The 1996 federal welfare reform law joined two approaches to changing welfare policy in the United States. The law put in place many policies reflecting a conservative approach to the goals of work, independence, and marriage. These included time limits on assistance, stricter work requirements, and demands that teen mothers live with their parents and finish school. The law also strengthened requirements that clients cooperate with child support enforcement efforts and established stronger sanctions for noncompliance.
However, the law also created a block grant giving states flexibility in fashioning their own policy and administrative strategies to achieve the goals of the law. State innovation and experimentation are seen as critical ingredients of policy change. Federal time limits and work requirements apply only to cash assistance funded by the federal block grant, Temporary Assistance to Needy Families (TANF). States can, however, devise programs without time limits or work requirements when they use their own money, spending state funds under TANF's "maintenance of effort" provision, which requires states to spend 75 percent of the state dollars they spent in 1994. States can even use federal TANF funds to provide benefits to low-income families without time limits if those benefits help pay the costs of working, such as child care or transportation. States can impose stricter work requirements or shorter time limits. They can change many other eligibility requirements for cash assistance, including asset and earnings disregards. They determine the services to be offered to low-income families and define who is eligible for such services. And they have wide discretion over which providers—public or private, secular or religious—carry out their programs.
This combination of work-focused policy mandates and increased state discretion raises several questions. Have states used their flexibility—and, if so, how? Have they advanced the philosophy of the federal legislation, or have they introduced different elements? Have states "raced to the bottom," competing with one another to make their policies more punitive and less attractive to low-income families? Or have they developed diverse approaches to welfare reform, responding to different economic conditions and political climates? And what do state choices suggest about changes that Congress may want to consider in reauthorizing TANF?
State Policy Choices
Initial evidence on the differences and similarities in state policies comes from an examination of choices states made under TANF. Table 1 summarizes several decisions and the number of states selecting them. The left column includes policy choices that use positive incentives to encourage the work, marriage, and childbearing objectives of the 1996 federal law. These policies tend to enhance access to services and income supports. The right column contains work, marriage, and childbearing policies that use negative incentives-typically, restrictions on benefits or supports-to meet the same objectives.
The policy choices listed in table 1 reflect the broad support that states have given to the employment goals of the federal legislation. Most states accepted the premise in TANF that assistance should be temporary. Six states have reduced the limits on payment of cash assistance below the 60 months in the federal law while fourteen states have introduced "intermittent" time limits (e.g., available for only 36 out of 60 months). Only a few states have chosen to eliminate the 60-month time limit, but several states have announced that they will apply time limits only to adults, apply broad exemptions, or otherwise limit the effects of time limits (not shown).
By 2000, forty-three states had strengthened the federal work requirements by demanding that caregivers engage in a work activity before the TANF-imposed deadline of twenty-four months; thirty-eight required adults to do so immediately. States also endorsed the federal law's emphasis on "work first" over education and training: the number of states counting full-time post-secondary education as an allowable work activity for two years dropped to twenty-six in 2000. The number of states extending the coverage of the work requirements to parents of children less than one year old rose from six states in 1996 to twenty-two states in 2000.
Other changes reduced assistance to people who failed to meet the new obligations. TANF did not require states to cut off all benefits to a noncompliant household, but seventeen states now levy 100 percent sanctions for first-time violations, and twenty-one states impose 100 percent sanctions for at least three months as an ultimate sanction. Twenty-two states also reduce or eliminate Medicaid and/or food stamp benefits if sanctions are imposed for violations of TANF work requirements. Twenty-three states have adopted family caps, which means that children born or conceived while a family receives welfare are not counted in determining cash benefits. Finally, thirty-four states have ended or reduced the "pass through" to families on welfare of $50 per month of child support collections that was required under the old Aid to Families with Dependent Children (AFDC) program.
Two other highly restrictive options are not now in effect in any state. By 1998, thirteen states had reduced or delayed assistance to new residents coming from other states. Such laws, however, were struck down by the U.S. Supreme Court as unconstitutional in 1999, so they no longer apply. And no state adopted a policy favored by the most conservative advocates of federal welfare reform in 1996—barring all cash benefits to teen parents. Overall, however, states accepted and often strengthened the restrictions on assistance found in the federal law. Few of these policies were adopted by a majority of the states, but most states adopted at least one such policy.
