This report is part of the Urban Institute's Assessing the New Federalism project, a multi-year effort to monitor and assess the devolution of social programs from the federal to the state and local levels. Alan Weil is the project director. The project analyzes changes in income support, social services, and health programs and their effects. In collaboration with Child Trends, the project studies child and family well-being.
The project has received funding from The Annie E. Casey Foundation, the W.K. Kellogg Foundation, The Robert Wood Johnson Foundation, The Henry J. Kaiser Family Foundation, The Ford Foundation, The John D. and Catherine T. MacArthur Foundation, the Charles Stewart Mott Foundation, The David and Lucile Packard Foundation, The Commonwealth Fund, the Stuart Foundation, the Weingart Foundation, The McKnight Foundation, The Fund for New Jersey, and The Rockefeller Foundation. Additional funding is provided by the Joyce Foundation and The Lynde and Harry Bradley Foundation through a subcontract with the University of Wisconsin at Madison.
The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders.
Contents
Executive Summary
Background, Purpose, and Sample
Devolution: Continuation of a Trend
The 13-State Sample
Overview of the Report
Pre-TANF Welfare Policies and Caseloads
Three Categories of Cash Assistance Experimentation
Administrative Program Characteristics
Caseload Characteristics
Moving to the TANF Era: States' Early Responses
Who Should Receive Cash Assistance?
Who Should Work and How Much?
How Long Should Families Receive Cash Assistance?
What Part Should Cash Benefits Play in the System?
Implications of States' Early TANF Policies
The Changing Nature of Cash Assistance in Our Sample States
The Importance of Implementation
Notes
Tables
About the Authors
Executive Summary
In 1996, federal legislation replaced the 60-year-old Aid to Families with Dependent Children (AFDC) program with a block grant to states to finance a new program, Temporary Assistance for Needy Families (TANF). The law greatly extended the devolution of federal authority to the states that had been occurring. When the law passed, many welfare reform initiatives already were under way through federal waivers that allowed states to experiment with key features of the federal AFDC program. Because states varied considerably in their use of waivers, in mid-1996 substantial diversity existed among states' welfare programs. This diversity was expected to influence states' responses to TANF.
On the basis of case studies conducted in 13 states in late 1996 and early 1997, as well as secondary information about the states, we examine how this diversity influenced initial state actions regarding TANF. The 13-state sample, chosen to represent all regions of the country and a wide range of income and employment circumstances, encompassed over 60 percent of the nation's welfare population in 1996.1
We describe the approaches to cash assistance in place in the sample states on the eve of TANF's passage, focusing on how far states already had departed from federal AFDC rules. We then discuss early state responses to the new legislation, highlighting trends and patterns in policy emphases and choices made in light of states' historical approaches to welfare. Our analysis shows that, for the most part, the diversity among states' cash assistance programs continues to increase under TANF.
The Pre-TANF Picture
Three factors that influenced states' early TANF decisions provide some insight into the differences among the 13 sample states:
- The extent of pre-TANF waiver experimentation,
- States' administrative approaches to welfare, and
- Differences in states' current caseload characteristics.
State experiments undertaken through waivers changed many of the fundamental features of AFDC--features intended to affect important recipient behaviors including marriage, having children, going to work, and the length of time spent on welfare.
Waiver activity among our sample states focused on four major policy areas: eligibility, work, time limits, and payment policies (including family caps). Changes in eligibility rules were designed to be more neutral toward marriage by making it easier for two-parent families to move on and off AFDC as they lost and found jobs. Changes in work policies were focused on increasing incentives to work and increasing work participation requirements and penalties for not complying with the new requirements. A few states were experimenting with time limits that would terminate benefits after a certain period of time, although individuals could again receive benefits after a period of ineligibility.
Based on whether states were experimenting with the four policy categories before the passage of TANF, we classify the AFDC programs operating in our sample states into three categories:
Minimal Experimenters: Alabama, Colorado, and New York
These are states that had changes under way in only one of the four policy areas. The group includes states with very different approaches to welfare. Alabama operated its AFDC program under federal rules with low benefit levels. Colorado operated a relatively traditional AFDC program with moderate benefit levels and experimentation in some counties designed to increase work. New York operated a high-benefit, fairly traditional AFDC program with some experimentation focused on helping mothers with child support achieve independence from welfare.
Moderate Experimenters: California, Michigan, Minnesota, and Washington
States in this category were experimenting with two out of the four policies highlighted. These states were, coincidentally, distinguished by their relatively high benefit levels and tended to focus on experimentation designed to expand eligibility for two-parent families and to increase incentives to work. Michigan included more aggressive work requirement policies in its statewide experimentation than others in this category.
Extensive Experimenters: Florida, Massachusetts, Mississippi, New Jersey, Texas, and Wisconsin
States in this group were experimenting with a wider variety of changes to the federal AFDC rules. All expanded eligibility to more two-parent families, were experimenting with a variety of policies to increase recipients' work activities, and were experimenting with time-limited welfare benefits, family caps, or both.
State approaches to administration of cash assistance programs differed in ways that may also affect their responses to the opportunities presented by TANF.
Our sample includes seven states in which state employees administer welfare through local offices under state control (state-administered systems) and six states in which county employees are responsible for local administration (county-administered systems). In county-administered systems, local offices traditionally have had flexibility in designing employment services for welfare recipients. As states moved toward more work-focused cash assistance programs during the pre-TANF era, however, even state-administered systems were allowing local offices to play a larger role in designing procedures to shift clients into work activities. Most states reported facing the challenge of moving welfare offices away from an eligibility-compliance culture toward forging a direct link between welfare and work.
