Appendix_A_Background on Subsidized Employment_18APR12

Appendix A

Background on Subsidized and Transitional Employment Programs and Evaluations

Appendix A

Background on Subsidized and Transitional Employment Programs and Evaluations

Subsidized Employment Models

Subsidized employment programs that seek to improve participants’ employability start from the assumption that structured work experience is an effective way to build human capital. However, there is great variation from model to model in how work experience is delivered, what kinds of skills are targeted, and other programmatic features. Four common elements of the programs – targeting and screening, work experience, supports, and transition services – are discussed below, followed by a brief discussion of program funding. Exhibit 1.1 shows a few key features of subsidized employment programs that have recently been part of random assignment studies.

Targeting and screening. As noted earlier, during economic downturns, subsidized employment programs may target a relatively broad range of people, including those who do not face major personal or situational barriers and are expected to return to work fairly readily when economic conditions improve. For example, Pennsylvania’s “Way to Work” program – a program that was funded under the TANF EF – was open to unemployed parents with income below 235 percent of the federal poverty level.

However, programs that seek to improve employability usually target the “hard-to-employ” and often operate in systems that interact with disadvantaged people, including both service systems such as TANF and vocational rehabilitation, and enforcement systems such as criminal justice and child support enforcement. (Subsidized employment programs for disconnected youth may operate at the community level, with fewer ties to public systems.)

Within the AFDC/TANF system, subsidized employment has often been targeted (or mandated) for particular subsets of recipients. For example, Philadelphia’s Transitional Work Corporation (TWC) was designed to serve recipients who had received benefits for at least two years and Vermont’s Welfare Restructuring Project required recipients to work in subsidized jobs after 30 months of benefit receipt if they were unable to find unsubsidized jobs. Other subsidized employment models have targeted recipients who were unable to find jobs after an initial period of job search.

Similarly, outside of TANF, subsidized employment initiatives use various strategies to identify who should receive a subsidized job. For example, the New Hope project, tested in the 1990s, reserved subsidized employment slots (called community service jobs) for people who were unable to find jobs after a period of job search. San Francisco’s TANF-EF funded initiative (which still operates today) starts with an initial vocational assessment which results in placement in one of three tiers. Those in the highest tier receive vouchers that can be redeemed by private employers for a 100 percent wage subsidy. Those in second tier are placed in public sector jobs, and those in the lowest tier receive more intensive and supportive transitional jobs with nonprofit organizations.

Subsidized employment programs for former prisoners are not always so carefully targeted. Parole agencies often do not have a systematic approach for assigning parolees to different kinds of employment services, so a broad range of people may be referred to subsidized employment programs. As discussed below, the Transitional Jobs Reentry Demonstration, a four-site test of transitional jobs programs for former prisoners, found that subsidized employment programs can have negative impacts on unsubsidized employment if they serve many people who could have found jobs relatively quickly on their own, highlighting the importance of careful targeting – a concern that was also raised in the 1970s by critics of the PSE program.

Work experience. Work experience is usually the core of subsidized employment programs. In some models, participants are placed in “regular” work environments with private, public, or nonprofit employers. The participant may or may not be on the payroll of the host organization but, in either case, his or her wages are fully or partially subsidized by the employment program. This model is used in “scattered site” transitional jobs programs like Philadelphia’s TWC, in on-the-job training (OJT) models, and in many of the programs that were supported with TANF Emergency funds. In other cases, the work experience occurs in a program setting that may look less like a typical workplace. For example, the participants’ co-workers may all be program participants. Transitional jobs programs operated by Goodwill Industries may fall into this category.

Supervision is critical in subsidized work programs because it is one of the primary vehicles for teaching participants appropriate workplace behavior. If the participant works in a program setting, such as a Goodwill agency, then supervision is typically provided by program staff. If participants work in a regular environment, supervision is usually provided by the host employer, with varying levels of support from the program. There are also hybrid models: For example, in the Center for Employment Opportunities (CEO) transitional jobs program for former prisoners, participants work in regular employment settings (for example, municipal office buildings), but they are organized in crews of program participants supervised by a CEO staff person.

