Series: CCF Briefs

The Role of Education and Training in Welfare Reform

Gayle Hamilton and Judith M. Gueron

In recent years, single mothers on welfare have gone to work in unprece-dented numbers. But with limited skills and work histories, they usually get low-paying jobs and remain in poverty. The situation is especially acute for the half of the caseload that does not graduate from high school. Since recipients with higher skills tend to get better jobs, it seems logical that education and training should play a central role in welfare reform. But what kind of role?

Alternative Strategies in Welfare-to-Work Programs

Welfare policy reflects an ongoing effort to balance two objectives-reducing poverty and ending dependency. Reformers on all sides favor these goals, but disagree on which should receive priority and on the most effective strategy for achieving them. As a result, states have used variants of three broad approaches to structure the welfare-to-work component of welfare reform.

Education or Training First

Adherents of putting adults on welfare into education or training programs before requiring them to find work stress antipoverty goals and view reforms that substitute work for welfare as insufficient if there is no increase in income. Before looking for work, they argue, welfare recipients need to improve their skills so they can get a job-especially one that is relatively stable, pays enough to support their children, and leaves them less vulnerable during an economic downturn. For those who lack a high school diploma or GED (high school equivalency) certificate, this view translates programmatically into referral to basic education courses, including remedial instruction in reading and math, English as a Second Language classes, or preparation for the GED test (much less common are programs that mix adult education and vocational training). For those with a high school diploma or GED, the education-or-training-first approach usually means assignment to vocational training, rather than to degree-producing, post-secondary academic courses.

Job Search First

Others place greater emphasis on reducing the welfare rolls and saving money. They advocate strategies that move people quickly into jobs, even if the jobs pay low wages. Some, focused on welfare reduction, see work as the most direct route to ending what they view as the negative effects of welfare on families and children. Some focus on the savings to be attained by both diminished caseloads and the relatively low cost of job search services themselves, reasoning that, given fixed budgets, they can serve more people using this strategy. A job-search-first strategy can also reflect antipoverty goals. Some hold that getting a job, even a low-paying one, is the best way to build skills that can eventually lead to better jobs. Others believe that, in any labor market, most welfare recipients will inevitably get low-wage jobs and that the best, most realistic way to reduce poverty is through more generous subsidies and services to working families. In job-search-first programs, virtually everyone must start by looking for a job independently or through a job club, which teaches such skills as résumé-writing and interviewing. After several weeks, participants typically get aided in their search by program staff. Job search first is usually not, however, job search only. People who fail to find work may be referred to education or training.

Mixed Strategy

Some reformers favor a more flexible approach, allowing staff and participants more choice in the initial and subsequent activities. Some participants, usually those lacking a high school diploma or GED, are assigned initially to basic education or training, while others are most often assigned first to job search. Subsequent activities vary for those still on welfare. Some mixed programs strongly emphasize employment: staff urge participants to find work and permit only short-term education or training activities. Others emphasize skill-building: participants may enter long-term education or training programs, and getting a job quickly is not paramount.

Education and Training in the Context of Welfare Reform

Since 1971, federal welfare legislation has required that an increasing share of welfare recipients participate in some form of work-directed activities as a condition of receiving full (or, more recently, any) welfare benefits. Even without any special welfare-to-work program, however, many low-income people enroll in school, training, community college, or some other program to help them gain skills and find work. This voluntary activity may have a big payoff, but it is not due to welfare reform and cannot reliably be captured in studies of reform programs.

Thus, asking about the value of education and training as part of welfare reform has a special meaning: does requiring education or training for people who may or may not want to participate have the intended positive results relative to what people would have achieved on their own or to other approaches such as job search? This question is particularly relevant to mandatory basic education, since few welfare recipients (only 8 percent in some studies) state that they want to go back to school to study reading and math; they have had poor experiences in school in the past and prefer to get specific skills training (around 60 percent) or help looking for a job (about 30 percent).