Two characteristics of state caseloads--family structure and the percentage of the caseload in work activities--highlight the substantial differences across state cash assistance programs in 1996.
"Child-only" AFDC cases--in which the children do not live with a parent or the parent they are living with is not eligible for benefits--make up vastly different proportions of states' caseloads. (Parents might not be eligible because they are not citizens, are receiving federal Supplemental Security Income benefits, or are sanctioned for not participating in required work activities.) Nationwide, one in five AFDC recipient units were child-only units, compared with more than one-third in states such as Alabama and Mississippi. The percentage of the caseload made up of two-parent families also varied substantially, with high proportions occurring in states combining high benefits with eligibility expansions for two-parent families.
The Early TANF Era: Trends and Responses
Although TANF generally increases state flexibility, in some ways the new legislation gives states less freedom in designing cash assistance programs than they had during the waiver era. The new law mandates important features of states' welfare programs and sets up financial incentives to further direct reform. To achieve the stated goals of reducing welfare dependency, preventing the incidence of out-of-wedlock pregnancies, and encouraging the formation and maintenance of two-parent families, the federal government mandated certain rules about eligibility, work, and time on welfare. States are constrained in using federal block grant monies and subject to financial penalties unless their programs conform in three areas designed to move states toward requiring work and reducing welfare dependence:
- States must achieve work participation targets of 90 percent for two-parent families by 1999 and 50 percent for all families by the year 2002. Work participation activities can include only a limited amount of education activities, and required participation for single-parent families must increase from 20 hours per week in 1997 to 30 hours per week by 2000. States must include all families with adults in their work participation rates unless the family has a child under age one. And all welfare recipients not exempt from work participation must be in a work activity by the end of two years to continue receiving assistance.
- States can only use federal monies to provide cash assistance for families that include an adult for up to five years in a lifetime, although states may exempt 20 percent of their average monthly caseload from the time limit.
- Unwed teen mothers must live with their parents or other responsible adults and must attend high school or other equivalent training to be eligible for cash assistance.
States can step beyond the federal program limits only by providing non-TANF state funds to support recipients who become ineligible for federal TANF support. Since the block grant allocations for states, based on their AFDC spending, are fixed for five years and requirements for states' continued financial contributions are set only at 80 percent of pre-TANF spending, states have the financial flexibility to provide non-TANF funds so long as caseloads remain low.
Balancing increased flexibility with new federal requirements, states made changes to many aspects of their cash assistance programs, including to the definition of who should receive cash assistance, who should work and when, how long cash assistance should be received, and what part benefits should play in the new system.
Who Should Receive Cash Assistance?
Continuing the pattern begun during the waiver era--a pattern designed to remove the marriage barriers and work disincentives inherent in the preferential treatment of single-parent families--nine of our sample states now provide full parity for two-parent families in their TANF programs, determining eligibility in the same ways as for single-parent families. Only Mississippi, a state in the extensive experimentation category, retained the AFDC eligibility policies; the remaining three states (California, Washington, and Massachusetts) retained some aspects of AFDC rules limiting eligibility for two-parent families to those with some recent work history.
One important change regarding family eligibility for TANF relates to "diversion assistance payments." These payments provide short-term cash assistance to applicants to prevent their formal entry into the welfare system. Eight of our sample states now offer short-term cash assistance. Diversion assistance payment policies cut across program models, relative benefit levels, and administrative models.
Who Should Work and How Much?
With the exception of Massachusetts (which retained its waiver policies regarding work), all sample states changed at least one important aspect of their work requirements. These actions indicated that the federal financial incentives to meet TANF's participation requirements were having their intended effect in motivating states to make their programs more employment focused. In general, states in our sample moved toward more generous earned income disregards (to encourage work); fewer exemptions (many more mothers with very young children will be required to participate in work activities); work requirement time triggers (setting a time limit after which a nonexempt recipient must be either employed, in a subsidized or unsubsidized job, or in community service); and tougher sanctions for those failing to comply with new work participation requirements (the toughest sanction states eliminate the entire family benefit after repeated failures to comply).
How Long Should Families Receive Cash Assistance?
Some states use intermediate or "tiered" time limits (usually in addition to a lifetime limit), some do not impose a lifetime limit, and some guarantee a limited amount of assistance beyond the federal limits. States also vary in how they define exemptions to the time limits, with some states permitting exemptions only for disability and others adopting more lenient policies that exempt individuals who cannot find employment or who face general hardship or "personal barriers."
What Part Should Cash Benefits Play in the System?
States were free to set their own benefit levels before TANF, but the new legislation gave them more flexibility over payment policies. The new law also offered states an incentive to cut benefits, since the block grant allocations were fixed over five years and states may use savings over what they spent on welfare for other purposes. Countering this incentive, however, was the "windfall" that most states received from the federal block grant. States have made few changes in their basic benefit structures in response to TANF, and only two states adopted new family cap policies that will no longer increase benefits if a child is conceived and born while the family is on welfare.
State policy choices since TANF are increasing diversity in cash assistance policies across the nation.
Some states are moving further toward policies that discourage the use of welfare, while others are maintaining large parts of the safety net that were available under AFDC. All are moving toward a stronger focus on work, but they emphasize different approaches to reach this goal. Even among states that already had been experimenting with broad changes to AFDC before TANF, welfare programs are undergoing a great deal of change. Somewhat fewer changes occurred in the states with extensive experimentation under way, but nearly all states in our sample changed two or more major policies.