Most subsidized employment programs aim primarily to teach “soft skills” – that is, general employability skills such as showing up on time and working well with others. In these models, the work tasks themselves may require minimal skills. This is true of all of the transitional jobs programs that have been rigorously studied (see below). Other models aim to improve skills in a particular occupational area through hands-on training in addition to building soft skills. For example, in YouthBuild programs, young high school dropouts learn construction skills while building or rehabilitating affordable housing.

Finally, programs use different approaches to a range of other issues that affect a participant’s daily experience, including work hours, how participants are compensated (including via welfare grants in “unpaid work experience” programs), evaluation of participants’ performance on the job, and others.

Supports. Subsidized employment programs for disadvantaged groups usually offer some ancillary supports designed to address barriers to employment. However, the nature and intensity of these services varies widely. In addition to workplace supervision, most programs provide some form of job readiness training, either individually in groups. This training typically covers issues such as how to develop a resume, how to behave in a job interview, and how to fill out job applications. Some programs provide educational activities, referrals for treatment or other medical services, parenting classes, child support advocacy, counseling, or other services (the specific kinds of services depends to some extent on the program’s target population). These supports may be provided before, during, or after the work experience period and may be provided in-house or via referral. Many programs use case managers – typically separate from worksite supervisors – to assess participant’s needs, refer them to services, provide counseling, and reinforce the worksite experience.

The transition. When the work experience occurs in a regular work environment – for example in an OJT program – there may be an expectation that participants will eventually “roll over” and become regular, unsubsidized employees of the host organization. If the work experience occurs in a program setting, it is almost always necessary for participants to make a transition from the program job into a regular job. Programs typically assign job developers to help participants find regular jobs, although, in some programs, this function is assigned to case managers. As in other employment programs, the role of job developers can range from providing general advice and job search tips to conducting proactive outreach to employers to identify job openings that match a participants’ skills and interests.

Some programs continue to work with participants after they are placed in regular jobs to help promote employment retention, though, once again, the nature and intensity of these post placement services varies. As discussed further below, some subsidized employment programs provide financial incentives to encourage participants to hold unsubsidized jobs.

Funding. Subsidized employment programs are relatively expensive, since program costs include the wages paid to participants. The programs are funded in a variety of ways. In the TANF context, programs may, in essence, convert welfare benefits into wages. When programs serve populations such as former prisoners who are unlikely to receive cash assistance, this may not be possible. Thus, programs may receive foundation grants or other types of public funding (for example, through the Second Chance Act or another funding stream dedicated to a particular population that allows funds to be spent on wages). Some programs sell a product or service and earn revenue that fully or partly supports participants’ wages. For example, programs operated by Goodwill Industries earn revenue from retail stores, contract assembly work, or other projects. CEO contracts with state and local government agencies to perform maintenance and repair work, with the revenue coming from the agencies’ maintenance budgets. In situations like this, it is difficult to know whether the term “subsidized” employment is totally appropriate. While the funds supporting CEO’s work may come from the government, the work is part of routine governmental functions; if CEO did not do the work, the agencies might contract with a private firm to do it, and the word “subsidized” would not be used. Regardless of how wages are funded, subsidized employment programs often scramble to fund case management and other ancillary supports.

