Welfare Reform and Devolution: Looking Back and Forward

Enactment of the 1996 welfare law was about many things: expecting and requiring work; cutting welfare caseloads; making welfare a transitional, time-limited system; reducing projected spending; discouraging out-of-wedlock pregnancies. But for many, a key theme was that the very existence of a federal program stood in the way of accomplishing these and other goals, and that national goals could best be advanced by giving states broad discretion in using funds and designing programs.

With four years of experience behind us—and reauthorization one year ahead—it is important to assess not only how Temporary Assistance for Needy Families has worked overall, but also how devolution has contributed to the positive (and less positive) aspects of the experience.

Since 1996, employment among low-income parents has risen, family and child poverty have fallen, and states have expanded many services for low-income families. But welfare caseloads have fallen far more rapidly than child poverty, many families have lost benefits without finding work, and many who have found work have had little or no increase in economic well-being. It is also becoming clear that moving to a highly devolved system has made it more difficult to assure accountability and describe or measure state performance. So when TANF is reauthorized in 2002, a critical question will be how to preserve and build on a framework of state and local flexibility while better articulating and advancing national policies of promoting work, engaging those who face the most serious barriers to employment, and enhancing the economic well-being of low-income families.

The TANF Structure

The new welfare reform law did not simply devolve functions to states; it established specific funding and policy rules, requirements, and signals to encourage and discourage particular state action.

First, the law abandoned open-ended federal matching funding for state welfare costs. Instead, each state receives a federal TANF block grant, set roughly to reflect federal welfare spending during 1994—95. To receive its grant, each state must meet a maintenance-of-effort (MOE) obligation: spending for low-income benefits and services must be at least 75—80 percent of a state's previous spending. Nationally, TANF grants have totaled about $16.5 billion a year, MOE an additional $10 billion—$11 billion.

Second, federal entitlements to assistance were eliminated, and each state was given broad discretion in determining both the rules of its cash assistance programs and how to use its TANF and MOE funds.

Finally, federal goals were advanced principally through penalties, bonuses, and rules governing spending. For example, a state risks a penalty if it uses federal funds to provide assistance to a family for more than 60 months (subject to limited exceptions). It risks a penalty if it fails to increase the share of families in specified work-related activities—though it can meet this requirement if its caseload falls by a specified amount. A state can qualify for a bonus if the share of out-of-wedlock births in the state falls while the number of abortions also falls.

As a practical matter, the law is sometimes quite prescriptive, particularly in identifying groups that states may not assist. Still, the law considerably expanded states' discretion in deciding how to spend federal and state funds, whom to assist, what services to provide, who should deliver services, and how to structure approaches to welfare reform.

What's the Story?

There seems to be surprisingly little disagreement about what has happened since 1996, though there are sharp disagreements about which parts of the story are most significant, how much is attributable to TANF, and what parts of TANF have mattered most.

The areas of general agreement are many. The nation's welfare caseload has seen a historically unprecedented decline, from 5 million families in 1994 to 2.2 million in June 2000. Child poverty has also fallen, but welfare participation has fallen much more rapidly. In 1994, 62 percent of poor children received benefits; by 1998, the share had fallen to 43 percent. Most people who have left the welfare rolls are working, and the labor force participation rate among low-income single-parent families has grown dramatically. Working parents who have left the rolls typically have low earnings, with a reported median wage of $6.61 anhour and annual earnings around $8,000—$12,000. Food stamp and Medicaid benefits fall sharply among those who leave welfare. Though child care spending has grown rapidly, most who leave welfare for work (and most other low-income working families) get no child care subsidy. Families still receiving assistance are more likely to face quite serious barriers to employment—little education, little work history, higher incidences of mental and physical illnesses and disabilities. Some speak no English or have a disabled parent in the home. Some families have left assistance without finding work. Many also face serious barriers to employment. Finally, it appears that family structure has not changed dramatically since 1996.

When Congress reauthorizes TANF next year, there will be disputes about whether to view TANF as a success or a mixed story, but few will dispute that, at least in a very strong economy, the experience has been more positive than many anticipated in 1996. And broad agreement exists about some new goals: helping low-income parents keep jobs and advance through the workforce; expanding participation in programs, such as child care, food stamps, and Medicaid, that support working families; and helping families with the most severe barriers to employment. A consensus also seems to be forming around the need to restructure the child support system so that families benefit when support is paid, as well as the need to expand employment and other services to connect or reconnect fathers with their children.

In other areas, divisions will be sharper. Should funding be changed? Should federal time limits and restrictions on education and training be reexamined? Should there be protections against arbitrariness in program administration? Should stronger federal mandates require states to run work programs and impose benefit cutoffs for rules violations? Should government do more to promote marriage?Should other federal programs be block-granted?