With some policies there is more conformity among the sample states than with others. Nine out of 13 states now treat two-parent families who need assistance the same as single-parent families, mostly eliminating all special work rules for two-parent eligibility. While 10 states had been experimenting with two-parent eligibility before TANF, only three states had eliminated all special rules applying to two parents. Eight out of 13 states adopted formal diversion assistance programs (compared with only one state pre-TANF).
Early TANF decisions were clearly affected by the states' historic approach to benefit generosity.
With the exception of Michigan, states with relatively high benefits and minimal to moderate pre-TANF experimentation (New York, California, Minnesota, and Washington) tend to prefer approaches that encourage families to move off welfare through incentives rather than penalties. These states have high-intensity earned income disregards (where families can earn more than $1,000 per month before losing eligibility for cash welfare) and low-intensity sanction policies (under which families never lose their entire benefit as a result of a sanction). All four of these states adopted diversion programs, offering families cash incentives for staying off welfare.
As a group, states with the most extensive pre-TANF policy changes altered policies somewhat less than others. However, even these states tended to move to stronger work-first models of welfare. For example, Florida and New Jersey adopted even higher earned income disregards and fewer exemptions from work activities. New Jersey maintained policies consistent with its relatively strong safety net by holding to moderately intensive sanctions and allowing two six-month extensions to its five-year time limit.
The debate in three states in our sample--Colorado, California, and New York (all operating county-administered models of cash assistance)--was characterized at least in part by an historic desire to give counties a greater role in social welfare policy.
Colorado moved furthest in this direction by allowing counties to determine diversion policies and work requirements. While New York's counties were not given control over specific policies, their TANF plan provides performance awards based on job placements, and local districts can apply for waivers from state regulations.
Most states in our sample did not adopt the optional features of family caps and basic benefit reduction; only two states adopted new family caps under TANF and only one state reduced its basic benefit.
The family cap generated considerable debate in the U.S. Congress, and several prominent versions of the federal welfare reform bill proposed to mandate family caps for all states. (But many states did not believe that this policy was effective or necessary.) The family cap tends to separate the pre-TANF experimenters from the more traditional states (only one state in the pre-TANF minimal to moderate experimenters has a family cap, compared with five out of six extensive experimenters, including Wisconsin, which has implemented a new flat benefit policy that does not vary by family size).
The ultimate key to success in these efforts will lie in implementation and organizational reforms in the way services are actually delivered to clients.
In many states, new cash assistance policies represent a major departure from the past. Since most states already had more employment-focused policies under way before TANF, diversion assistance and time limits clearly present the newest implementation challenges. The strongest message we heard in our site visits to state and local welfare offices was that most states will adopt new procedures or complete transitions begun under waivers to ensure that new work-focused programs will be a success. But many respondents were worried that the TANF time clocks for work participation and benefit termination were ticking away while these reforms were still getting under way, raising the specter that--even in a good economy--some state changes focused on service delivery might come too late for those whose clocks are ticking.
Background, Purpose, and Sample
On August 22, 1996, the federal government replaced the 60-year-old Aid to Families with Dependent Children (AFDC) program with a block grant to states to finance a new program, Temporary Assistance for Needy Families (TANF). This federal legislation was remarkable in two respects. It eliminated the federal entitlement to cash assistance for such families, and it gave states unprecedented flexibility to design and operate cash assistance programs. Henceforth, states could only use federal TANF funds to provide assistance on a temporary basis (no more than five years in a recipient family's lifetime). Receipt of the full block grant allocation depended on documentation proving that increasing shares of a state's caseload participated in allowable work activities.2 Twenty-five percent of states' TANF families must participate in work activities for at least 20 hours a week in 1997, with the level of participation to increase to 50 percent for at least 30 hours a week by 2002. Participation rates specific to two-parent families--a small share of the total cash assistance caseload--are 75 percent and 90 percent. Within these broad requirements, however, states have considerably more authority to design cash assistance for families with children.
Devolution: Continuation of a Trend
In fact, this devolution of federal authority was, in many ways, an extension of business as usual. States have always had the authority to determine their cash assistance benefit and eligibility levels, and, before TANF, benefits were paid for out of both state and federal revenues according to a federally prescribed, open-ended matching formula. More important, explicit welfare reform initiatives were under way in many states long before passage of the federal legislation, through federal waivers of certain rules that allowed states to experiment with changes to their AFDC programs.
Some states used these waivers to implement comprehensive statewide reform. Others used them to experiment with changes to specific AFDC provisions or implement reforms on a pilot basis in selected counties. Thus, there was considerable diversity in state welfare programs in mid-1996, which influenced states' TANF actions and the scale of changes they had to make to comply with federal TANF rules.