Exhibit 1.1

Selected Features of Subsidized Employment Programs In Recent Random Assignment Evaluations

	Personal Roads to Individual Development and Employment (PRIDE)	Center for Employment Opportunities (CEO)	Transitional Work Corp (TWC)	Goodwill Easter Seals Minnesota	Safer Foundation	Goodwill Industries of Greater Detroit	New Hope Project (as part of the Transitional Jobs Reentry Demonstration)
Target group	Welfare recipients with work-limiting health conditions	Ex-prisoners	Long-term welfare recipients	Ex-prisoners	Ex-prisoners	Ex-prisoners	Ex-prisoners
Location of work experience	Nonprofit or public agencies	Public agencies	Nonprofit or public agencies	In-house Goodwill enterprises	Private or public agencies	In-house Goodwill enterprise	Nonprofit organizations or small businesses
Worksite supervision	Worksite staff	Program staff	Worksite staff	Program staff	Worksite staff	Program staff	Worksite staff
Skill goals	Soft skills	Soft skills	Soft skills	Soft skills, but also offered paid occupational training concurrent with work experience	Soft skills	Soft skills	Soft skills
Compensation	Unpaid; mandatory welfare-to-work activity	Paid daily	Paid bi-weekly	Paid bi-weekly	Paid weekly	Paid bi-weekly	Paid weekly

Potential Target Groups for Subsidized Employment Programs

Many public policies designed to encourage work and reduce poverty focus on single parent families, usually headed by mothers. Indeed, the poverty rate for children living with a single mother is five times higher than the rate for children in married couple families. Starting in the 1970s and 1980s, these policies have focused on increasing employment among recipients of TANF (formerly AFDC), which primary serves single mothers and their children. While welfare-to-work policies have been quite successful in some respects, a substantial proportion of single mothers still struggle to maintain employment. States and localities have increasingly targeted these “hard to employ” parents for services, and subsidized employment models have been seen as particularly promising.

At the same time, policymakers have increasingly realized that most single parent families will continue to struggle economically unless they receive significant financial contributions from the non-resident parent(s). Researchers have also learned that child support is associated with better child outcomes and that positive interaction and engagement by noncustodial fathers can promote child well-being.¹ The two roles are intertwined; fathers who are able to contribute financially are also more likely to sustain relationships with their children.²

Thus, over the past two decades, policymakers have increasingly focused on the “other half of the equation” – the noncustodial fathers of low-income children in families headed by single mothers. A critical goal is to increase the extent to which these fathers provide both financial and emotional support to their children. The Obama Administration’s multi-faceted fatherhood initiative reflects this emphasis. According to the Administration’s Fatherhood and Healthy Families Taskforce, “Responsible, engaged fathers are critical to the financial, emotional, intellectual and spiritual well-being of children.”³

Efforts to encourage and assist noncustodial fathers confront two key challenges. First, broad economic trends over the past four decades have reduced the availability of well-paying jobs for workers (particularly males) without postsecondary education. Adjusted for inflation, the earnings of young men with a high school diploma or less dropped 23 percent between 1973 and 2006.⁴ The Great Recession produced huge job losses in manufacturing and construction, industries that once offered relatively high paying jobs to many non-college educated men.

Second, it is difficult to reach and engage disadvantaged men because they are seldom involved in the public systems that support low-income families and do not benefit from many of the work supports established to assist custodial parents. At the same time, however, many disadvantaged fathers are connected to public enforcement systems, notably the criminal justice and child support enforcement (CSE) systems, which serve overlapping populations of low-income men.⁵ In recent years, these two systems have experienced parallel changes that are driving both systems to broaden their missions – to move beyond their traditional enforcement roles to try to improve outcomes for their “clients.”

In 2009, the CSE system collected and distributed more than $25 billion to families. Despite this success, there is an increasing realization that ever tighter enforcement within the system has its limits, and may even be counter-productive in some cases. Although the CSE system serves a diverse range of families, most of the $100 billion in unpaid child support is owed by a relatively small number of noncustodial parents who have no or low reported income. In addition, some studies have found that the system’s growing ability to locate noncustodial parents and withhold their wages may discourage less-educated men from working in the formal labor market.⁶

Thus, starting the 1990s, states and localities began to experiment with a variety of service-oriented models for noncustodial parents. One of the earliest research projects was the Parents’ Fair Share (PFS) Demonstration, which targeted noncustodial parents whose children received welfare and who were behind in their child support payments. The PFS programs combined employment services, more flexible child support enforcement practices (for example, temporary downward modification of support orders), peer support groups built around a curriculum called Responsible Fatherhood, and mediation services to address disputes between noncustodial and custodial parents. PFS modestly increased employment and earnings for the least-employable men but not for the men who were more able to find work on their own. MDRC’s researchers concluded that subsidized employment might be needed, in part to retain fathers in programs long enough to deliver more intensive services, since many had no regular income source.