Related Books

The Studies

The research on these three strategies, based on programs that operated between 1985 and 1999, is unusually reliable because it:

  • covers programs representing a variety of specific approaches and conditions;
  • includes results from almost 100,000 single parents, a sufficient number for reliably assessing the programs’ effects;
  • follows people for five years, long enough to determine whether an up-front investment in education or training pays off;
  • measures what the three strategies produce when implemented under real-world conditions; and
  • uses random assignment, the most powerful research design, in which welfare recipients are placed through a lottery-like process in a mandatory welfare-to-work program or in a control group. Control group members are not required to participate in any activities but can (and very often do) seek out such services in the community.

The last factor is the most fundamental. By assigning people randomly to either a welfare-to-work program or a control group, the studies can safely attribute any subsequent difference in their or their children’s behavior to the particular program strategy. These differences are called the program’s “impacts.” Throughout this brief, saying that a program increased some outcome—for example, earnings—does not refer to how people’s behavior changed over time, but to how people subject to a particular welfare-to-work strategy performed relative to the study’s control group.

The findings come primarily from comparing results across twenty programs in five of the largest welfare-to-work studies-the National Evaluation of Welfare-to-Work Strategies (NEWWS; 11 programs), the evaluations of California’s Greater Avenues for Independence Program (GAIN; 6 programs), Los Angeles’s Jobs-First GAIN, Florida’s Project Independence, and San Diego’s Saturation Work Initiative Model (SWIM)-and from a head-to-head test in NEWWS of the first two approaches. Thus, this brief builds on the work of many people, especially researchers at the Manpower Demonstration Research Corporation in New York City who conducted these studies and analyzed the results.

The studies were launched prior to the 1996 welfare reforms (some of the programs continue today with modifications) and thus assessed the impact of different pre-employment strategies before there were time limits on welfare receipt, more generous limits on what people can earn and still receive welfare, and larger penalties (sanctions) for noncompliance with the program. The implementation of the 1996 welfare reforms might change somewhat the magnitude of the impacts, but would be unlikely to affect the relative success of the three strategies.

The Findings


All three strategies increased single parents’ work and reduced welfare receipt compared to what would have happened in the absence of the programs, but they did not increase people’s income or have many or consistently positive or negative effects on children, except for adolescents.

People in job-search-first programs took jobs sooner. Those in education-or-training-first programs eventually caught up, but the larger up-front investment had no clear payoff in higher wages or income, or in improved outcomes for children, relative to job-search-first programs.

The best results came from programs that used a mix of initial activities, where some people started with job search and others with short-term, work-focused education or training. These findings hold true for high school graduates and nongraduates alike.

Twenty-Program Comparison

Figure 1 shows the impact, over the five years after a person enrolled, of programs that used different variants of the three strategies. Each bar represents one program and shows the difference between the average total earnings (top panel) or welfare payments (bottom panel) of all single parents required to participate in the program and all those in the study’s control group. The top panel shows that all the programs increased earnings, with almost all differences reaching statistical significance, but those that used a mixed strategy tended to have the largest impacts. The bottom panel shows that welfare savings were largest in job-search-first and mixed programs that focused on employment.

The mixed-strategy programs that were employment-focused (Portland and Riverside GAIN) emerged as clear winners, producing unusually large earnings gains and taxpayer savings and, for Portland (not shown), more stable employment and higher wages. The Florida results in figure 1 show, however, that this approach does not guarantee success. Other features of the Florida program-limited child care funding, weak job search activities, and a rigid method for determining who received education or training-probably compromised its success.

Although most programs increased earnings, they reduced welfare and food stamp payments by a similar amount. Over five years, people derived more of their income from earnings but were generally not better off financially as a result of the program compared to control group members. These findings hold even when estimates of Earned Income Tax Credits, state and federal taxes, and Medicaid are included in the calculations.

Figure 1 is persuasive because it shows patterns replicated over a number of locations. But there is always a question as to whether such cross-site comparisons reflect differences in the value added by the welfare-to-work strategy itself, the characteristics of the people studied, the local economy, or the welfare and community context.

