Our country has funded a lot of community-based employment programs in an effort to help people out of prison find work and reduce high recidivism rates. Most are never rigorously evaluated. When they are, the results are disappointing. 

In 2005, a team of researchers at the Urban Institute conducted a meta-analysis to understand the impact that community employment programs had on recidivism. They reviewed eight of the highest quality experiments they could find conducted between 1971 and 1994. They concluded that, overall, the eight programs had no significant effect on the likelihood that participants would be rearrested.¹

The transitional job (TJ) program model was embraced as a more promising solution. TJ programs provide temporary, subsidized jobs, support services, case management, and job placement help to program participants. 

Over the last 20 years there have been eight rigorous evaluations of TJ programs targeting people out of prison.

 The Center for Employment Opportunities (CEO) evaluation which began in 2004 used a random assignment design that compared a TJ program with job search assistance. There were no significant increases in unsubsidized employment, but there were reductions in two measures of recidivism that persisted over three years. CEO reduced the probability of misdemeanor re-conviction by 12% and jail re-incarceration by 11%. The reductions were driven by a subset of participants who were at higher risk to re-offend and entered the program within 90 days of their release from prison.²

 The Transitional Jobs Reentry Demonstration (TJRD)  tested four employment programs for former prisoners in Chicago, Detroit, Milwaukee, and St. Paul. In 2007-2008, more than 1,800 men recently released from prison were randomly assigned to a TJ program or to a program providing basic job search assistance but no subsidized jobs. The study found that the TJ programs did not significantly affect measures of recidivism over the two-year follow-up period. About half of each group was arrested, and a similar number returned to prison.³

In 2010 we gave it one more shot.  The Enhanced Transitional Jobs Demonstration (ETJD)  tested three more transitional jobs programs targeting people recently released from prison in Fort Worth, Indianapolis, and New York City. The programs were “enhanced” in various ways compared to the TJRD sites. On average, the three ETJD programs did not have a statistically significant impact on whether participants were arrested, convicted, or returned to incarceration in either jail or prison. When evaluated separately, the Indianapolis program was successful in reducing recidivism by about 8%. There was no impact in Fort Worth, and the New York City program led to a small increase in recidivism.⁴

Like I said, the results are pretty disappointing. It’s not surprising that TJ programs are being written off as a bad bet.⁵⁶ But that’s a mistake.

It’s not that TJ programs can’t reduce recidivism. They haven’t been designed to do it.

We can do way better than reducing jail misdemeanors by 12%. Way better. TJ programs designed correctly and implemented well can be a game for reentry.

Six design problems and how we can fix them

Reducing recidivism is notoriously difficult. A program must be both designed and implemented well to have any chance of success. But a good design comes first. And the TJ program model has a few critical design mistakes that have limited its ability to produce results. 

Don’t get me wrong. Employment isn’t going to solve all of our reentry problems. But it can be used to reduce recidivism for some people, some of the time. And that’s more than worth our time and investment to get it right. 

The TJ model starts with a faulty theory of change.

Programs are designed (and funded) on the assumption that giving a person out of prison a job and successfully transitioning them into the labor force will make it less likely they will commit crimes or violate parole and return to prison. But research just doesn’t support this assumption.

In his analysis on labor markets and crime, criminologist Shawn Bushway explains that “Ex-felons are at a higher risk for crime, and while change can and does happen, the reality is that the social science evidence in support of work…as a change agent for ex-offenders is relatively weak.”⁷

Being unemployed and having an unstable work history are predictors of recidivism. But giving someone a job doesn’t address the underlying reasons why a person hasn’t had success at work. A 2010 study demonstrates that young people exhibit unstable work histories well before the first time they’re incarcerated.⁸

Shawn Bushway and Robert Apel considered “Why work doesn’t work” to reduce recidivism. They concluded that “the brutal truth is that individuals have difficulty holding onto jobs; providing them with a job probably does little to improve their employability.” Outcomes of the National Supported Work Demonstration (one of the studies in Visher’s early meta-analysis) support their point. Among the participants with a criminal history, 33% were fired from their program job, and another 20% were terminated for other negative reasons such as re-incarceration.⁹

Research is unequivocal here. Re-offending behavior is correlated to criminal thinking patterns, attitudes supportive of crime, and personality traits such as impulsivity and aggression. People cycling in and out of prison often lack the problem-solving and self-control skills needed to maintain employment. They have unrealistic expectations about what life after prison will be like. They have anti-social peers, poor family relationships, negative attitudes about work and authority, and substance abuse problems.¹⁰

Center for Employment Opportunities did not improve employment outcomes yet it reduced recidivism. The TJRD experiment resembled CEO in many ways. But those TJ programs didn’t improve employment outcomes or reduce recidivism. Why? Researchers hypothesize that it wasn’t the transitional employment itself that lead to CEO’s positive recidivism outcomes. Instead, the crime reduction came from the extensive coaching or other services that are part of the CEO program.¹¹

Securing employment makes one aspect of a person’s life more pro-social. But it’s not nearly enough.

