The 1996 federal welfare reform law is discouraging states and localities from experimenting with some of the most promising ideas for getting people back to work, according to a recent report by the U.S. Government Accountability Office (GAO). The independent auditing arm of Congress found that incentives baked into the nation's largest cash assistance program -- Temporary Assistance for Needy Families (TANF)-- actually undermine efforts to employ the poor.

The problem, according to the report, is that the current system encourages states to focus on activities that help them meet a flawed federal performance measure: the work participation rate. The measure, said Elizabeth Lower-Basch, a welfare policy analyst at the Center on Law and Social Policy, doesn't reflect how many people on TANF are working, or even doing the things most likely to lead to work. "It's a process measure," she says. "It doesn't look at whether people get jobs."

MORE: This appears in the Human Services e-newsletter. Click to subscribe.

Instead, the participation rate requires states to make sure that at least half of eligible TANF families participate in one or more prescribed activities -- such as searching for work or job training -- for a certain number of hours per month. The report also notes that the measure adversely encourages TANF agencies to concentrate on job-ready participants or participants that might not need as much help as "hard-to-employ" participants who have health problems, disabilities, criminal records, dependence on drugs, limited education or a responsibility to care for a disabled relative. In either case, states appear to be meeting the letter of the law, but not its intent.

To better promote innovative, research-driven employment approaches, the GAO report points to four evidence-based strategies, detailed below, being implemented in 10 places around the country: subsidized employment, treatment coupled with employment services, career pathways and modified work first.

With subsidized employment, public funds create or support temporary work for someone who would otherwise be unemployed. One example is the San Francisco Jobs Now! program, which gradually reduces its contribution to a participant's wage over five months. Public funds cover 100 percent of wages in the first month, 75 percent in the second month and $1,000 per month for the next three months. Employers have to agree to try to retain the participant once the subsidy runs out. San Francisco administrators told the GAO that the employee retention rate is between 75 percent and 80 percent. Subsidized employment strategies are also being tried in Los Angeles, Erie County, N.Y., and Kentucky.

Under treatment and employment services, officials try to address barriers to employment, such as mental health needs, substance abuse or a physical disability, while helping people look for work. Utah's Licensed Clinical Therapist Program still assists people with job searches and resume building, but also offers a clinical assessment of mental health problems and clinical therapy sessions. These services are also offered in New York City and Ramsey County, Minn.

Under the career pathways model, TANF participants receive basic education while also learning skills needed for a specific job and industry, usually with guidance from local employers. In Washington state, 34 community and technical colleges train people in classrooms for careers with a demonstrated market demand in the region, such as health care, early childhood education and advanced manufacturing. Kentucky and Minnesota are also experimenting with career pathways models.

"Work first" refers to mandatory work-related activities, such as job searching, rather than education and training that might lead to a job later on. The district's modified work-first program differentiates people who are ready for a job and those who need to upgrade their skills, experience or education before looking for work.

The GAO review of 10 jurisdictions experimenting with employment-focused approaches underscores a paradox: All these programs aim to place people in work, but administrators from half the programs said they didn't prioritize meeting the federal work requirement. They could afford to do that, in part, because these jurisdictions are part of larger TANF programs with other components that focus on the participation rate.

Despite the fact that research suggests the highlighted approaches might be some of the best ideas today for helping the poor get jobs, administrators in three of the jurisdictions said it would be difficult to secure state funding for their programs -- in light of other demands for the same money - if they started today. The GAO authors concluded that it is unlikely the lessons learned from these programs will be widely adopted until Congress changes the law that created TANF.

The federal government also ought to offer incentives to states for evaluating experimental TANF programs, the GAO authors wrote. Since the work participation rate fails to track employment outcomes, the report also suggested that TANF agencies need to conduct additional research on which approaches work best. Currently, the federal government isn't funding state research centered on employment-focused programs in TANF, and states are unlikely to take it upon themselves to do it.

There are other good reasons why TANF agency heads would be reluctant to conduct a scientific assessment of their programs -- besides the money, obviously. Rigorous evaluation comes with the political risk that it will reveal a current program isn't working. Plus, the randomized design in evaluations means that some needy residents will end up in control groups and won't benefit from a potentially effective service. But the biggest hurdle? The funding for a study is "coming out of money you could have been spending on actual services," Lower-Basch said, "which is always a hard case to make."