This report explores how this diversity influenced initial state responses to TANF, on the basis of case studies and secondary information about 13 states. Together these states comprise more than 60 percent of the nation's welfare population. Site visits were conducted in late 1996 and early 1997.3 Our case studies included interviews with state political and policy leaders, heads of state departments responsible for cash assistance programs, and local offices responsible for delivering cash assistance benefits to welfare clients.4 The intent was to learn about their social welfare policies just before the new legislation: how they made decisions about welfare policy, what goals they set for their programs, and how these goals were translated into policies and procedures. At the time of our visits, some states were still debating how to respond to the new federal legislation. Other states had already made final decisions about their new policies. The implementation status of these policies varied across the states--some policies had already been implemented for quite some time, others were just beginning to be put into practice, and others had yet to be implemented. Documents from the 13 states--legislation, new program regulations, and caseworker handbooks--supplemented information obtained through our extensive interviews.5
These states' social welfare policies encompass a wide range of social supports for low-income families, including child care, transportation, and health coverage. These supports clearly play a large role in helping low-income families achieve independence from cash assistance, and ultimately in the success of welfare reform. However, given its importance to the safety net and the potential for major changes resulting from federal welfare reform, this report focuses on the more narrow topic of cash assistance for low-income families as a first step in understanding the changing social safety net in the states.6
The 13-State Sample
The 13-state sample represents all regions of the country and a wide range of income levels (figure 1). Washington and California in the West, Wisconsin and Michigan in the Midwest, and Florida in the South were within 5 percent of the national average in per capita income in 1996. Colorado in the central part of the country, Minnesota in the Midwest, and New York, New Jersey, and Massachusetts in the Northeast were 5 percent or more above the national average. Texas, Mississippi, and Alabama, all in the South, were 5 percent or more below the national average. Their unemployment rates were similarly varied (figure 2). Colorado, Massachusetts, Minnesota, and Wisconsin enjoyed unemployment rates at least a full point below the national average (5.4 percent in 1996). Rates in California and Washington, in contrast, were at least a full point above that average.
AFDC program size also varied across our sample states (table 1). Obviously, the more populous states had the bigger programs, but even when expressed in per capita terms, the variation was considerable. Compared with a national average of about 6 percent of nonelderly families receiving AFDC in 1996, for example, the share was only about 3 percent for Alabama and Colorado versus almost 10 percent for California. The proportion of all poor families with children receiving AFDC in 1996 varied from about a quarter or lower in Alabama, Mississippi, and Texas to about a half or higher in California, Massachusetts, Michigan, New Jersey, New York, and Washington.
Program size is clearly related to program generosity. Other things being equal, the higher the benefit level, the larger the caseload, because more families qualify for assistance. Among our sample states, Alabama, Mississippi, and Texas had the lowest AFDC benefit levels in 1996; California and New York had the highest.
Total AFDC expenditures varied by program size and generosity. The proportion paid for by nonfederal sources, however, also varied because the federal matching formula used to finance the AFDC program provided more generous matching funds for poorer states. For the country as a whole, the federal government paid 46 percent of the $23.5 billion spent on AFDC in 1996. Mississippi, with the least expensive AFDC program at $68 million in 1996, paid only 22 percent of the total. Alabama, near the low end with a $94 million program, paid 34 percent of the total. California and New York, at the high end with program costs totaling $6.2 billion and $3.9 billion respectively, each paid half of the total.
In no state did AFDC absorb a large share of the state general fund.7 In Alabama, Mississippi, and Texas--all states with low benefit levels and the lowest percentages of poor families with children receiving cash assistance--AFDC program costs accounted for only about 1 percent of the state general fund. At the other extreme, California, with the highest benefit level and the second highest share of poor families with children receiving cash assistance, spent more than 6 percent of its general fund on AFDC.
Overview of the Report
The following section of this report describes the major characteristics of cash assistance programs in place in the sample states on the eve of TANF's passage, focusing on how far states already had departed from traditional federal AFDC rules. On the basis of these descriptions, we group states into three categories: minimal experimentation, moderate experimentation, and extensive experimentation. We also describe other program and caseload characteristics likely to affect states' TANF choices. The next section discusses early state responses to the new federal welfare reform legislation, highlighting policy emphases and choices made in light of the state's historical approach to welfare. The final section summarizes the variations in policy emphases across these states and discusses the implications in the context of their pre-TANF program structures.
Pre-TANF Welfare Policies and Caseloads
President Clinton's famous 1992 campaign promise to "end welfare as we know it" resonated with states already frustrated by caseload increases during the early 1990s. The increases roughly paralleled the recession-induced rise in unemployment (figure 3) but lasted longer and were more sustained than caseload increases during previous recessions.8 State frustration was somewhat alleviated by a federal provision (Section 1115 under the Social Security Act) that allowed states to apply for waivers of federal AFDC policies to pursue alternative strategies. The federal government responded to the states' growing interest in reforming their welfare programs by increasing waiver approvals between 1993 and 1996. By the time federal welfare reform was passed by Congress and signed by the president, waivers had been approved for 43 states.9 During the period of waiver expansion, both unemployment and caseloads were declining.
State experiments undertaken through these waivers changed many of the fundamental features of AFDC--features intended to affect important recipient behaviors, including marriage, childbirth, work, and time spent on welfare. Waiver activity among our sample states in 1996 was considerable in four major areas (table 2):
Eligibility. AFDC limited two-parent eligibility to families whose principal wage earner worked fewer than 100 hours per month, was unemployed for at least 30 days, and could demonstrate substantial previous work experience. It was widely believed that these rules created disincentives for two-parent welfare recipients to move back into the workforce quickly and contributed to the formation of single-parent families. AFDC also allowed teen mothers to apply for assistance as a separate unit but permitted states to require them to live with their parents to gain eligibility. Many believed that allowing unmarried teen mothers to form assistance units encouraged out-of-wedlock pregnancy.