Responsible fatherhood programs have expanded dramatically in the last decade. Many programs continue to combine parenting classes or support groups with employment services and mentoring. Most programs encourage participants to support their children financially, but they have varying degrees of connection to the CSE system. Programs walk a fine line between reinforcing financial obligations and emphasizing that fathers can play an important role in children’s lives even when they cannot afford to pay support. However, programs often find that it is difficult to recruit and engage fathers without formal links to the CSE system.⁷

The criminal justice system is experiencing a similar reorientation. Driven mostly by changes in sentencing policy, the number of prisoners per 100,000 U.S. residents grew from 110 in 1975 to 484 in 2004.⁸ Today, more than two million people, more than 90 percent of them male, are incarcerated in a prison or jail, and corrections costs approach $70 billion per year.⁹

Within the past 10-15 years, several factors have prompted a reassessment of the role of corrections agencies. First, corrections are consuming an increasing share of state budgets, putting pressure on other areas. Second, rates of recidivism remained stubbornly high. The most recent national data show that two-thirds of released prisoners are arrested and half are reincarcerated within three years of release.¹⁰ Third, research has documented some of the deleterious consequences of mass incarceration for families and communities.¹¹

While sentencing reforms to reduce the number of people who are imprisoned can still be controversial, there is broad agreement on the need to improve outcomes for ex-prisoners who have “paid their debt to society” – and reducing recidivism is another way to reduce the prison population. Individuals leaving prison often face daunting obstacles to successful reentry, including problems finding housing and reconnecting with family. Many experts believe that stable employment is critical to a successful transition, but former prisoners tend to have low levels of education and skills and other characteristics that make them hard to employ. In addition, studies have shown that many employers are very reluctant to hire people with criminal records, and state laws sometimes bar convicted felons from working in certain fields. Thus, employment rates for recently released prisoners are very low, often well below 50 percent.¹²

In response, many states and localities have developed multi-faceted prisoner reentry initiatives. These efforts have received Federal support from the Serious and Violent Offender Reentry Initiative, the Prisoner Reentry Initiative and, most recently, the Second Chance Act. Reentry initiatives often include a range of services and supports, but almost all include a strong emphasis on employment.

Interestingly, there is some disagreement in the field about the role of employment services in reentry programming. Some experts argue that steady work is the key to a successful transition, while others believe that employment should not necessarily be the central focus of reentry programs. For example, a recent “coaching packet” funded by the Bureau of Justice Assistance in the U.S. Department of Justice notes that “while substance abuse and employment may be important to address, they do not represent the most influential criminogenic needs. These top criminogenic needs often need to be addressed before other interventions (such as employment) will take hold.”¹³ Indeed, a recent meta analysis of employment programs that served ex-offenders found that, overall, the programs did not have statistically significant impacts on recidivism.¹⁴

Broader meta analyses of evaluations in the criminal justice field have concluded that the most effective programs are cognitive-behavioral and behavioral interventions that directly address antisocial attitudes and build problem solving skills, that incentives can increase individual motivation, and that quality of relationships between staff and ex-offenders may be as important as the content of interventions.¹⁵ Criminologists also stress the importance of assessment and targeting, noting that services can actually increase recidivism when targeted to individuals who are at low risk of recidivism.

Finally, it is critical to note the overlap between programming for fathers and programming for ex-prisoners. Research has shown that family relationships are critical to the reentry process and, indeed, many reentry programs (including CEO) offer extensive fatherhood programming and child support advocacy. In fact, an ongoing research project sponsored by the U.S. Department of Health and Human Services is evaluating programs that seek to sustain and strengthen family relationships among incarcerated fathers.