Work isn’t the change agent. It’s the motivator.

People being released from prison list finding a job as a top concern. Many believe that if they can find a good job, everything else will fall into place. But most people in prison have worked. Many were employed at the time of their arrest.¹²

Do you know what doesn’t usually make it to the top of the priority list? Managing aggressive behavior. Changing attitude about rules and authority. Adjusting unrealistic expectations. Learning impulse control and problem-solving skills. 

It’s not likely that people released from prison will voluntarily enroll in a program focused solely on developing thinking and coping skills. But without them, people struggle to stay employed and out of prison.

Herein lies the power of employment as a key reentry strategy. Simply giving a person a job is an  ineffective strategy to change criminal behavior. It is effective, however, as a way to motivate people to participate in treatment shown to reduce re-offending.

The TJ program evaluations demonstrated that men returning from prison have a high level of motivation to work.

In the TJRD experiments, about 85% of the men who were assigned to the TJ programs worked in a transitional job. The study did not find that employment reduced recidivism. It did find “concrete evidence that men coming out of prison are strongly motivated to work,” and that “transitional jobs programs are a good engagement strategy.”¹³

Giving a person a job doesn’t reduce recidivism. However, subsidized work can be an incredibly effective strategy to engage an at-risk group in treatment. 

TJ programs aren’t targeting the specific subset of people who will benefit the most.

There’s consensus among experts that no one criminal justice intervention is appropriate for all people and all situations. Types of rehabilitation programs are more effective with some people than others. Experts analyzing the relationship between work and crime have been making this point for years. 

Shawn Bushway argues that “work programs can, and perhaps should, be developed to serve the specialized and (small) group of at-risk individuals who are likely to be most responsive to work incentives.”¹⁴

In their report, “What have we learned from longitudinal studies of work and crime?” Christopher Uggen and Sara Wakefield conclude that “providing employment to offenders and at-risk groups works only for some kinds of offenders in some situations.”¹⁵

In her analysis of crime, work, and reentry, scholar Ann Piehl finds “It simply makes sense to target work initiatives to the part of the population where they are likely to have the greatest return.”¹⁶

All eight TJ programs failed to target the specific subset of people returning from prison who were most likely to benefit.

Target the group of people who will benefit the most.

Subsidized work can promote behavior change and reduce crime. But the same program design will not be equally effective for all people. And there are also some groups of people who will not benefit from employment regardless of the quality of the program design or services. The trick is to target a group of people who will likely benefit, and then design the program to the needs of that group.

The TJRD and ETJD experiments were designed as intense post-release programs (30+ hours a week), but relatively short in duration (3-4 months). The group of people most likely to benefit from this program design is adult men 25 years and older, recently released from prison, scoring medium to high on a risk assessment, without major substance abuse problems who are ready and willing to exit a life of crime.

Targeting this sub-group of the prison population may be criticized as “creaming” to get the “best” candidates into a program.

But this strategy is consistent with the best practices literature. Research shows us that programs are more effective when they’re matched to people who can most benefit from them.¹⁷ Bushway points out that “creaming might not be popular among those looking for programs to serve large numbers of at-risk individuals, but it represents a realistic response to the research literature on labor market programs aimed at crime prevention.”¹⁸

Changing criminal behavior is never easy.

• 83% of people released from prison are re-arrested at least once during the nine years following their release.¹⁹
• 68% of our state prison population is between 25-50 years old, many of whom are at a higher risk to re-offend.²⁰ 
• 41% of the prison population has no alcohol or drug problem.²¹ 

A TJ program targeting an older, more motivated group without major substance abuse problems can still have a massive impact on reducing recidivism.

Target Population:

None of the experiments emphasized CBT.