Three-Site Test of Two Approaches

To eliminate this uncertainty, NEWWS fielded a highly unusual study in Atlanta, Georgia; Grand Rapids, Michigan; and Riverside, California. In each of these sites, welfare recipients were assigned at random to one of three groups: a job-search-first program that allowed short-term education or training only for those who did not initially get work through job clubs (an approach labeled “Labor Force Attachment,” or LFA, in this evaluation); an education-or-training-first program that assigned most people to education or training before requiring job search (called “Human Capital Development,” or HCD); or a control group. A comparison of results for the LFA and HCD groups, presented in figure 1, reveals few differences: the HCD approach did not produce the expected added benefits. Any differences for particular years or measures or subgroups that did occur favored the LFA programs.

The five-year results shown in figure 1, mask a strong difference in the pattern of impacts over time. People in the LFA programs found jobs and got off welfare sooner, a clear advantage when welfare is time limited. People in the HCD programs caught up with those in the LFA programs some time after leaving education or training, but did not end up in higher-paying, more stable jobs, even though the HCD programs ultimately cost 40 to 90 percent more to operate. Finally, there was no difference in the effects of the two approaches on the well-being of children, despite some hope that the HCD parents’ greater attendance in education or training might lead their children to do better in school. In NEWWS, both types of programs had few effects or, in the case of adolescents, some negative effects on a few outcomes such as grade repetition.

Looking at different subgroups within the welfare caseload, this basic pattern held true for most groups, including those with different skills, work history, and race/ethnicity. The findings were particularly disappointing for those without a high school diploma (or GED) and for other highly disadvantaged groups who were expected to benefit most from the initial investment in basic educa-tion. Whether because of the quality of the services or the short time that most people stayed in them, people without a high school diploma in the HCD programs did not measurably improve their reading or math literacy or end up with better jobs than those in the LFA programs. Quite the contrary: where differences showed up, it was the LFA programs that led to higher earnings and income. In a nonexperimental analysis, however, Johannes Bos of the Manpower Demonstration Research Corporation finds some evidence that the small number of women in the HCD programs who obtained a general equivalency degree may ultimately have had higher earnings, particularly if they went on to vocational skills training. But he also found that staying longer in general equivalency degree test preparation classes would not have appreciably increased the proportion of women who obtained this credential.

The findings on training for welfare recipients with a high school diploma or GED are somewhat more mixed. The most relevant data, again from the HCD/LFA comparison, show no added impact from the HCD strategy. Two major evaluations of voluntary programs suggest some reasons why the training-focused programs did not perform better. The National Job Training Partnership Act Study (JTPA), led by Larry Orr of Abt Associates, found that, on average, classroom skills training did not increase the earnings of welfare recipients, although other JTPA activities that included a combination of on-the-job training and job search did. In the Minority Female Single Parent Demonstra-tion, John Burghardt at Mathematica Policy Research studied four remedial education and skills training programs for single mothers, most of whom were on welfare. One program, the Center for Employment Training in San Jose, increased earnings and wages. Researchers attributed this success to the program’s strong connection to the job market, its integration of education and training curricula, the absence of entry tests, and easily accessible child care. These findings suggest that the unimpressive results from past training programs may derive, in part, from the inflexible structure of the courses (education preceding skills training), the people enrolled, the support services, or the types of training women are placed in. However, the evidence is thin and, importantly, does not encompass rigorous studies of training provided by community colleges or of degree-granting, post-secondary academic programs.

Characteristics of the Most Successful Programs

The welfare-to-work programs that were the most successful overall for both high school graduates and nongraduates—Portland and Riverside GAIN—were flexible about initial activities. Both programs strongly enforced participation requirements, had experience operating job search programs, stressed the importance of finding jobs (a message that permeated all aspects of Riverside’s operations), and used job developers. In Portland, however, job search participants were counseled to wait for jobs that paid well above the minimum wage and that offered the best chance for long-lasting, stable employment, whereas Riverside participants were advised to take the first job offered, since any job was viewed as a good job.

Regarding education and training, staff in both programs communicated that improving people’s employability was the goal-assignments were limited in duration (usually six months or less), and people were not allowed to “languish” in these activities without making progress. Most people not ready to enter the labor market—based on such factors as work history, education, and literacy test scores—were first assigned, in both programs, to basic education or, in Portland, to three- to five-week life skills classes or occupational training. The others—usually those who had a high school diploma or GED—were most commonly assigned first to job search or, in Portland, to life skills, vocational training, or work experience. Finally, the small number already enrolled in degree-granting, post-secondary academic programs when they entered the program were allowed to continue, provided they could obtain their degree in a short time.