The preceeding discussion highlights the critical importance of the three groups that will be targeted by STED and ETJD interventions – low-income noncustodial parents, ex-offenders, and hard to employ TANF recipients – and illustrates how the projects lie at the intersection of several important trends in social policy.

What We Know from Rigorous Evaluations of Subsidized Employment Programs

Exhibit 1.2 summarizes the results from a number of random assignment evaluations of programs for various disadvantaged populations that included a significant subsidized employment component.¹⁶ The focus is on long-term (post-program) impacts on employment and earnings, as well as some indirect outcomes. The discussion below briefly summarizes the results for different types of program approaches.

Transitional models. The most commonly studied models have been transitional programs that placed individuals in temporary subsidized paid jobs and then helped them move to regular jobs. The 1970s National Supported Work Demonstration was the first rigorous evaluation of a subsidized employment model. Supported Work was explicitly designed as a transitional program that aimed to improve the long-term employability of hard-to-employ groups. The 15 Supported Work programs offered 12 to 18 months of highly structured paid work experience. Participants worked in crews to promote peer group support, and the model emphasized “graduated stress” – that is, expectations at the worksite were supposed to increase over time until they approximated those in a regular job. Almost all of the programs offered job placement assistance, though the intensity and quality of this service varied. Some of the Supported Work programs were social enterprises that sold products or services to offset the costs of running the program.

A random assignment design was used to measure the impacts of the model for four target groups: female long-term recipients of Aid to Families with Dependent Children (AFDC); and three mostly male groups: young high school dropouts; recently released former prisoners; and former drug addicts.

Supported Work initially generated very large increases in employment and earnings for all target groups, suggesting that the programs successfully enrolled hard-to-employ participants who would have been unlikely to work without the program. (This initial period, when many program group members are working in subsidized jobs, is often called the “in program” period). However, the differences in employment rates faded rapidly for all four target groups as program group members left the subsidized jobs. Two or three years after people entered the study, the program group was no more likely to be working than the control group. Nevertheless, results for the long-term AFDC target group remained positive – the program group worked more hours per month and had higher average earnings than the control group for at least three years, and the program group also received less welfare. Within the long-term AFDC target group, impacts were largest for individuals who had been on welfare for a very long time (more than 7 years) and those who had never held a job. These individuals were least likely to find employment on their own, without supported work, leaving more room for the program to make a difference.

In general, results for the other three target groups were disappointing, though there were decreases in crime and some hints of long-term employment impacts for the former addict group. In addition, it appears that Supported Work reduced recidivism for older ex-prisoners (i.e., those over age 26), perhaps because they had reached a point in their lives when they were determined to avoid further incarceration and the jobs program helped them further this goal.¹⁷

The most recent rigorous evaluations of subsidized employment models, still ongoing, have focused on transitional jobs (TJ) programs. TJ programs vary along several dimensions, but they generally combine temporary paid jobs, ancillary supports, and job placement assistance. TJ models are similar in many ways to the Supported Work programs tested in the 1970s, though the work experience component is generally much shorter (less than six months) and there is, in theory, a greater focus on moving participants into unsubsidized jobs. MDRC is currently leading three evaluations including a total of six TJ programs: the evaluation of the CEO program for ex-prisoners, an evaluation that includes the TWC program for long-term welfare recipients (both part of the HHS/ACF Hard-to-employ project), and the four-site Transitional Jobs Reentry Demonstration targeting ex-prisoners. In most of these tests, a TJ program is being compared to a program providing basic job search assistance but no subsidized jobs.

Although none of these evaluations is complete, the basic story is clear.¹⁸ As in the Supported Work demonstration, all of the TJ programs produced large increases in total employment initially, but these increases were driven either mostly or completely by the transitional jobs themselves. None of the six programs have generated sustained increases in unsubsidized employment; in fact, several of them produced statistically significant (though temporary) decreases in unsubsidized employment. This seemed to occur because the TJ programs served some people who could have found jobs on their own or with help from the job search programs. This type of substitution of subsidized for unsubsidized employment appeared to be most prevalent in sites where the labor market was relatively strong and the TJ program placed all participants into subsidized jobs very quickly.