In thinking about next steps, it is important not just to describe where we are, but to consider how we got here. If TANF is seen—at this point—as either a success or a mixed story, which of its parts have mattered most? Has devolution itself been a key to TANF's successes or to its limits?

How Has Devolution Mattered?

Many states would surely say that time limits, stronger sanction policies, greater participation mandates, and expanded supports for the working poor have been crucial to their success. This may be true, but that's a story about policy changes, not about devolution. Each of these changes was already in place in many states before 1996 under the old "waiver" system, whereby states could get permission from Washington to experiment with new policy approaches. Congress could simply have adopted a transitional, time-limited, work-focused system, either with a uniform national approach or with allowances for state discretion in filling in the details of a national policy. Devolution was not essential to accomplish this result.

What, then, has been devolution's role? Recalling the three key features of the law—the funding structure, broad state discretion, and the federal requirements and signaling—we can draw some conclusions with important implications for reauthorization.

Devolution and Funding

The amount and flexibility of money are both important. In 1995—96, many critics of the pending legislation feared a "race to the bottom" if federal standards were eliminated. The concern was not that states were uncaring, but that they were rational and that if funding was inadequate to meet needs, states would cut eligibility and benefits rather than commit new state funds when there was no obligation to do so and no corresponding federal commitment.

The experience has been different, because states have not yet faced a time of inadequate resources. With block grants set to reflect 1994—95 federal spending and caseloads having already fallen by 1996, TANF meant substantial new money for states. The U.S. General Accounting Office estimated that had states participated in TANF throughout 1997, the combined federal and state resources that year would have been $4.7 billion more than under the old system. Moreover, when states were initially hesitant to spend all available federal funds, members of Congress and others urged them not to leave money unspent while needs remained to be addressed. The result has probably been increased spending on child care, employment and training, substance abuse treatment, mental health counseling, domestic violence services, transportation, and related services. And under the TANF rules, as cash assistance declined, states could redirect resources to other low-income families, including those who had never received welfare.

Today, state and local administrators regularly say that for the first time in their professional careers, they feel they have significantly more resources to respond to service needs. They clearly value the ability to redirect funds flexibly at the state and local level. So this is probably, in part, a story about the importance of flexibility and discretion, but it's also a story about how it is easier to make choices when resources are expanding. The same discretion would likely be exercised quite differently in a time of scarcity.

For example, one dispute during the 1995—96 debate over welfare reform was whether to have a maintenance-of-effort requirement. Since 1996, state TANF-related spending in most states has hovered at or slightly above the required MOE level. If that level were less, state spending would likely be less. Indeed, had there been no MOE requirement, state spendingand programmatic choices might well have been quite different.

Another dispute in 1996 concerned how block grants would work in a recession. We still don't know. But as caseloads fell, states were encouraged and pressured to commit block grants to an array of needs. So if caseloads suddenly began going up, states would be forced to cut other programs, restrict assistance, or commit more state funds. In such a context, the meaning of state discretion would be very different.

Devolution and Measurement Issues

Fostering accountability and measuring performance in a context of extensive devolution is difficult. During 1996, a popular slogan was that Washington should give states broad flexibility for welfare reform and then hold them accountable for results. It has been much easier to implement the slogan's first half than its second.

Unless one is in favor of devolution for its own sake, the reason for promoting devolution is that it is seen as a better way to advance goals. But knowing whether welfare reform goals are being met requires answering a set of questions. What policy choices were made? How were they implemented? How was funding used? Who received help? What kind of help? How did it matter? What states met goals most effectively? What approaches worked best, and what can they teach about promoting best practices in other states?

Answering these questions is hard, partly because of limits in the federal law and partly because of difficulties inherent in description and measurement in a devolved framework. TANF requires each state to submit a biannual state plan, but requires little content in that plan. Many plans provide only skeletal information about policy choices or how funds are being used. Some private groups have tried to describe state policy choices, but, at best, they describe a limited number of formal policies at a given time. It is much more difficult to provide any sense of how the policies are implemented—for example, how the state's work requirement works in practice, what happens when an applicant walks in the door, what services go to which families, who gets sanctioned and why. And in states that have devolved further to counties or localities, it is sometimes not even meaningful to talk of a state policy. States report only minimal data about broad categories to describe how they are spending $27 billion in federal and state funds.

