This report presents early evidence on the effectiveness of four diverse programs that aim to help current or former welfare recipients maintain stable employment and increase their earnings. The four programs are part of the Employment Retention and Advancement (ERA) project, which is testing 15 such programs across the country. The ERA project was conceived and funded by the Administration for Children and Families (ACF) in the U.S. Department of Health and Human Services (HHS) and is also supported by the U.S. Department of Labor (DOL). The project is being conducted by MDRC, a nonprofit, nonpartisan research organization, under contract to HHS.

The results described in this report are far from the final word on ERA. For the most part, results are presented for only one year after individuals entered the four programs — a short follow-up period for interventions focused on long-term employment goals. In addition, at this point, the analysis includes only a subset of the study participants — typically, those who entered the four programs during the early, formative months of operations. Finally, results are currently available for only 4 of the 15 ERA experiments. A series of additional reports by MDRC will present more definitive results for a larger number of ERA sites.

The Policy Challenge

There is great interest in finding effective strategies to help low-wage workers maintain employment and advance to better jobs. Broad economic trends have decreased the number of stable, well-paying jobs for workers without a college education. At the same time, welfare reforms have encouraged and required millions of single mothers with low skills to enter the labor market and have reduced the availability of long-term welfare support. For many former welfare recipients, stable employment and wage progression have become a matter of economic survival.

Yet, although a great deal is known about effective strategies to help welfare recipients and other disadvantaged groups find jobs, there are almost no proven approaches for helping them keep jobs or advance in the labor market. The most comprehensive test of programs providing postemployment case management services to welfare recipients who went to work — which was conducted in four sites in the 1990s — found that such services did not improve employment outcomes.

The ERA Project

The ERA project was designed to improve on past efforts by identifying and testing innovative models designed to promote employment stability and wage progression among welfare recipients or other low-income groups. The project began in 1998, when HHS issued planning grants to 13 states to develop new programs. The following year, MDRC was selected to conduct an evaluation of the ERA programs.

From 2000 to 2003, MDRC and its subcontractor, The Lewin Group, worked closely with the states that had received planning grants — and with several other states — to mount tests of ERA programs. MDRC, Lewin, and Cygnet Associates also provided extensive technical assistance to some of the states and program operators, since most were starting programs from scratch with no proven models on which to build.

Ultimately, a total of 15 ERA experiments (also called “tests”) were implemented in eight states. Almost all the programs target current or former recipients of Temporary Assistance for Needy Families (TANF), the cash welfare program that mainly serves single mothers and their children, but the program models are extremely diverse. One group of programs targets low-wage workers and focuses strongly on advancement. At the other end of the spectrum, another group of programs targets individuals who are considered “hard to employ” and aims primarily to place them in stable jobs. Finally, a third group of programs has mixed goals and targets a diverse set of populations, including former welfare recipients, welfare applicants, and low-wage workers in particular firms. Some of these programs initiate services before individuals go to work, while others begin services after employment.

The evaluation design is similar in most of the sites. Individuals who meet the ERA eligibility criteria (which vary from site to site) are assigned, at random, to a program group (also called “the ERA group”) or a control group. Members of the program group are recruited for — and, in some sites, required to participate in — the ERA program, while those in the control group are not eligible for ERA services. The extent and nature of services and supports available to the control group vary from site to site, but it is important to note that, in most sites, the ERA program is not being compared with a “no services” control group.

To track both groups over time, MDRC is using surveys and administrative records (data on welfare and food stamp payments and quarterly earnings in jobs covered by unemployment insurance). The random assignment process ensures that the two groups were comparable at the start; thus, any differences that emerge between them over time (for example, in employment rates or average earnings) are attributable to the ERA program.

The Sites Discussed in This Report

This report presents early results from four of the earliest-starting ERA sites:

• Illinois. Operating in Chicago and St. Clair County, the Illinois ERA program targets a group that appeared to be “stuck” in low-wage jobs: TANF recipients who reported full-time employment to the welfare agency for at least six consecutive months and yet continued to qualify for cash assistance. Operated by contracted service providers, the program provides a range of services designed to help participants increase their earnings in their current job or, more typically, to find a higher-paying job. The control group is not referred to an ERA service provider but may receive services from welfare office staff.



• Riverside County, California. The Riverside program targets newly employed TANF recipients and aims to promote advancement by testing two alternative models designed to encourage and assist participants to enroll in education and training activities. One model is operated by the welfare agency and requires recipients to continue working at least 20 hours a week while participating in education or training. The other model is operated by the workforce development agency and allows participants to reduce their work hours or stop working to participate in education or training. Recipients also may be randomly assigned to a third group, similar to a control group, that receives some postemployment follow-up but no strong encouragement to participate in education or training.



• South Carolina. Operating in six rural counties that make up the Pee Dee Region, the South Carolina ERA project targets former TANF recipients who have been off welfare for a long period. The program reaches out to these individuals and seeks to help improve their labor market outcomes. Depending on the client’s circumstances, the program might provide job placement help, employment retention services, or advancement-focused activities. There is no outreach to the control group.



• Texas. Operating in Corpus Christi, Fort Worth, and Houston, the Texas ERA program targets TANF applicants and recipients, most of whom are not employed. Using financial incentives (a stipend of $200 per month for individuals who leave welfare and work full time) along with team-based case management and other services, the program seeks to move participants into jobs, stabilize their employment, and help them advance. The control group is subject to the state’s regular welfare-to-work program, which includes extensive preemployment activities but only limited postemployment services and no stipend.