Prior to TANF, 10 states in our sample had waived one or more rules to make it easier for two-parent families to move on and off AFDC as they lost and found jobs. In most states the new eligibility rules applied statewide. One state, Texas, had begun implementing a diversion program that would provide short-term cash assistance to help families avoid entry into welfare. Only five states in our sample--Massachusetts, Minnesota, Mississippi, New York, and Wisconsin--had adopted the optional federal rules requiring unmarried teen mothers to live with their parents or other responsible adults to be eligible for cash assistance.10
Work. The AFDC program specified rules for the treatment of earnings while on welfare, and states were required to operate Job Opportunities and Basic Skills (JOBS) programs offering a range of education, training, and work activities to recipients. However, a large share of the adult caseload was exempt from participation; participants were required to work only 20 hours a week; and states only had to reach a 20 percent participation rate for their nonexempt caseload.11 The general consensus among many observers of the federal JOBS program was that it was not as effective as hoped in moving recipients into work.12
In the pre-TANF period, nearly all states in our sample adopted some policy to increase work among welfare recipients. Most states expanded earned income disregards to encourage employment. Only 4 out of 13 sample states--Alabama, New Jersey, Texas, and Washington--still retained the AFDC earned income disregard rules statewide in 1996. Most states in our sample obtained waivers or made major changes in their JOBS programs to implement tougher requirements designed to increase participation in activities leading to employment. Eleven states had increased the percentage of the caseload that was required to participate in work and work preparation activities. Many of these states also increased the sanctions imposed on people who failed to comply with new requirements.
Dependency. Under AFDC, families could receive assistance as long as they had a dependent child under age 18 and met their state's income eligibility criteria. Many critics of the welfare system argued that the entitlement nature of AFDC led to long-term dependency.1313 However, lifetime time limits were largely absent from states' waiver policies in part because the federal government would approve time limits only if certain safeguards were included. Some states, including Massachusetts and Wisconsin, were prepared to go forward with lifetime limits under certain circumstances before TANF but were prevented from doing so by the federal government. Some states secured approval to experiment with temporary benefit termination. At the time TANF passed, most states still retained AFDC's guaranteed assistance to families with children.
Prior to TANF, only one state in our sample--Texas--had adopted a statewide "benefit termination" time limit before the 1996 legislation, which was being implemented gradually across the state beginning in April 1996. The Texas time limit was "tiered" so that recipients could receive benefits for 12, 24, or 36 months depending on their readiness for work, and the time limit resulted only in a loss of eligibility for the adult for five years (children continued to receive benefits). Washington adopted a unique time limit that would reduce a family's monthly benefit by 10 percent a year after adults had received welfare for four out of five years. Florida and Wisconsin had time limits that were not statewide, and both of these policies allowed the clock to be "reset" after a period of ineligibility. Waiver approval for time limits, in all cases, required states to continue aid or allow participation in a work program in cases where the adult complied with program rules but was unable to find employment despite "best efforts."1414
Payments. States have always set basic benefit levels for families, and benefit levels have traditionally increased with family size. Some argued that this benefit policy encouraged welfare recipients to have additional children.15
Prior to the passage of TANF, four states in our sample--Massachusetts, Mississippi, New Jersey, and Wisconsin--adopted "family caps" that either reduced or eliminated any increase in assistance for families when another child was conceived and born while the family was receiving welfare.16 New Jersey adopted a family cap first, with a waiver going statewide in 1994, that eliminated any increase for children so born.
Thus, the most common waivers for states in our sample focused on increasing equity for two-parent families and increasing earned income disregards to encourage work. While these were expansionist policies (both increased the number of families eligible for welfare), they were designed to be more neutral toward marriage and to encourage work. States with more generous earnings disregards also increased the proportion of the caseload that was required to participate in work-related activities relative to the federal JOBS requirement and placed more emphasis on activities designed to lead immediately to employment (such as job search activities rather than education or training). Some states also increased the penalties for nonexempt recipients who failed to participate in work activities. Other states chose to impose time limits during the waiver era, although these time limits did not result in a lifetime limit on benefits.
Three Categories of Cash Assistance Experimentation
Based on whether states were experimenting with the four policy categories shown in table 2 (eligibility, work, time limits, and payments) before the passage of TANF, we classify the AFDC programs operating in our sample states into three categories. Minimal experimentation refers to states that were operating with or without waivers in one of our four selected policy areas. Moderate experimentation refers to programs that had changed two policy areas for at least some portion of the states' welfare population. Extensive experimentation refers to programs experimenting with changes in three or more policy areas. Table 3 shows the states in our sample grouped into these three categories, with policy changes and dates approved (to show how long the waiver policy had been in effect), maximum benefit levels, and relative caseload size. Table 3 also shows whether the state received waiver approval to implement the new policy statewide or limited experimentation to some counties.1417
Minimal Experimenters: Alabama, Colorado, and New York
This group includes states with very different approaches to welfare. Alabama's AFDC program operated under federal rules with such low benefit levels in 1996 that the federal Food Stamp program was the main income support program for its low-income families with children. Alabama did strictly enforce sanctions for noncompliance in JOBS, but the sanctions were thought not to stimulate compliance for those required to participate because they eliminated only the adult portion of the benefit, which was partially offset by increased food stamp benefits. (Families lost $30 per month in AFDC benefits when they were sanctioned, and food stamps made up one-third of this loss.)
Colorado, a fairly rich state, also had a small, relatively traditional AFDC program with a medium benefit level. Only 3 percent of nonelderly families received assistance in Colorado, which matched Alabama as the smallest caseload in our state sample. Like Alabama, welfare historically has not been a prominent policy issue in Colorado, although the state maintains a strong interest in educating and training its low-skilled population--welfare recipients and working poor alike. Colorado did have a waiver to increase work among its caseload, and the state experimented with a combination of earnings incentives, tougher work requirements, and sanctions in five counties. However, these counties still emphasized skill development work activities over immediate employment. These policies reflected Colorado's goal of creating incentives for recipients to leave and stay off of welfare, rather than move recipients off quickly into low-paying jobs with the high probability that they would return. Colorado also reported that it fostered an environment in which counties were encouraged to tailor programs to fit their communities while remaining within the federal AFDC and JOBS rules.