In the final evaluation of the TJRD experiment, the authors argue the need to develop and test enhancements to the TJ model. They proposed “extending the period of the transitional job, including vocational training as a core program component, or focusing more on the transition to regular employment by offering stronger financial incentives for participants.”³²

A better recommendation would’ve been to include cognitive behavioral therapy (CBT) as a core program component. Structured cognitive behavioral interventions target the thoughts, feelings, and behaviors associated with criminal recidivism. A large body of research has demonstrated that CBT is a particularly effective intervention for reducing recidivism among juveniles and adults.

The authors of CEO’s final impact evaluation got it right.

They finish their report with this insight: “Future evaluations of programs serving former prisoners should also consider looking closely at program components that address criminal thinking and behaviors. Some criminologists believe that cognitive behavioral approaches may be key to reducing recidivism. Indeed, the findings from the CEO evaluation suggest that the program’s promising impacts on recidivism may have been driven by the positive peer and staff influences that extended beyond the basic provision of employment, though paid employment may be needed to engage participants in activities designed to influence behaviors.“³³

Unfortunately, none of the enhanced TJ experiments incorporated CBT as a major program component.

Only one of them provided any CBT services at all. Fort Worth’s Next STEP offered CBT workshops. At the one year follow-up, the Next STEP program significantly reduced arrests by 6 percentage points and incarceration in jail by 5 percentage points. The program reduced recidivism by 19 percentage points among those who were at high risk of recidivism, even though it never produced impacts on employment.³⁴

Participants in the Next STEP program group demonstrated lower levels of “criminal thinking.”  The evaluators suggest that the reductions in criminal thinking may have led to the program’s positive interim impact on recidivism. They state: “The workshops…may have helped participants learn new problem-solving and self-regulation skills, and the job-readiness classes may have also led to improvements in thinking and attitudes. Cognitive behavioral approaches and motivational techniques are becoming increasingly common in programs that work with individuals involved in the criminal justice system, and these approaches and techniques are considered to be effective in reducing recidivism.”

The Next STEP experiment was not able to hold on to its positive results at the 30-month follow-up. But the program put only a very minor emphasis on CBT. It also lacked a strong subsidized work program component. Only 39% of program participants worked in a subsidized job.

What impact could the program have made if CBT was designed as a core program component and more people were placed into transitional jobs?

Use subsidized work to support high-quality, high-dosage CBT.

Giving a person a job is the easy part. Helping them change attitudes and behavior is a whole lot harder and takes a lot more time. Research shows us that it can take 100+ hours of cognitive behavioral interventions to change the attitudes, beliefs, and behavior of people at a higher risk to re-offend.³⁵  

How can we effectively engage adults out of prison in high doses of cognitive behavioral treatment? A well-designed TJ program model isn’t the only solution, but it’s a big one.

CBT must be designed as the core TJ program component.

Subsidized work enables the delivery of high-quality and high-dosage CBT. The two major program components, CBT and subsidized work, must be as integrated as possible. Participants attend group therapy classes and go to work each day. Classes are delivered by program staff in the same building where the participants work. CBT principles are reinforced consistently throughout the day by all staff.

The TJ programs were missing a tightly structured behavior management system.

People cycling in and out of prison need help to establish new, more constructive, forms of behavior. They have difficulty with long-term planning and impulse control. The promise of a “better life” or “good” job in the future is too remote to influence daily decisions. Changing behavior requires incentive structures that offer almost immediate payoffs rather than learning skills with which to build a career.³⁶

The TJ programs lacked a tightly structured behavior management system of immediate rewards and consequences shown to promote behavior change. They were designed without the framework to encourage compliance with program expectations and full participation in the treatment process.

Programs without a behavior management system suffer from a variety of problems. 

Rules enforced inconsistently can maintain and even increase behavioral problems. Anti-social behavior is easily ignored and thereby reinforced. Participants commonly graduate through the program based on a subjective measure of attitude rather than on actual conduct. People who have been in the program the longest are often rewarded, even if their behavior is only good enough to keep them from being terminated. 

Use a token economy.

Behavior management systems are built on principles of operant conditioning, which suggests that all behaviors are learned based on consequences that result from actions. Programs that attempt to change behavior through a system of reinforcement and punishment are often referred to as contingency management programs.

One popular form of contingency management is the token economy.

It involves the use of points that can be earned for specific behaviors and exchanged for backup reinforcements such as tangible goods (money, material goods), desirable activities (sports, recreation, TV, socialization), and social reinforcers (praise, approval).