Early Results

Two aspects of the research design are critical to interpreting the early ERA results and comparing them with the results from earlier studies. First, the states participating in ERA —  like almost all states — have implemented aggressive measures to promote employment among welfare recipients. Such measures have profoundly shaped the outcomes for both the program group and the control group and have created high benchmarks for the ERA programs to overcome. In other words, any effects produced by ERA must be over and above the already substantial effects of state welfare reform efforts.

Second, none of the ERA projects targets motivated volunteers. All are reaching out to individuals who, based on their characteristics, were seen as potential beneficiaries of retention and advancement services. Moreover, unlike most preemployment welfare reform strategies tested in the past, ERA programs generally do not have the means to mandate participation in postemployment services. Yet, because of the random assignment research design, all potential participants are part of the program group, and the analysis of program effects includes both participants and nonparticipants.

• The four programs discussed in this report, like virtually all the other ERA programs, have faced a substantial challenge in increasing participation in retention and advancement services.

In order to achieve their ultimate goal of improving employment stability and earnings, it is assumed that the ERA programs must first ensure that program group members receive a substantially greater “dose” of retention and advancement services than the control group. This has proved to be a daunting challenge for two reasons. On the one hand, many program group members have been difficult to locate or were reluctant to participate, particularly in postemployment services; typically, these are single parents struggling to balance low-wage work with family responsibilities, and they may have little time or energy for additional activities. ERA staff — assisted by MDRC and its partners — have developed many innovative strategies to sell ERA services to potential participants; in fact, most of the programs have managed to have face-to-face contact with a very high percentage of the program group. Yet almost all the programs have struggled to keep participants engaged and active over time.

On the other hand, preliminary results from client surveys suggest that services similar to those provided by ERA are sometimes available to control group members who want them. In addition, in some sites, members of the control group may be required to participate in employment-related services as a condition of receiving public benefits. The services received by control group members are probably less intensive than those provided by ERA — and are probably less focused on retention and advancement — but they may affect employment outcomes.

Together, these two factors mean that the difference in service receipt between the two research groups may not be as large as expected, a result that may diminish the ability of some of the ERA programs to affect employment outcomes. MDRC is currently conducting additional analysis to better understand the extent and nature of the service difference in each site.

• The early effects on employment outcomes are mixed: Some of the ERA programs appear to be promoting retention or advancement, while others seem to be less successful.

The ERA program in Chicago has generated modest increases in earnings and employment outcomes. (Results are not reported for St. Clair County because of the small sample size.) For example, the program group earned, on average, $539 (9 percent) more than the control group during the first year after enrollment. Effects seem to be particularly large for individuals who did not work in jobs covered by unemployment insurance (UI) in the months prior to entering the study. It appears that ERA may be moving some participants from informal jobs that are not covered by UI into UI-covered employment. This is likely to be a positive result, since UI-covered jobs may be of higher quality; for example, they may be more likely to offer fringe benefits.

The Illinois program has also generated a large decrease in TANF receipt. By the end of Year 1, only 40 percent of the ERA group were still receiving TANF cash assistance, compared with 55 percent of the control group. The decrease in TANF receipt may have occurred because ERA participants obtained higher-paying jobs that made them ineligible for welfare. Or, in contrast, some program group members may have closed their welfare cases — without obtaining higher-paying jobs — to avoid the obligation to participate in ERA.

The early results in Texas vary among the three sites. The ERA program in Corpus Christi, which was implemented more smoothly than the programs in Fort Worth and Houston, produced some improvements in employment retention outcomes. For example, the proportion of sample members in Corpus Christi who worked in four consecutive quarters was 31 percent for the ERA group and 26 percent for the control group; the difference was larger — almost 8 percentage points — among the subgroup of individuals who worked in UI-covered jobs just prior to enrollment. The Corpus Christi program has not generated impacts on other key outcomes, however, and the Fort Worth and Houston programs show few early effects.

The South Carolina ERA program generated some statistically significant increases in employment, but these effects were inconsistent and short-lived. Individuals who enrolled during the early months of operations experienced some increases in employment during their first year after enrollment. However, these gains appeared to evaporate by the end of Year 1, and there were few employment gains for people who enrolled later. The program has had few effects on retention or advancement outcomes.

So far, there is little evidence that either of the education- and training-focused models being tested in Riverside is generating improvements in labor market outcomes, although a one-year follow-up period may be too short to reveal such effects. Preliminary data from a client survey raise questions about whether the two models succeeded in increasing participation in education and training above the level of the control group; MDRC is collecting more data to better understand these findings. If participation did not increase, the study may not provide solid evidence about whether postemployment education and training can improve labor market outcomes — although it may provide important data on the “normal” patterns of participation in such activities among single parents working in low-wage jobs.

• It is too early to draw broad conclusions about the effectiveness of employment retention and advancement services or to determine why some ERA programs appear to be working better than others.

Although the overall story is mixed, it is encouraging to see that some of the ERA programs appear to have positive effects. As noted earlier, past research has identified few, if any, successful retention or advancement strategies. That said, the results presented here are far from definitive. The findings are from only 4 of the 15 ERA tests; they cover a short follow-up period; and they mostly focus on people who enrolled in the programs during their startup months. Despite these cautionary notes, HHS and MDRC felt that it was important to publish results and begin stimulating discussion, rather than waiting for more definitive evidence. A series of future reports will provide additional evidence on the effects of the ERA programs.