It is difficult to reliably deconstruct the results of these studies, but it appears that the programs failed to increase long-term employment both because a substantial number of TJ participants were never placed into regular jobs (typically about half of TJ workers successfully transitioned to regular employment) and because many of those who were placed lost their jobs relatively quickly.

As in the Supported Work demonstration, the results from the recent TJ studies are somewhat stronger for the mostly female TANF population than for the mostly male ex-prisoners. TWC generated some increases in unsubsidized employment, though they faded by the end of an 18-month follow-up period. In another echo of supported work, TWC’s impacts on unsubsidized employment were larger (though still relatively short-lived) for the most disadvantaged sample members – those with little recent work experience and long histories of TANF receipt.

Although the employment results are generally disappointing, some of the TJ programs have produced other positive effects. TWC reduced welfare receipt and welfare payments, at least for a time. CEO generated decreases in recidivism that lasted through three years, a rare result in any reentry evaluation. However, the four TJRD programs did not consistently reduce recidivism through the first year of follow-up – even during the early period when there was a large impact on employment. This result challenges the “conventional wisdom” that there is a direct link between employment and recidivism. Even in the CEO study, where there were decreases in recidivism, these impacts did not occur at the same time as the employment impacts, and non-experimental evidence about this link is mixed. It may be that another aspect of CEO’s program – for example, relationships that participants built with staff or attitudinal changes that were trigged by the work crew experience – is responsible for the recidivism impacts.

Finally, there is some suggestive evidence from the TJRD study that financial work incentives may strengthen the impacts of a transitional jobs program. The St. Paul, Minnesota site introduced a program of employment retention bonus payments mid-way through the study enrollment period. Sample members who enrolled after the payments were made available could receive a series of monthly payments totaling up to $1,400 over 6-9 months if they provided proof that they had worked full-time throughout the month. The analysis found that the impacts on earnings from unsubsidized jobs were significantly larger for the cohort of sample members who were eligible for the payments than for those who were not. It is impossible to attribute these results to the bonuses alone, since other aspects of the program and the context also changed over time, but these results, when combined with other studies of earnings supplements, suggest that this may be a promising strategy.

Unpaid work experience. The 1980s saw the rise of workfare, or community work experience programs (CWEP), in the welfare system. Under this model, instead of receiving paying jobs, recipients are required to work in community service positions in exchange for their benefits, with work hours typically calculated by dividing the monthly welfare grant by the hourly minimum wage. Those who fail to comply can be penalized by having their benefits reduced or canceled. Critics argue that wage-paying models like the transitional jobs programs discussed earlier are preferable to CWEP, both because participants can qualify for the Earned Income Tax Credit and because wage-paying jobs are more like regular jobs and thus provide better preparation for the regular labor market. The first point is usually true, but there is no rigorous evidence about whether the presence or absence of a paycheck – as opposed to, for example, the nature of the work experience – is related to long-term employment outcomes.

Although much discussed, workfare has only been implemented on a large scale in a few places. The few studies that were designed to isolate the impact of unpaid work experience from other program components (conducted in the 1980s) found little evidence that it led to increases in employment or earnings, though some advocates argued that the main purpose of workfare was to enforce a reciprocal obligation – that is, to transform AFDC into a work-based income support program – not to prepare recipients for unsubsidized jobs. Moreover, surveys of CWEP workers and supervisors found that the participants performed real work and that their productivity was about the same as regular workers.¹⁹

Interestingly, a more recent program, New York City’s Personal Roads to Individual Development and Employment (PRIDE), evaluated by MDRC as part of the ERA project, found more positive results, with employment and earnings gains lasting at least three years. PRIDE targeted a very hard-to-employ group – welfare recipients with health-related barriers to employment – and placed many participants into unpaid work experience positions, though there is no way to isolate the impact of that component.²⁰ Some more recent work experience programs, such as those in Erie County, New York and Hamilton County, Ohio, are designed with a more explicit focus on building skills in addition to enforcing a reciprocal obligation, though these models have not been evaluated.²¹

Subsidized private sector employment. A variety of different models have used public funds to provide full or partial wage subsidies to private employers who hire and/or train disadvantaged workers. These models can be loosely grouped under the heading “on-the-job-training” or OJT.