At the same time, several policies increasing access to services or supports became quite widespread. Earnings disregards were liberalized in forty-seven states compared to AFDC standards. Families with earnings were allowed to keep more assistance than before, thereby increasing the incentive to work. Sixteen states strengthened positive incentives by creating a state Earned Income Tax Credit for low-income families with earnings and children. However, these credits can be used only to lower tax liability rather than being received as a cash income supplement in five states.
Nearly all states increased their asset disregards-limits on what a family could save or own and remain eligible for assistance. All states increased their vehicle disregards, primarily because the $1,500 ceiling for an automobile under AFDC was viewed as a barrier to employment for people who needed a reliable car for work. A large majority of states also made it easier for two-parent families to get cash assistance. For example, AFDC restricted eligibility for two-parent families by limiting how many hours the parents could work in a month, but forty states had eliminated those restrictions by 2000.
Taken together, the changes in disregards and the elimination of restrictions on two- parent families have expanded the range of working families eligible for cash assistance. One indicator of this shift is a substantial increase in "break even" points for families on assistance—the income recipients may earn before losing eligibility.
By contrast, states have not made major changes in their maximum cash assistance levels; i.e., the money families receive if they have no other income. Indeed, the most common pattern is a continuation of the pre-1996 pattern of real benefit levels being eroded by inflation. Between 1994 and 2000, twenty-nine states made no change in the nominal value of the benefit a family with no earnings receives; benefits in these states lost about 14 percent of their real value. Fifteen states increased their nominal maximum benefits, though only three of these states increased their real value; seven states cut their nominal benefits.
The increasing disregards and the declining real value of maximum benefits for families with no other income shifted the distribution of cash benefits away from persons without income and toward those with earnings. However, there is substantial diversity across states. In 2000, thirty-nine states offered some benefits to persons working 35 hours a week at the minimum wage, at least initially. But in eighteen states, a three-person family with one minimum wage worker who worked 35 hours a week would get no TANF benefits in their fourth month; in another ten states they would get less than $100. In short, only in a minority of states does TANF provide major wage supplements for workers working full-time even at minimum wage. Wage supplementation is even more modest for workers earning $8 per hour or more. Because maximum benefits remain low in most states, even generous disregards whittle down assistance to small sums as parents increase their earnings.
In short, states generally accepted the "sticks" elements in TANF, those provisions that punish noncompliance with work requirements. Many states used their discretion to stiffen work requirements or penalties for non-work over those in federal law. A majority of states adopted stiffer initial work requirements, and a large minority of states strengthened the federal sanction policies and cut the time limits. Yet the vast majority of states also adopted "carrot" policies-those aimed at rewarding work by enhancing access to certain benefits or supports, especially by eliminating provisions that discouraged work. As table 1 shows, however, most policy choices, especially those that restrict access to benefits, are in the two intermediate categories in terms of breadth of diffusion, having been adopted in more than ten but fewer than forty states.
States engaged in substantial innovation in both access-enhancing and access-restricting policies before 1998 and in some cases before 1996 (through AFDC waivers prior to passage of the federal legislation). A leveling off of innovation occurred thereafter. In some cases, the leveling off occurred because policies had been adopted by almost every state (eased auto asset limits, for example). In other cases, the apparent political limits of the policy had been reached (notably family caps). Instead of a "race to the bottom"—a continuing expansion of restrictive policies and little or no expansion in access-enhancing policies—many states adopted both types of policies. Yet states have differed dramatically both in their overall degrees of policy change and in their mix of "carrot" and "stick" policies. Rather than an emerging homogeneity, either around a "race to the bottom" or a consensus set of "best practices," there remains substantial heterogeneity in packages of state choices.
Factors Related to State Variations
Why do states differ in their policy choices? To explain variation across states in their policy choices under TANF, we used statistical techniques designed to find the relationship between characteristics of states and the policies they choose, while controlling for other attributes of those states. Table 2 summarizes our analyses of five policy choices: family caps, time limits shorter than those required under TANF, immediate work activity requirements, stronger sanctions than required under TANF, and the generosity of work supplements for working families. In the cells of table 2, a "+" means that the variable or factor is associated with a policy choice that is comparatively "liberal"; i.e., makes assistance more widely available or more generous. A "-" sign means that the variable is estimated to have a "conservative" impact on the dependent variable; i.e., constrains access to public assistance. Several points emerge from this analysis.