As to performance, we can look at state data about caseload declines, but in the absence of common eligibility rules across states, one cannot readily tell to what extent a decline in a state reflects reduced need or contracted eligibility. Five states "won" a bonus for reducing out-of-wedlock births in each of the past two years, though there is little or no evidence that state policies or performance, as opposed to demographic shifts or other factors, explained the winners. The federal government put in place a "high performance bonus" to measure state success in maximizing employment entries, sustained workforce participation, and earnings growth, but again it is unclear why the winners are winning or to what extent performance is being measured.

Some of these problems are not new and would arise whether or not devolution had occurred. Before 1996, state performance was largely judged on accuracy in determining eligibility and payment amounts. Any effort to measure broader outcomes would have posed challenges. But it is important to acknowledge that devolution has entailed real losses in information about state policies, practices, use of funds, and provision of benefits and services.

Devolution and Federal Signaling

Clearly, federal signaling can affect state behavior, and the federal "caseload reduction" signal has had troubling consequences. Critics of the 1996 law feared the elimination of federal protections, as states were given broad discretion in determining whom to assist, how long to provide assistance, and when to terminate assistance. While allowing discretion, the 1996 law was accompanied by strong messages to states. The strongest message, particularly in the earliest stages of TANF implementation, was the importance of caseload reduction.

States responded to that message. Many of the most troubling stories about state conduct since 1996 concern efforts to cut caseloads. Most states did not cut benefits or directly restrict eligibility, except in areas such as assistance to legal immigrants, where federal law required or encouraged it. However, states and localities did restrict assistance in an array of day-to-day practices, as many jurisdictions have imposed new requirements on applicants for assistance and have reduced and terminated grants when requirements are not met. While most people leaving assistance are working, a large group—about 40 percent—is not. As noted, families who lose benefits without finding work face multiple obstacles to employment. According to the Urban Institute, 71 percent of families who left assistance with no more than one obstacle to employment were working, while only 24 percent of those who left with two or more obstacles were working. In some states, the use of sanctions—grant reductions and terminations for noncompliance with program rules—has accounted for a major part of the caseload decline. Studies have consistently found that sanctioned families have less education and less work history than other families receiving or leaving welfare. A recent Michigan study found that they were also more likely to have been victims of domestic violence and to be mentally ill. A three-city study found that sanctioned leavers had an 89 percent poverty rate after leaving welfare.

The federal TANF law doesn't force states to sanction mentally ill parents or victims of domestic violence, but it doesn't discourage it either. Some states have begun to recognize that reducing caseloads without providing safeguards causes families with the most severe employment barriers to fall out of the public system. But current law neither requires nor encourages states to work with families to identify and address obstacles to employment. So when a parent misses an appointment or otherwise fails to meet program requirements, there is a federal fiscal incentive to terminate assistance, but not to make additional efforts to work with the family.

While the law stressed the need to reduce caseloads, it provided no clear message to states about what to do after that. It gave no comparable messages about addressing job quality; assuring food stamps, Medicaid, and child care for families beginning work; or using freed-up block grants to improve the economic well-being of low-income families. And where there was no clear federal message, state responses have been far more uneven.

Implications for Reauthorization

Thinking about what devolution has and hasn't accomplished since 1996 suggests some important issues for reauthorization.

First, funding must be adequate. As Congress considers block grant levels and maintenance-of-effort obligations, some will urge spending cuts in light of caseload declines. But we still have no way to judge the extent of need during a recession. And as welfare caseloads have declined, states have begun addressing the needs of a much larger group of low-income families. If states are expected to pursue this broader mission, they need more, not fewer resources. An essential point, though, is that the story of the past four years is not about devolution per se, but about state and local flexibility amidst ample resources—at least for the goals articulated in 1996. If Congress wishes to sustain the momentum and takefurther steps to advance the economic well-being of families, the funding must be sustained.

Second, accountability should be strengthened. Debate should focus on how, without unduly restricting state flexibility, to generate better information about state policy choices and use of funds; to develop better data about who is being assisted, how, and with what outcomes; and to develop better measures of state performance. For example, surely state plans should describe what states intend to do, and annual reports should explain what they did. Financial reporting should be detailed enough to give a clear picture of how funds are being used. And a serious effort must be made to reach consensus on outcomes for which it is reasonable to hold states accountable.

Third, signaling is important. When the law clearly signaled the importance of caseload reduction, states responded. Clearer signaling is now needed in two areas: enhancing the economic well-being of working families and working with families with the greatest barriers to employment. The law could advance these goals in many ways without restricting creativity and flexibility—reflecting them in the purposes of TANF, requiring more explicit state plans, developing poverty reduction goals to guide state performance, requiring states to develop explicit benchmarks against which to measure performance.

In short, as Congress faces reauthorization of TANF, the discussion of state flexibility needs to be closely tied to the discussions of resources, accountability, and clear goals of promoting work and enhancing family economic well-being.