Mr. Chairman, members of the Subcommittee, it is a pleasure to be here today to discuss the role
of research as one critical element in the success of welfare reform. This Subcommittee
played a key role in making research and evaluation activities integral to TANF. To illustrate
how ACF is using the authority and funding provided by the Congress, I will describe our state
evaluation activities, in terms of both what we have already begun to learn and what we can
expect to learn in the future.
A central focus of ACF's welfare reform research and evaluation strategy is to develop reliable,
credible information about how different strategies are working in order to inform the state
policy choices that TANF flexibility provides. This information can also inform the public and
the Congress about how welfare reform is progressing.
A key to developing this wealth of information is through the use of experimental evaluations of
outcomes. These evaluations can shed light on whether state reforms are meeting the goals of
TANF as set forth in the statute--providing assistance to needy families with children, ending
dependence through increasing employment and marriage, reducing out-of-wedlock pregnancies
and encouraging two-parent families.
My testimony today will focus on current evaluations of state programs, primarily of existing
state waiver demonstrations, and future areas to consider for evaluations. I also will share some
of our early lessons from these current evaluations. But before beginning this discussion I first
would like to take a quick, broader look at other research activities funded by ACF.
The waiver evaluations are only one part of a broader welfare reform research and dissemination
effort taking place in ACF. Last year, for example, we were able to fund nine projects (from
over 100 applicants) that are examining a variety of key issues in welfare reform, including
implementation of tribal TANF programs, identification of problems and solutions in
implementing TANF in rural areas, and a study of different diversion programs that states are
operating. In addition, we are making careful investments to ensure that knowledge gained
through evaluation is widely disseminated in formats that program operators find accessible,
including the very important Manpower Demonstration Research Corporation (MDRC) "How-to
Guide" entitled Work First, How to Implement an Employment-Focused Approach to Welfare
EFFECTIVENESS EVALUATIONS OF STATE PROGRAM INNOVATIONS
Over the past twenty years, the application of experimental approaches to studying the effects of
welfare reform has proven to be enormously important to finding out what is effective and
what is not. Because of its proven track record, in almost all cases ACF and the state agreed on
experiments as part of the approval of their waiver demonstrations begun prior to enactment
of the Personal Responsibility and Work Opportunity Reconciliation Act. States randomly
assigned families either to continue receiving welfare under the old AFDC rules, the "control
group," or to another group that received welfare under the reform regime, the "experimental
In authorizing the Department to fund the continuation of these evaluations under TANF,
Congress provided the vehicle for obtaining the earliest reliable information on the effects of
different state welfare reform strategies in meeting the objectives of TANF. ACF has funded
nine states to continue their evaluations with only minimal changes and an additional eight to
modify their evaluations.
Since most of the nine states are continuing waiver policies into TANF with minimal change, by
comparing the experimental and control groups over time in a state, we can reliably determine
the causal effect of a state's reform effort on such key outcomes as welfare dependency,
employment and earnings, total family income, and family structure. And by looking across
states, we can begin to compare the relative effectiveness of different strategies in achieving the
various goals of reform.
The ability to look across states is enhanced by the fact that the nine states embody significantly
different approaches to key policy decisions. For example, they include states with a variety of
time limits: states with strict and standardized time limits, states with individualized time limits,
states that prescribe work after a time limit, states that eliminate the benefit of the entire family,
states that only remove the benefit of the adults and states with no time limits.
The outcomes in the studies described above primarily concern adults. To better measure the
effects of welfare reform on children, we created a team of twelve states to think collectively
about what outcomes, both positive and negative, they thought their reforms might produce. We
also included researchers to inform the team on how best to measure these outcomes.
We now have been able to fund five of these states to add a common core of child outcome
measures to their evaluations-- Connecticut, Florida, Indiana, Iowa and Minnesota. Because
these states represent varying approaches to welfare reform, the results will produce rigorous
information on how different reform policies affect children.
A LOOK AT FUTURE EVALUATIONS
The main question that the waiver evaluations answer is, "What has the totality of the change in
state policy brought about?" But there is another equally important area of study to examine
decisions that states need to make about the parts of their programs, if they are to invest their
TANF and state funds most effectively.
