The findings of the literature review and the advice of the project's expert panel both suggest that rigorous impact research is needed to evaluate the effectiveness of different approaches used by employers and intermediaries. Such research should be carefully designed and implemented to identify the relative effectiveness of several important practices, especially the use of labor market intermediaries.
However, it would be difficult to implement such experiments in the near future, largely because not enough is currently known about the characteristics and prevalence of specific employer practices, the extent to which hiring and employment functions are performed by labor market intermediaries, and which employer or intermediary practices are likely to be most critical to the key employment outcomes for TANF recipients. For these reasons, we recommend that additional research be undertaken in the short run and that, once more is learned about employer and intermediary practices, the feasibility of experiments be assessed.
The optimal way to obtain much of the needed information is to conduct a national survey of employers and the labor market intermediaries with whom they partner. Such a survey would provide comprehensive knowledge of recruitment, hiring, and employment practices for different types of employers and intermediaries. It also would provide a basis for identifying critical points in the hiring and employment process. More generally, such a survey would inform decision making by policymakers, government agencies, employers, and other institutions for years to come.
Other types of research are also attractive. One is additional analysis of data from the telephone survey of employers directed by Harry Holzer in the late 1990s. It would be particularly helpful to analyze the survey's data on labor market intermediaries. Another appealing research option is to analyze data from the Abt Associates survey of intermediaries that was conducted in the same period. Finally, natural experiments might be conducted in localities where TANF work-activity assignments are random or nearly random. Before undertaking such studies, however, research would need to identify potential sites, assess differences between the service providers to which TANF recipients are assigned, and determine the extent to which assignment is truly random. All three of these research options would be relatively inexpensive and potentially very valuable. In addition, they would inform consideration and design of future random assignment experiments, and could be designed with this in mind.
"report.pdf" (pdf, 212.99Kb)