In the early 1980s, a follow-up project to Supported Work, the Structured Training and Employment Transitional Services (STETS) demonstration, targeted developmentally disabled young adults. The program model reflected a growing preference for placing individuals with disabilities into competitive employment as quickly as possible. In the first phase of the intervention, emphasizing training, participants were paid by the program; some participants worked in sheltered settings and others worked in regular employment settings. In the second phase, all participants worked in competitive settings, though they sometimes stayed on the program’s payroll, and the employers often received full or partial wage subsidies. A random assignment evaluation found that STETS substantially increased the percentage of young people working in competitive employment in the post-program period.²²

At about the same time, the AFDC Homemaker Home Health Aid demonstration tested a subsidized employment model for nearly 10,000 welfare recipients in seven locations. Participants received four to eight weeks of training, followed by up to one year of subsidized employment as home health aids. The evaluation found that the program increased employment and earnings in the post-program period in most of the sites.²³

Two other 1980s studies, in Maine and New Jersey, tested versions of OJT for welfare recipients.²⁴ These programs reimbursed employers for half of the wages paid to a program participant during a transitional period; employers were expected to hire the participant into a permanent job if the transitional period was successful. Both programs significantly increased earnings, though both were fairly small programs that targeted selected groups of recipients. These results were largely consistent with the National Job Training Partnership Act (JTPA) evaluation, which examined impacts for participants who were recommended for OJT or job search assistance (JSA). There were significant earnings gains for adult women in the OJT/JSA service stream, and the impacts were largest for AFDC recipients.²⁵

Several models designed to accelerate reemployment for unemployment insurance recipients have also included on-the-job training, though usually as a minor program component. Although the random assignment evaluations of these programs were not designed to isolate the impact of OJT, one nonexperimental analysis found that this service led to higher employment and earnings for participants.²⁶

Models for youth. As far back as the 1930s, there has been a distinct strand of subsidized employment programming for disadvantaged youth. These programs have often included a link to education or training, and many have emphasized service; participants work on projects that produce visible benefits to communities. Participants sometimes receive stipends instead of wages.

As noted earlier, high school dropouts were one of the populations targeted in the Supported Work project; results were generally disappointing. Shortly afterwards, MDRC managed a very large youth-focused demonstration project, the Youth Incentive Entitlement Pilot Projects (YIEPP), which provided subsidized part-time and summer jobs to all low-income youth in particular geographic areas who agreed to attend school regularly. The YIEPP provided jobs to 76,000 young people and a comparison site design found that the program virtually erased the large gap in unemployment rates between black and white youth in the target areas, though it did not affect school outcomes. The project was terminated before post-program impacts could be fully measured.²⁷ A sub-study tested different levels of subsidies designed to induce private employers to hire YIEPP youth, finding that employers were quite sensitive to the subsidy level that was offered. Specifically, about 18 percent of employers agreed to employ youth when a 100 percent subsidy was offered, but only 10 percent agreed when the subsidy was 75 percent.²⁸

Youth service corps, descendents of the New Deal’s Civilian Conservation Corps, started receiving federal funding in the 1970s. In the 1990s, a random assignment evaluation of this model found a variety of positive results, particularly for African-American males, though the follow-up period was short.²⁹ Another evaluation of this model is ongoing, sponsored by the Corporation for National and Community Service.