The two programs differed, however, in how they provided education and training. In Portland, program administrators took the unusual step of partnering with the community college system to design and implement the courses and provide comprehensive case management. In contrast, the Riverside welfare department solely administered its program and, while using some community colleges to provide education and training, relied primarily on adult education schools, offering payments based on measures of student performance to several of the schools.

Lessons and Implications for TANF Reauthorization

These findings point to several lessons about the role of education and training in welfare reform. First, whether the goal is reducing poverty, reducing dependency, saving money, or helping children, there is no evidence to support a rigid education-or-training-first policy. The findings are particularly discouraging for mandatory basic education.

Second, there is a clear role for skills-enhancing activities in welfare reform. The unusually successful Portland and Riverside GAIN programs suggest a balanced approach, which emphasizes employment but uses some work-focused, short-term education and training.

Third, the solution to low earnings is not more of the same kind of training used in the past. Historically, training programs have often had no direct link to jobs in demand in the local economy or to local employers. They have also often shut out the most disadvantaged. Remedial education and GED test preparation programs could not retain people or, conversely, kept them for years without clear progress. All this suggests that program operators need to identify and systematically evaluate alternative pre- and post-employment approaches. The approaches should include training that fosters career advancement, integrates basic education and skills training, and engages local employers. Welfare recipients should also have access to support services that will increase program retention.

Finally, the findings reported here show that while well-designed welfare-to-work programs can substantially increase earnings and reduce dependency, there are limits to this approach. In the typical NEWWS program, after five years, working women’s annual earnings were about $12,500, 25 percent were still on welfare, and children were doing worse in their school performance and social behavior than children nationally. Recent research showing that children do better in school when work leads to higher income points to the importance of services and policies, like the Earned Income Tax Credit, that augment the efforts of the working poor to support themselves and their families. The welfare-to-work pendulum has swung from quick employment in the early 1980s, to skills enhancement in the late 1980s, and then back to quick employment in the mid-1990s—when the federal welfare reform legislation gave states great flexibility but sent a clear pro-work, anti-education message through its detailed language on what activities would “count” in federal participation rates. The Temporary Assistance for Needy Families (TANF) shift was prompted in part by the demonstrated success of Riverside’s GAIN program, but reflected Riverside’s strong pro-work message while ignoring its more balanced service mix.

This brief suggests that the pendulum has swung too far. TANF’s focus on employment is well placed, but does not encourage states to maximize the payoff that education and training can have. The frustration for policymakers is that, while the potential payoff to the flexible use of work-focused, short-term training and GED preparation is clear, the research leaves many questions unanswered. We still know little about the success of more innovative pre- and post-employment training and community college programs, yet innovation is clearly called for if welfare reform is to deliver on its potential not only to save money but also to help families increase their income.

Additional Reading

Bloom, Dan, and Charles Michalopoulos. 2001. How Welfare and Work Policies Affect Employment and Income: A Synthesis of Research. New York: Manpower Demonstration Research Corporation.

Burghardt, John, and others. 1992. Evaluation of the Minority Female Single Parent Demonstration: Volume 1, Summary Report. New York: Rockefeller Foundation.

Freedman, Stephen, and others. 1996. “Five-Year Impacts on Employment, Earnings, and AFDC Receipt.” GAIN Evaluation Working Paper 96.1. New York: Manpower Demonstration Research Corporation.

Hamilton, Gayle, and others. 2001. How Effective Are Different Welfare-to-Work Approaches? Five-Year Adult and Child Impacts for Eleven Programs. Washington, D.C.: U.S. Department of Health and Human Services and U.S. Department of Education.

Orr, Larry L., and others. 1996. Does Training for the Disadvantaged Work? Evidence from the National JTPA Study. Washington, D.C.: Urban Institute.

Riccio, James, and others. 1994. GAIN: Benefits, Costs, and Three-Year Impacts of a Welfare-to-Work Program. New York: Manpower Demonstration Research Corporation