First, ideological factors are correlated with policies restricting cash assistance. Stronger sanction policies, shorter time limits, and immediate activity requirements are more common in conservative states than in liberal states (liberalism is measured here by the percent of the state's popular vote going to Bill Clinton in the 1996 presidential election, but other measures of state public opinion produce similar results).
Second, policies restricting cash assistance—such as shorter time limits, more severe sanctions, and family caps—are also more common among states that have a high percentage of African Americans on the caseload. Having a high percentage of Hispanics in a state's caseload is associated only with stiffer time limits.
Third, a state's resources under the TANF block grant are strongly related to policy choices regarding income supplements through earnings disregards and state earned income tax credits. Earned income disregards are more generous in states that were given relatively large grants per needy person—here, measured as the size of the TANF grant per child living in a low-income household. Because the formula for distributing federal TANF funds was based on state and federal spending in 1994, states that spent a lot on a per-case basis then got a comparatively large block grant under TANF and now have greater resources to spend on each of their families.
Fourth, policy decisions among the states were generally not statistically related to the severity of social problems in the states once other factors are controlled for. Out-of-wedlock birth rates, welfare dependency (measured by the percentage of the population on welfare at its highest point in the early 1990s), and unemployment showed weak marginal effects on state policy choices.
Policies that restrict assistance are thus most responsive to factors likely to affect a state's politics, particularly in the area of social policy, such as its electoral tendencies and the racial and ethnic composition of the caseload. Policies offering positive incentives to work, by contrast, are most strongly affected by a state's resources, especially the resources per needy family member provided through the TANF block grants.
Implications for TANF Reauthorization
Most states responded to the federal TANF program by endorsing both the employment goal and the means to achieve it, including time limits and work requirements. Perhaps the most surprising finding is the large expansion of eligibility for cash assistance among working families. States increased the rewards of work and lowered barriers to employment by increasing earnings and asset disregards, by eliminating anti-work regulations aimed at two-parent families, and by increasing their funding for child care and other services that directly support employment.
Expanded access to assistance for working families was in no way mandated by TANF. It emerged out of the new flexibility accorded to the states. Some aspects of TANF may have encouraged this tendency, including the block grant funding formula, the performance requirements, and the political popularity of the law's employment goals. The strong economy may also have been a factor. Although employment levels were not significant in accounting for differences among the states in their earnings disregards, it is still possible that the general prosperity of the late 1990s made policymakers willing to spread benefits to a wider range of working families.
However, there were important differences among state responses. States that were politically conservative and those that had large numbers of African Americans on their welfare rolls tended to adopt policies-such as stricter time limits, work requirements, and sanctions—that made assistance less attractive and less widely available. Contrary to the hopes of some welfare reform proponents, the new welfare law does not seem to have dissipated the image of the program as disproportionately aiding minorities, or the negative impact that this image has on support for the program in many states.
Several implications flow from this analysis. First, the absence of evidence that a "race to the bottom" in state policy choices is under way weakens the case for tightening federal limits on the range of state choice. But it should be noted that the good economic conditions that have existed until recently are those least likely to produce a "race to the bottom."
Another issue for reauthorization is whether differences in policy choices across states are problematic and, if so, what can be done about them. If one finds the divide between states that rely heavily on "sticks" and those that put greater emphasis on "carrots" to be troubling, it may be necessary to increase the funding levels per poor family in the states that had smaller relative grants in the first years of TANF. Giving more states the resources to pay for income supplements and child care might push TANF to become a stronger work support program in a larger number of states, not just in the traditional high-benefit states.
Encouraging work could also be addressed by maintaining or increasing the required work participation rates while revising the caseload reduction credit in calculating state performance levels. One possible revision might be to transform the caseload reduction credit into an employment credit for former TANF recipients who are in the work force.
Increasing hours of work required for individuals as well as state work participation rates is another option, with its own distinct challenges. Income supplements for families with full-time workers remain small or non-existent in most states-even when they earn only the minimum wage. Since work participation rates are based on the number of families on assistance who work the required hours, basing those rates only on full-time workers would make it difficult for most states to increase or even maintain their work participation rates.
It is still unclear, moreover, how states will respond to new challenges. What policies will they develop in dealing with timed-out families? How will they react when and if they have to meet 50 or 70 percent work participation rates year after year, especially if they face higher caseloads in a weakened economy? Evidence from the first five years of the TANF program suggests that there could be wide variation in state responses.