New York contrasts sharply with the other two states in the minimal experimentation category with its relatively high benefit level and large caseload. New York had minimal reform under way before TANF. The state had implemented administrative changes to strengthen requirements for work and work-related activities, but it had no active waivers from federal AFDC policy with the exception of the Child Assistance Program (CAP), a voluntary program for AFDC mothers with child support orders. CAP, implemented in 14 counties, was designed to help families with child support court orders achieve independence. The program encouraged work through higher earned income disregards and increased support services in exchange for lower AFDC benefits.
Moderate Experimenters: California, Michigan, Minnesota, and Washington
States in this category were experimenting with two out of the four policies highlighted. As shown in table 3, all states in this category paid relatively high benefits, and, with the exception of Minnesota (the one state in this group with higher-than-average median income in 1996), all had relatively high proportions of nonelderly families receiving cash assistance. All had waivers to experiment with more generous eligibility for two-parent families, and, with the exception of Washington, these states had more generous earned income disregard policies than required under AFDC. None had waivers introducing family caps. Only Washington had introduced a time limit policy, and it was relatively lenient.18
While Michigan's approach to welfare has many features in common with the other states in this category, it also stands out as the state with the most aggressive approach to moving recipients into work. As shown in table 2, Michigan is the only state in this category that adopted stricter sanctions along with greater incentives to increase work activities among its caseload. Michigan's policy that mandates work activities for all but the disabled and mothers caring for a child who is disabled or under three months of age is unique among this group of moderate experimenters.
Extensive Experimenters: Florida, Massachusetts, Mississippi, New Jersey, Texas, and Wisconsin
States in this category were experimenting with a wider variety of changes to the federal AFDC rules. All states in this group expanded eligibility to more two-parent families, were experimenting with a variety of policies to increase recipients' work activities, and were experimenting either with time-limited welfare benefits, imposing family caps, or both. States in this category are mixed in terms of their relative benefit levels, and they all had a relatively moderate caseload size in 1996 (from 4 to 6 percent of nonelderly families). Some limited their extensive experimentation to select sites while others experimented on a statewide basis.
Mississippi and Florida moved toward reform by experimenting with new policies in three out of the four categories shown, but most of the new policies were adopted only in some parts of the states. Mississippi shifted from its traditional AFDC program by experimenting with a work-first demonstration in six counties in 1995 that emphasized employment and immediate placement in jobs over education and training. This demonstration also incorporated subsidized job placements and stricter work participation requirements with tougher sanctions for noncompliance, which included reducing or eliminating food stamp benefits. At the same time, Mississippi adopted a statewide family cap. The most notable component of Florida's waiver was one of the nation's first benefit time limits, also implemented only in parts of the state. Florida was poised for broader, statewide reform just before the federal legislation passed. Florida had passed its Work and Gain Economic Self-Sufficiency (WAGES) Act, legislation that shared many features found in Florida's waiver, in June 1996.19
Wisconsin's reforms designed to move welfare recipients into the workforce were widely known and influenced the federal welfare reform debate. Wisconsin's policy was strongly focused on work, including a new sanctions policy, implemented statewide in March 1996, that included part of the food stamp benefit in the penalty. New Jersey was also an early experimenter, with reforms in three out of four of the policy categories that were implemented statewide between 1992 and 1994. New Jersey was the first state to experiment with a family cap, and it drew considerable national attention as a result. New Jersey's waiver policies regarding work, however, contrasted sharply with those in Wisconsin. New Jersey required more families to participate in work activities than the federal JOBS program did (and those who did not participate faced tougher sanctions), but those work activities focused on education and job training aimed at helping welfare recipients achieve independence rather than immediate employment.
Massachusetts and Texas adopted extensive new policies before federal reform had passed, but they only began implementing these reforms in 1996, just before TANF. When Texas made its change, it was a dramatic one--going directly from a very traditional AFDC program to a strong work-first policy with a time limit, a move that was accompanied by a new statewide focus on workforce development. Massachusetts moved toward a strong work policy but applied it to a small percentage of its caseload (because of a broad exemptions policy), and it adopted a family cap.20
Administrative Program Characteristics
States vary in their approaches to administering cash assistance programs in ways that may also affect their response to the opportunities presented by TANF. Some states take full responsibility for administering programs, while others supervise county administration of welfare programs. Most states with county-administered programs require counties to share the cost of assistance programs. States that have historically given counties more flexibility in the delivery of welfare services may be better poised to devolve more responsibility to counties under TANF.
Our sample includes state-administered models and county-administered models of welfare service delivery (table 4). In state-administered models (seven states), state employees administer welfare through local offices. In county models (six states), county employees under state supervision are responsible for local administration.21 Both types of models extended little flexibility in determining welfare eligibility and imposing participation requirements, in large part because of the strong role of federal regulations in the pre-TANF period. In county-administered models, states have traditionally given local offices greater flexibility in designing employment services. In contrast, in nearly all of the state-administered models, local offices reported that they had little or no flexibility in designing employment services for welfare recipients.