A particularly important example that illustrates this kind of evaluation is the increasing efforts
by states to improve job retention and advancement. Over the last several years, it has become
clear that many welfare recipients do work, but their work is sporadic with substantial periods of
unemployment and dependency interspersed with work. In addition, many of these families
continue to have low incomes. We have learned a lot over the last twenty years about ways that
are most effective to move recipients into work, and most states have moved to "work first"
approaches, in part based on that knowledge. Now we need to turn our attention to learning
about how to keep individuals employed more persistently, and how to raise working families,
As part of the creativity that is coming from a flexible, work-focused system, many states are
now experimenting with what are frequently called "post-employment services" in order to
increase employment persistence. For example, some states are working with employers to
provide mentors to individuals with little work experience; others are providing opportunities for
skills enhancement for former recipients who show their commitment to work; and others are
addressing access to health insurance. Rigorous evaluation can tell us which of the approaches,
or the details of them, are actually effective in increasing employment and earnings.
This example suggests how, if we continue to work with states to ensure that the evaluations we
fund answer their most pressing program design questions, we have a tremendous opportunity to
learn which strategies really achieve the goals of welfare reform. And this kind of learning can
play a critical role in the success of state welfare reform efforts.
WHAT WE HAVE BEGUN TO LEARN
In turning to what we have learned so far from these waiver demonstrations, I want to stress that
it is too early to draw inferences with a high degree of confidence. Our ability to learn from the
evaluations will certainly grow over time. Nonetheless, based on the information we have now,
the following lessons are suggested.
- Most state evaluations indicate significant impacts on the proportion of recipients and former
recipients who are employed. These impacts range from modest to quite large, depending on the
reform intervention and the population subject to it. Among the states at the higher end of the
range, the proportion of participants working in the last quarter of follow-up was 8 to 15
percentage points higher for the welfare reform group than the AFDC control group.
- States are dramatically increasing the mandatory aspect of work programs. As with time
limits, states have implemented a variety of policies for sanctioning individuals who fail to
comply with program requirements, including eliminating a portion of the grant, eliminating the
entire grant, and progressive reductions for repeated instances of failure to comply. These
sanctions, including whole family sanctions, are much more commonly applied than anyone
would have predicted. For example, in Delaware nearly half of the participants were sanctioned
for failure to cooperate, and in Florida the sanction rate was 31 percent for the welfare reform
group compared to 7 percent for the control group.
- Although states are getting out the message that welfare reform is occurring, and despite state
efforts to simplify and explain their programs, many recipients do not have a clear understanding
of the new rules. For example, they may know that there's a time limit, but do not have a good
idea of how months are counted toward it. Thus, in one state almost 90 percent of participants
knew they were under a time limit, but only about half understood the circumstances under
which a month counted. This is not a new phenomenon since recipients seldom understood
AFDC rules either.
- To date, the studies confirm earlier research on the effectiveness of more generous earnings
disregards. Earnings disregards appear to be more effective than other policies being
evaluated in raising family income for single parent families. This was most dramatically
illustrated in Minnesota where work incentives were a very central feature of the state's
In theory, more generous disregards should increase the employment of single parents, and it
appears likely that this is happening in current state reform efforts. At the same time,
they probably also increase the length of time, and amount of welfare received over time,
although the welfare received is now essentially a smaller earnings supplement. For example,
again in Minnesota, under its demonstration, which had no time limit, both duration and amount
of welfare were significantly increased. But even in states with time limits and tough work
requirements, welfare duration prior to the exhaustion of the time limit was not reduced.
Previous research suggested that the effects of more generous disregards on earnings are mixed,
stimulating participants who would not have worked, or worked only a little, to earn more, but
also stimulating those who would have worked more substantially to earn less than they would
have. The results in Minnesota seem to confirm this with average earnings increasing
substantially for long-term urban recipients, but declining substantially for two-parent families
The effects on family income of the common package of welfare reform, which combines
more stringent work requirements, more generous disregards and time limits, are not clear. At
this point, most programs appear not to have increased or decreased average family income by
much, although undoubtedly, these averages mask the fact that some families have experienced
positive effects and some have experienced negative effects. More generous programs
combining high benefits with generous disregards and no time limits appear to do much more to
increase family income, but probably also increase governmental costs at least in the short run.
Thus, over an 18-month follow-up period Minnesota reduced poverty by nearly 15 percentage
points for long-term urban recipients, but increased welfare costs by about 8 percent.
These early lessons, I believe, illustrate how the results of our evaluations hold the promise to be
highly informative to policy makers at all levels of government. our perseverance in carefully
evaluating the results of these shifts in states, programs can make a major difference in the extent
to which the goals of welfare reform are realized.
I would be happy to answer your questions.