Other service-oriented models for youth include Civic Justice Corps, which target youth involved in the justice system (currently being tested using random assignment in New York City) and YouthBuild, mentioned earlier, which operates at more than 200 sites nationwide. YouthBuild participants split their time between worksites, where they learn construction skills, and classrooms, where they study for a high school diploma or GED. Participants receive wages or stipends for their time in the program. A national random assignment study of YouthBuild is now underway.

Summary and implications. As discussed earlier, subsidized employment programs may have multiple goals, including providing work-based income support to people who are unable to find jobs, and improving the employability of disadvantaged groups. Many subsidized employment programs have achieved the income support goal by operating at scale and providing opportunities for “real” work to people who would not otherwise have been employed (as indicated by large in-program impacts on employment).

In terms of the second goal, a few subsidized employment models have been at least moderately successful at building employability and increasing earnings over time. However, two key caveats are in order. First, most of the subsidized employment programs that generated long-term employment gains targeted women; there have been very few positive post-program results for adult men.

Second, most of the positive results for women were generated by models that had strong links to regular employment. For example, the on-the-job training models in Maine and New Jersey placed participants directly into regular jobs and subsidized their wages, and the Homemaker Home Health Aid model placed participants into subsidized positions as home health aides in regular work environments. Notably, studies suggest that the most effective employment programs for individuals with disabilities also place participants fairly quickly into competitive employment.³⁰

With the exception of the Supported Work results for the AFDC target group, there are not many examples of programs that achieved long-term employment gains by placing participants into subsidized jobs outside the regular labor market and then helping them transition into regular jobs. As discussed further below, a key question is whether OJT-like models that attempt to place participants directly into private firms can be designed to serve harder-to-employ groups; most of the programs tested to date have operated on a small scale and served fairly selective populations.

1 Carlson and Magnuson, Forthcoming.

2 Nepomnyaschy, 2007.

3President’s Advisory Council on Faith-based and Neighborhood Partnerships, Fatherhood and Healthy Families Taskforce, 2010.

4 Sum et al., 2009.

5 In recent reentry evaluations, at least half of the sample members are fathers. Similarly, in the Parent’s Fair Share project, which targeted noncustodial parents, 70 percent of the sample members had a prior arrest.

6 Sorenson, 2010.

7 Fatherhood programs also face challenges related to multiple partner fertility: The Fragile Families project found that 59 percent of the low-income parents in that study had children with more than one partner (Carlson and Furstenberg, 2006), a pattern that complicates the targeting of services to increase father-child engagement.

8 Raphael and Stoll, 2009.

9 West, 2010; Bureau of Justice Statistics, 2010.

10 Langan and Levin, 2002.

11 Travis and Waul, 2004.

12 Bushway, Stoll, and Weiman (eds.), 2007.

13 Domurad and Carey, 2010.

14 Visher, Winterfield, Coggeshall, 2005.

15 Aos, Miller, and Drake, 2006.

16 The table and subsequent discussion is adapted from Bloom, 2010.

17 Hollister, Kemper, and Maynard (1984); Gueron (1991); Uggen (2000).

18 Redcross et al. (2009); Bloom et al. (2009).

19 Brock, Butler, and Long (1993). It is important to note that, in assessing impacts on employment and earnings, the evaluations generally did not count work experience participants as employed.

20 Bloom, Miller, and Azurdia (2007).

21 Derr (2008).

22 Kerachsky et al. (1985).

23 Bell and Orr (1994).

24 Auspos, Cave, and Long (1988); Freedman, Bryant, and Cave (1988).

25 Orr et al. (1996).

26 Corson and Haimson (1996).

27 Gueron (1984).

28 Diaz, Ball, and Wolfhagen (1982).

29 Jastrzab, Masker, Blomquist, and Orr (1996).

30 Several random assignment studies have found positive results for the Individual Placement and Support model of supported employment. (See, for example, Bond et al. (2001). One study compared an IPS model to a model called “transitional employment” in which the employment agency, not the client, contracts with the employer. IPS generated much higher rates of competitive employment Mueser et al. (2004).

A-7