The majority of states in the minimal and moderate experimentation categories have county-administered welfare systems. In contrast, states in the extensive experimentation category tend to have state-administered systems. This difference may indicate somewhat more difficulty in reaching a consensus on bold statewide reforms when counties play a strong role in the state's welfare system. Of course, other factors (such as the governor's strength and will to achieve reform) also play an important role in consensus building.
With the exception of Wisconsin, all of the states with county-administered systems require their counties to pay a share of the benefit costs and local administrative costs (table 4). The required share of nonfederal benefit costs varies from a high of 50 percent in New York to a low of 5 percent in California. The required share of local administrative costs ranges from a high of 100 percent in New Jersey and Minnesota to a low of 30 percent in California. County funding responsibilities do not necessarily imply county flexibility, however. Local offices in California and New Jersey reported limited flexibility in designing their welfare programs, while others reported great freedom to do so. Counties in Colorado, for example, were permitted to experiment with welfare reforms before TANF (Denver, for example, experimented with JOBS-like case management techniques in its AFDC program).
At the time of our case study visits, most states, regardless of whether they employed a state- or county-administered model for their cash assistance program, reported that efforts were under way to strengthen the link between welfare and employment. And most states reported that forging a direct link between welfare and work within an eligibility-compliance culture was a challenge. The traditional AFDC culture, in most of our sample states, as well as elsewhere, has focused on determining client eligibility for cash assistance, ascertaining whether the client was required to participate in work activities, informing the client about other services available in the county, and making sure that payments were correctly calculated and delivered on time. Communication between the welfare eligibility system and employment services typically was conducted in writing, and there was no perceived obligation on the part of the eligibility caseworker to encourage or link a client with work.
States that had well-developed work-first programs before TANF had further strengthened the link between cash assistance and work. In this process some states with state-administered welfare systems (including Florida, Washington, and Texas) gave local areas a larger role in designing and developing service delivery systems. Florida, for example, was establishing community-based private/public boards to oversee work and support services. Overall, these changes have introduced more flexibility at the local level, although welfare is still operated under a state-administered system.
Washington and Texas provide two more examples of state-administered welfare systems that were moving toward more local flexibility before TANF. Washington state retained control over policy development, technical assistance, program monitoring, and evaluation. However, it divided its program operations and administration into six regions in 1996 to ensure that community needs were identified and community resources were used in operating the AFDC and JOBS programs. In 1995 Texas shifted responsibility for job training and employment for welfare clients from the state welfare offices to 28 regional workforce development boards.
Caseload Characteristics
Other things being equal, a state's program model will influence the characteristics of its caseload. Although it is not possible to sort out program influences on caseload from those of other important factors (such as employment conditions and education levels), the basic characteristics of a state's caseload in 1996 can still help us understand state responses to TANF. Two characteristics--family structure and percentage of the caseload in work activities--are particularly important because they indicate how states' caseloads compare with the new federal requirements that states meet work participation targets for families. These work participation targets are higher for two-parent families than for single-parent families, and units without adults are not included in the targets.
Family Structure
For the country as a whole, only 8 percent of the AFDC caseload consisted of two-parent families (table 5). As is to be expected, California, Michigan, and Washington--all states that extended eligibility to two-parent families statewide--had higher-than-average proportions of two-parent families in the AFDC caseload. Although Massachusetts and Texas had also implemented waivers extending eligibility to two-parent families statewide, both had lower-than-average shares of two-parent families in their caseloads. It is unclear whether these states had low two-parent participation despite their waivers because the rule was relatively new (and possibly not well known), because the benefit was too low to attract eligible two-parent families (in Texas), or because the AFDC work history rule applying to two-parent families remained in effect (in Massachusetts).
Child-only cases consist of cash assistance units in which children either are not living with a parent or are living with a parent who was ineligible for benefits. Child-only cases without parents do not count as part of the caseload for the purpose of calculating states' TANF work participation targets. Thus, states with a large percentage of child-only cases without parents have fewer adults to move into work activities. On the other hand, these states' caseloads may not decline much because child-only cases will not be subject to new policies designed to move parents into work. Nationwide, 21 percent of AFDC recipient cases in 1996 were child-only cases. Among our sample states, more than a third of the caseloads in Alabama and Mississippi consisted of child-only cases. Texas and Wisconsin also had higher-than-average shares of child-only cases.
The composition of child-only cases also varied widely by state. Nationally, nearly 40 percent were eligible children who did not live with their parents; among our sample states, this percentage was higher in Colorado, Florida, Minnesota, New Jersey, and Washington. Nationally, 47 percent of child-only cases involved children living with parents who were not AFDC-eligible--21 percent involved parents on Supplemental Security Income (SSI), 16 percent involved parents who were not citizens, and 10 percent involved parents who had been sanctioned. Among our sample states, Alabama, Colorado, Massachusetts, Michigan, Minnesota, Mississippi, and Wisconsin had a higher-than-average percentage of parents on SSI. California, New York, and Texas, consistent with their relatively large immigrant populations, had a much higher than average percentage of parents who were not citizens. Alabama, New Jersey, Texas, and Wisconsin had a higher-than-average percentage of parents sanctioned. Alabama, the only state in our sample without an active waiver from federal AFDC rules, had strong JOBS sanction enforcement policies during the pre-TANF era, as noted earlier. The other three states, all in the extensive experimentation category, had implemented tough sanctions to enforce participation in their new work-first programs.
