Appendix A

Appendix A

Summary of Studies Aimed at Improving Employment Outcomes of Low Income Families

The reason that many children grow up in poverty is that they live in single-parent, woman-headed households with low earnings. Fewer two-parent households mean fewer two-parent working households, which are now the mainstay of the standard of living for so many American families. In addition, the responsibilities of a single parent mean greater difficulty balancing work and family, as well as less ability to invest time in human capital development.

Despite the very large increase in the proportion of low-income mothers that work, many still work inconsistently, at low wages or less than full-time (Acs & Loprest, 2005). In addition, for many of these mothers, because they start at low earnings levels, wage growth does not translate into significantly higher earnings (Gladden & Taber, 2000). Many low-wage jobs lack health insurance and other employee benefits. Due at least in part to low family income, the children in these families do less well on a variety of different dimensions, including attaining educational credentials and avoiding legal problems and risky behaviors.

Nor do the contributions of two parents guarantee economic security. While employment rates for single mothers rose in the 1990s, that was not the case for low-skilled men, especially for inner-city African American men without a high school diploma (Holzer, Offner, & Sorenson, 2004). Wage-earning difficulties mean that many men who are fathers are unable to contribute fully to their children’s material well-being. These problems exist for low-skilled fathers who are present in the household, as well as for those who do not reside with their children.

Findings from the various types of studies aimed at improving employment outcomes of low income families are summarized below:

• Mandatory welfare-to-work programs—studies of programs that operated in the 1980s and early 1990s and tested employment interventions that required participation by welfare recipients, but did not modify the AFDC program in other ways. Well-evaluated programs consistently, but not uniformly, show employment and earnings impacts, but the average annual earnings impacts are modest. Considerable evidence exists that the “work first” programs with an emphasis on initial job search that states began to operate in the mid-1990s, produce larger impacts, at lower cost, than earlier programs that emphasized basic education programs. In addition, there is a small amount of evidence that the effectiveness of services to improve labor market outcomes is related to the way in which the services are delivered; one rigorous study showed that an integrated service delivery model that combined employment and income maintenance functions in a single caseworker produced better employment outcomes than a model that divided those functions between caseworkers. However, little rigorous research to evaluate different approaches to service delivery has been conducted. Some have argued that “mixed” models that do not require all individuals to participate in job search as an initial activity, but rather tailor activities based on some form of assessment have more positive impacts, but the evidence for this is mixed and not strong.

• Waiver demonstrations—includes mandatory work programs like those described above, but that also typically modified other AFDC rules. The most significant finding related to employment and self-sufficiency is that financial incentives (i.e. more generous earnings disregards), combined with mandatory work programs typically increased employment and earnings for welfare mothers and increased family income. This increase in income appears to improve young children’s school performance. There is some evidence that requiring welfare recipients to participate in work activities, independent of the content of those activities, has positive effects on labor market outcomes, but no experiments have directly tested the effectiveness of different sanction regimes (neither levels nor associated services) separately from the employment services with which they are packaged. The effects of time limits and requirements less directly aimed at employment are also unclear.

• Voluntary training programs for the low-income adult population—voluntary, broader population programs, such as the Job Training Partnership Act (JTPA), for which low-income parents (including welfare recipients) sometimes qualify. Often the particular services in these programs are the same as, or are similar to, those provided in mandatory programs, but frequently they also include occupational skills training, which is less typically provided directly by welfare agencies. A number of evaluations of voluntary programs have shown earnings impacts more sizeable than those of mandatory welfare-to-work programs; however, direct comparisons of impacts between these two kinds of programs are questionable. Multiple evaluations have shown sizeable earning impacts for low-income women resulting from both skills training and various forms of subsidized work including grant diversion, on-the-job training, and the provision of transitional jobs. One limitation of the current research is that several of the evaluations of subsidized employment have included it as part of a larger intervention, in which case it has not been possible to estimate the separate effects of the direct employment part of the program. A major problem with using these results in program design is that the interventions have been limited in scale, either by the current skill level of the population that could qualify for, and benefit from, them (in the case of skills training), or by employer interest (in the case of subsidized employment). An open question remains whether these programs can be operated at larger scale.

• Work-based alternatives to welfare—at one time there was considerable interest in designing alternatives to welfare that were intended to support work. Although these programs sometimes had other elements, their centerpiece was monthly supplements to earnings. The central findings for these programs are consistent with results of studies of within-welfare supplementation of earnings using more generous earnings disregard. The programs increased employment, earnings and income, especially for those who had a lower probability of employment in the absence of the supplement. However, for those with the highest likelihood of employment and higher earnings, the programs sometimes reduced work effort. Supplements conditional on full-time work had bigger employment and earnings effects for welfare populations, but at the cost of affecting a smaller proportion of the population.

• Post-TANF projects—recent and on-going, large-scale ACF projects have examined what might be effective in helping welfare recipients and former welfare recipients sustain themselves in work and advance. Overall, in comparison to the evaluations of programs in the 1980s, almost all of which had significant positive effects, the majority of programs evaluated in these projects have thus far failed to show significant effects on employment and earnings. While the two types of programs have operated in very different times and conditions the point of the recent evaluations was to find new programs that produced stronger employment outcomes than standard TANF programs, and the failure of most to do so thus far is, therefore, a disappointment.

• Overcoming personal and family problems—another important post-TANF category of programs which is currently being studied are interventions that attempt to remove “barriers” (e.g. substance abuse) that are thought to impede employment. The literature on “barriers to employment” is quite mixed. Although more recent studies have found a statistical relationship between employment and some of these barriers for welfare recipients, there is little evidence that this relationship is causal, and if so, in what direction. Thus, the effectiveness of this approach as a strategy to increase employment is unclear.

Appendix A A-3