Contingency management has a solid foundation in empirical literature as an effective strategy to reduce substance use. Token economies have also been found to reduce a wide variety of undesirable behaviors across a range of criminal justice settings.³⁷ Behavioral reinforcement of treatment attendance has been shown to increase treatment retention, reduce unexplained absences, and improved employment and social adjustment while decreasing criminal behavior among violent offenders.³⁸

Yet, contingency management techniques are rarely applied in community reentry programs to increase pro-social behavior and reduce recidivism.  

A program providing intense services to a high-risk population needs a well-designed behavior management system that incorporates forms of contingency management. Incentives are provided to increase the frequency of pro-social behaviors and disincentives are used to reduce antisocial behavior. The system directly targets specific, observable, and concrete behaviors. It places responsibility onto the person to manage their behavior and holds people accountable to program rules and expectations fairly and objectively. 

A well-designed behavior management system has both “micro” and “macro” components. The micro system encourages pro-social behavior in the day-to-day. It immediately reinforces things like attendance, participation, being on time, passing drug tests, and being in uniform.

The macro system moves people forward and backward through the program based on their behavior. It includes a system of levels that are integrated with the micro system. A daily point system moves people towards “leveling” up which comes with additional rewards or privileges. The levels act as intermediary rewards to increase participation and retention. Program completion results in a high-quality job placement which is the ultimate incentive and reward. 

The effectiveness of the behavior management system, of course, relies on staff to enforce it well.

And that requires the use of particular behavior management techniques. Core correctional practices (CCP’s) describe how the behavior management system should be enforced. They are the best practices for staff to use in the delivery of the program and interactions with participants. CCP’s include effective reinforcement, effective disapproval, effective use of authority, and prosocial modeling. These techniques are shown to enhance program delivery and outcomes.³⁹

The two programs that reduced recidivism did not outsource subsidized work.

TJ programs take many forms. In some models, participants work directly for the program in a social enterprise. In others, they work for nonprofits or businesses in the community but remain on the payroll of the program sponsor. A third model places participants with businesses and subsidize a portion of their wages for a time. 

In the TJRD sites, The Detroit and St. Paul programs were operated by Goodwill Industries affiliates. Participants worked in jobs in existing Goodwill enterprises such as retail stores or a light manufacturing plant. In Chicago, workers were employed by a staffing agency that placed almost all participants in garbage recycling plants around the city. The New Hope program in Milwaukee used a scattered-site model. Participants were placed in positions with local nonprofit organizations or small businesses but were employed by New Hope, which paid their wages.⁴⁰

None of the TJRD programs supervised the subsidized work experience. None improved long-term employment outcomes or reduced recidivism.

One hypothesis for the poor results was that the TJRD experiments didn’t help participants move from relatively sheltered positions into unsubsidized jobs. Thus, two of the three later “enhanced” experiments made structural changes to the program design to promote a smoother transition from subsidized to real jobs.⁴¹

Two of the “enhanced” experiments also did not supervise the entire subsidized work experience. Those two programs did not reduce recidivism or improve employment outcomes.

New York’s Pathways program used a “staged” model. After a one-week orientation, participants worked on the program’s street-cleaning crews for six weeks, then move into subsidized internships for eight weeks. At the Fort Worth site, participants begin with a two-week “boot camp” and were then placed in jobs with private employers. Neither program had any long term positive impact on employment or recidivism. 

The only two programs that reduced recidivism operated and supervised the entire subsidized work component.

The third ETJD site, Indianapolis’s RecycleForce program, and the Center for Employment Opportunities were the only two programs that positively impacted recidivism. They also were the only two experiments that directly operated and supervised the entire subsidized work component of the program.

New York and Fort Worth made a mistake by designing their programs to transition participants quickly into “real” jobs. That strategy was the exact opposite of what they should’ve done. Research shows us that the success of CEO and RecycleForce was the result of something that happened in the program other than the provision of a job. Whatever this “something” was, it didn’t happen on jobs outside of the program sponsor.

Outsourcing the subsidized work program component is a problem for three major reasons.

1. Changing criminal behavior requires specific expertise not found in the average nonprofit or business. Anti-social behavior is easily reinforced if it’s overlooked or addressed inappropriately.
2. It’s impossible to apply a behavior management system consistently throughout all aspects of the program if subsidized work is outsourced. Inconsistencies in enforcing rules can maintain or increase behavioral problems.
3. The feedback loop between work supervisors and other program staff must inform individual treatment decisions. Communication gaps naturally exist between different organizations and treatment quality can suffer as a result.

