5,829 welfare recipients.

Random assignment to 2,983 program and 2,846 control group members.

To determine the overall effects of the EMPOWER demonstration and to provide detailed information about the outcomes attributable to the separate provisions of the EMPOWER demonstration.

To provide information about the replicability of the program and its impacts.

To examine if families were better off with EMPOWER than they would be without the intervention (based on gains in earnings that surpass losses in benefits) and if the government’s budget is better off with EMPOWER (based on declines in assistance payments that offset operating costs).

• EMPOWER served to increase employment among those who were TMA recipients when the reforms were first implemented.
• EMPOWER succeeded in moving people off cash assistance and food stamps, but unintentionally reduced Medicaid rates for those receiving cash assistance when the reforms were first implemented.
• However, the reduced participation in welfare programs was not achieved through increased employment, as the reforms had intended.
• EMPOWER decreased the proportion of families in which unwed minors gave birth.

• The first-round site visits conducted in the fall of 1996 found that EMPOWER was being implemented largely as intended and largely in a consistent fashion across the four original research sites.
• The most important finding regarding the implementation of EMPOWER in the original research sites is the increasingly blurred distinction in the policies that applied respectively to the experimental and control groups.
• Another important finding regarding the implementation of EMPOWER was that the “office culture” in the four research sites had not changed since the first-round visits. That is, these offices had not implemented the “Work First” philosophy intended by Redesign.
• Much of this has changed in the Redesign offices. The Work First philosophy is being implemented in the four Redesign offices, although not to the same degree in each of the four offices. The large centers in Phoenix are not implementing the philosophy as well as the Fort Lowell and Tempe sites. This may be due to their large size. However, communication among the three agencies (FAA, JOBS, and CCA) has been helped by the fact that the workers are all in the same location.
• The Redesign offices are more effective in conveying to clients that they need to get out and find a job, however. The factors that are having the greatest influence in accomplishing this are the new, more stringent sanctions; fewer exemptions from mandatory participation in JOBS; and the change in the way clients are processed, from seeing an eligibility interviewer before being sent to a JOBS office, to first seeing a JOBS case manager or Job Service worker and participating in Job Search. This is the essence of the Work First policy, and it is working according to EIs, unit supervisors, and JOBS staff.
• The Navajo reservation is an entirely separate situation. They are implementing EMPOWER and Redesign quite differently than the other sites. A major difference on the reservation is that they are exempt from the two-year time limit.
• At the time of the site visit to the Chinle office, the JOBS workers were less likely to impose sanctions for clients who did not comply with JOBS than in the Phoenix offices. Another difference is that JOBS workers in Chinle seemed to be more genuinely concerned with clients’ needs. The reservation staff were moving toward implementing welfare reform themselves. Therefore they would not become Redesign offices similar to the Phoenix and Tucson area offices. Finally, because of the lack of jobs on the reservation, and the high unemployment rate, moving people off welfare is much more difficult there.

• Job quality increased in the Phoenix sites, with increased percentages employed full-time, at high wages (> $7.50/hr), and at jobs providing health insurance and dental benefits. In contrast, the employment gains in the Navajo site were predominantly at lower wage jobs ($6.50 or less per hr).
• In the Phoenix sites and Navajo site, there was reduced dependence on public benefits: cash assistance, food stamps, Medicaid, and housing assistance.
• Total household income showed little net change between follow up months 30 and 48; reduction in monthly cash assistance and food stamp benefits offset the modest gains in income.
• Substantial improvements have been made in the delivery of program services in local welfare offices as perceived by clients recently approved for cash assistance.
• Clients seemed generally well informed on key program rules. Some rules however such as school attendance and child immunization requirements, appeared inadequately explained.
• Clients reflected an awareness of the importance of finding employment, so that cash assistance could indeed be transitional. The use by clients of employee related services appeared low.
• Personal interviews and focus group sessions with selected former recipients of cash assistance provided additional insight that, although not generalizable, were nonetheless instructive in confirming other evidence and raising other questions.