole condition to seek or maintain employment.

6. draft - do not quote or cite6 The CEO Evaluation During 2004-2005, ex-prisoners who showed up at CEO were assigned, at random, to: Program group: Regular CEO program, or Control group: Limited job search assistance. Random assignment only in weeks when there were more participants than slots. Sample size: 977 (P: 568; C: 409)

7. draft - do not quote or cite7 The CEO Evaluation: Target Group Evaluation targets a subset of CEO’s population: parolees who showed up at CEO after referral by regular parole officer. Other key CEO populations in special programs were not included in the study for contractual reasons.

8. draft - do not quote or cite8 Data Now Available Baseline data (self-reported). Data on quarterly UI-covered employment (not earnings) in New York State. Criminal justice data: arrests, convictions, parole status, and incarceration in NYS prisons. Data from CEO Management Information System.

9. draft - do not quote or cite9 Characteristics of Participants Average age: 33 Over half are fathers; few live with children. 2/3 African-American; 1/3 Latino. 43% no high school diploma or GED. 81% ever worked. Average of 7 prior convictions; 5 years in prison. Time since release: 40% came to CEO within 3 months after release from state prison; model was designed for them. 60% came to CEO longer after release (not expected).

10. draft - do not quote or cite10 Participation in CEO Activities Good test: 79% of program group completed initial 4-day life skills class. 72% of program group worked in a CEO work crew. Average of 8 weeks in work crew.

11. draft - do not quote or cite11 Impacts on Employment and Recidivism: Background In addition to full sample, two groups are studied: Recently released ex-prisoners: enrolled within 3 months after release (n=377; P=222, C=155) Not recently released ex-prisoners: enrolled more than 3 months after release (n=551; P=314, C=237) Currently have 1 year of follow-up (3 planned) Reincarceration data do not yet include local jails. Employment data include CEO jobs and non-CEO jobs but can’t yet distinguish (will be able to later). Statistical significance levels: *=10% **=5% ***=1%

12. draft - do not quote or cite12 Impacts for Full Sample Large but short-lived increase in employment. Small but statistically significant declines in two key recidivism measures (felony convictions; incarceration for new crime); no impacts on other measures. But overall results hide a potentially important pattern.

13. draft - do not quote or cite13 Employment Impacts for Recently Released: initially large but decline Employed (%) Program group Control group Difference Year 185.453.332.1*** Quarter 169.926.043.9*** Quarter 265.530.934.6*** Quarter 347.835.012.8** Quarter 438.933.75.2 Statistical significance levels: *=10% **=5% ***=1%

14. draft - do not quote or cite14 Recidivism Impacts for Recently Released: large in many areas Year 1 (%)Program group Control group Difference Arrested21.827.0-5.2 Felony conviction1.76.2-4.5** Parole revoked or absconded18.827.0-8.2* Incarcerated in state prison for any reason9.619.7-10.1*** Incarcerated in state prison for new crime0.55.1-4.7*** Statistical significance levels: *=10% **=5% ***=1%

15. draft - do not quote or cite15 Employment Impacts for Not Recently Released: faster decline Employed (%) Program group Control group Difference Year 177.756.221.5*** Quarter 165.424.740.7*** Quarter 256.935.521.4*** Quarter 339.235.14.1 Quarter 431.333.8-2.5 Statistical significance levels: *=10% **=5% ***=1%

16. draft - do not quote or cite16 Recidivism Impacts for Not Recently Released: none Year 1 (%)Program group Control group Difference Arrested24.723.41.3 Felony conviction1.92.4-0.5 Parole revoked or absconded23.018.54.6 Incarcerated in state prison for any reason 12.711.71.0 Incarcerated in state prison for new crime 1.02.1-1.1 Statistical significance levels: *=10% **=5% ***=1%

17. draft - do not quote or cite17 Summary of Subgroup Findings Recently released group: Large but short-lived increase in employment. Substantial decreases in felony convictions, parole revocation, incarceration for new crimes, overall incarceration. Rare to see recidivism impacts in random assignment study. Not recently released group: Employment impacts similar, but decay faster. No impacts on recidivism measures.

18. draft - do not quote or cite18 Coming Attractions Up to 3 years of follow-up. Additional data: Survey of 500-550 sample members. NYC jail data. Individual-level earnings data and employer ID #s. Child support payment data. Implementation data. Report planned for 2008. The Joyce Foundation has developed another 4-site random assignment evaluation of TJ programs in Midwest.