The most effective TJ subsidized work model for reducing recidivism is the social enterprise run by the program sponsor.

The Indianapolis RecycleForce program was the only experiment to produce substantial impacts on both employment and recidivism. The Indianapolis site used an intensive, highly supportive model in which participants worked in a program-run electronics recycling social enterprise. Bingo. For their high-risk target population, RecyleForce got it right.

There are many reasons why a social enterprise run by the program sponsor is the most effective TJ model for reducing recidivism.

First, it maximizes the number of participants placed in subsidized jobs. Out of the three EJDP experiments, the Indianapolis site placed 100% of program participants in subsidized jobs versus 79% in the New York City site and 39% in Fort Worth.⁴²

Having a high participation rate is an important part of a program’s ability to reduce recidivism. But high participation doesn’t guarantee higher success rates. Three of the earlier TJRD programs had an 87% or higher placement rate in transitional jobs, yet they didn’t reduce recidivism. CEO had a transitional job participation rate of 71% but it did reduce recidivism.

Ultimately, reducing recidivism is not about the job. It’s about something that happens within the program. Running a social enterprise amplifies the potential for impact because it ensures a higher participation rate and it allows for a more immersive program experience.

A social enterprise is also the ideal work model to integrate a sufficient dosage of CBT alongside transitional employment.

• Participants can easily move between classes and work as needed throughout the day.
• The behavior management system can be applied consistently across the entire program. Inappropriate behavior is quickly identified and corrected. 
• Working on-site controls for potentially risky unstructured and unsupervised time.
• Work supervisors communicate with other program staff more regularly resulting in higher quality treatment. 
• Subsidized work can easily be extended if a person isn’t ready yet to transition to permanent employment.
• A social enterprise model allows a program to embed an organizational-wide culture shown to promote behavior change.

There’s no question about it. The social enterprise work model has the most potential to both improve employment outcomes and reduce recidivism.

People need the skills to succeed in new pro-social roles and they need a pathway into high-quality employment.

It’s a mistake to assume that all jobs will have equal influence on helping a person transition to a more pro-social lifestyle and maintain it over time. Research indicates that the quality of the job can matter when it comes to helping older adults exit a life of crime.⁴³ 

Criminologists Christopher Uggen and Jeremy Staff studied the relationship between job quality and recidivism in the Supported Work Demonstration Project. Participants who were more satisfied with their job recidivated less often. The researchers conclude that “employment quality may be more important for crime reduction than the simple presence or absence of a job.”⁴⁴ 

Employment of poor quality is positively related to crime.⁴⁵ Yet, simply connecting someone to a high-quality job is unlikely to reduce their chances of returning to prison. People need help developing the skills to succeed in a new pro-social lifestyle. They also need a pathway to higher quality employment.

The TJ programs weren’t designed to do either. They didn’t provide the intense treatment needed to help participants change thinking patterns and attitudes associated with criminal behavior. They were also unable to transition many participants from low-quality subsidized work into more satisfying employment upon the program’s end. A successful TJ program must be designed to do both. 

A high-quality job placement has three important roles in the design of an effective TJ program.

1. A higher-quality job is more likely to have low staff turnover, regular full-time hours and permanent work status, which can limit a person’s exposure to risky people and situations.

   The promise of a “good” job can initially motivate people to enroll and participate in the program. It’s an integral part of the program’s behavior management system. Job placement is the ultimate reward for meeting program expectations. 
2. People who are employed are less likely to be re-arrested. A job placement that offers better pay, benefits, and potential for advancement can increase the chances that a person will remain employed after the program ends (when they have developed pro-social skills to be successful at work), and thus reduce their chances of returning to prison.
3. A higher-quality job is more likely to have low staff turnover, regular full-time hours and permanent work status, which can limit a person’s exposure to risky people and situations.

Brokering higher-quality jobs for people out of prison is challenging, but not impossible. 

Opportunities can be found in small to medium-size manufacturing, logistics, or construction companies. They can also be found in county, city, or state government. Program leadership starts the process by engaging funding partners, community supporters, and political connections to find initial job partners. A well-designed and well-executed program can then build on a small portfolio of companies willing to hire.

Delivering high-quality cognitive behavioral classes and running a successful in-house work model is impressive to potential employers. People with soft-skills training are in high demand. Delivering one great employee sells the program and opens other doors. One success at a company often leads to another hire. 

Let’s finally design a transitional jobs program that reduces recidivism. 

It’s not easy. But it’s possible.