Work Activities
The new federal rules require states to show that 25 percent of TANF families engaged in work activities in 1997, to increase to 50-percent in 2002, and increasing participation in job search and employment was a major focus of state waiver activity. About one-third of all adult single females receiving AFDC already participated in work-related activities in an average month during 1996 (table 6).22 Among our sample states, Michigan, Minnesota, and Wisconsin--all pursuing a variety of policies to increase work among their caseloads--had more than 40 percent of their caseloads participating in some work activity. As noted earlier, Michigan had implemented multiple policies to increase work activities among its caseload, which applied to a large share of the caseload. In 1994, Minnesota implemented an incentive-focused waiver in eight counties that allowed participation in welfare until a household's earnings reached 140 percent of the federal poverty level. The waiver also required all recipients to complete an employability plan after two years on welfare.
Nationally, 10 percent of recipients were combining welfare with employment. Among our sample states, Michigan and Wisconsin--states with bold, employment-focused waivers under way for at least two years--had more than 20 percent of their caseloads doing so. California--a moderate experimentation state that had significantly increased earned income disregards and reduced the proportion of the caseload exempt from work--was not far behind, with about 16 percent of its caseload working in unsubsidized jobs.
Only New Jersey fell considerably behind other sample states in achieving work activities for its adult female caseload. New Jersey's waiver policy focused on education and training and addressing the needs of the entire family to help families achieve independence; respondents in our case studies, however, reported that caseworkers lacked the time required to implement the ambitious policy.
Moving to the TANF Era: States' Early Responses
The period of waiver experimentation ended with the passage of the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA). PRWORA created the TANF program, stating the following purposes:
[To] increase the flexibility of States in operating a program designed to: 1) provide assistance to needy families so that children may be cared for in their own homes or in the homes of relatives; 2) end the dependence of needy parents on government benefits by promoting job preparation, work, and marriage; 3) prevent and reduce the incidence of out-of-wedlock pregnancies . . . ; and 4) encourage the formation and maintenance of two-parent families.
Along with increased state flexibility, the statute was equally clear in its intent to remove cash assistance to low-income families with children as a federal entitlement:
No Individual Entitlement--this part shall not be interpreted to entitle any individual or family to assistance under any State program funded under this part.23
To achieve these goals, the new legislation also mandated certain rules about eligibility, work, and time on welfare. In particular, states are constrained in how they use federal block grant monies and are subject to financial penalties unless their programs conform in three key areas designed to move states toward requiring work and reducing dependence:
- States must achieve work participation targets of 90 percent for two-parent families by 1999 and 50 percent for all families by the year 2002. Work participation can include only a limited amount of educational activity, and required participation for a single-parent family must increase from 20 hours per week in 1997 to 30 hours per week by 2000. (Two-parent families must participate in work activities for at least 35 hours per week beginning in 1997.) States must include all families with adults in their work participation rates unless the family has a child under age one. All welfare recipients not exempt from work participation must be in a work activity by the end of two years to continue receiving assistance.
- States can only use federal monies to provide cash assistance for families that include an adult for up to five years in a lifetime, although states may exempt 20 percent of their average monthly caseload from the time limit.
- Unwed teen mothers must live with their parents or other responsible adults and attend high school or other equivalent training to be eligible for cash assistance.
If states wish to assist individuals not eligible for federal TANF assistance, they are permitted to do so, as long as they use their own monies.
States were required to bring all aspects of their cash assistance programs into compliance with the new federal law. The one exception was that a state could retain its waiver policy until the scheduled expiration date and would not be required to comply with the aspects of the federal law that are inconsistent with the waiver.
The federal financing of the new system played an integral role in states' responses. The federal legislation established a block grant, fixed over five years, to finance TANF benefits and employment services for welfare recipients. The allocations to states were based on AFDC spending (including JOBS and Emergency Assistance) over the 1994Ð96 period. These allocations were expected to exceed costs in most states, at least in 1997 if not beyond, because caseloads and spending were declining. States that wished to be generous could use the extra federal money to provide relatively expensive services required to move more recipients into jobs. The federal law also included a state maintenance-of-effort (MOE) requirement. States must spend at least 80 percent compared with what they spent in the baseline year (or 75 percent if they met the federal work participation requirements).
The 1997 federal block grant allocations for all the sample states except Colorado exceeded federal 1996 spending for AFDC (table 7). What is interesting is that many states did not receive their entire block grant allocation during 1997. This is because states were not required to submit a TANF plan until July 1, 1997, and could operate temporarily under AFDC federal financing rules. The states that received their entire block grant allocation--Florida, Massachusetts, Michigan, Mississippi, and Wisconsin--submitted TANF plans early and, with the exception of Florida, received the largest windfalls (more than 123 percent of 1996 funding). These states were experimenting with welfare reform before TANF (all but Michigan fall in the extensive experimentation category) and were ready to respond quickly to the new legislation. Some states submitted TANF plans conforming to new federal guidelines even before they made final decisions about how to structure their programs in order to begin block grant funding. For example, California submitted its initial plan November 26, 1996, but only completed its legislative debate concerning TANF in late summer of the following year. In contrast, Colorado and Minnesota submitted their TANF plans at the latest possible date. While Colorado expected a slight cut in federal monies (unless its caseload dropped during 1997), Minnesota was poised for a significant bonus (111 percent).
How much change TANF required at the state level, of course, depended on how much each state had already changed its traditional AFDC practices. The states that were experimenting with policies similar to the new federal law needed fewer policy changes to come into compliance with TANF. Most states in the extensive experimentation category either did not reopen the
welfare debate at all (e.g., Massachusetts and Texas, states with very recent, extensive waivers) or simply waited until the federal legislation passed before adopting new policies already debated and approved by the state legislature (e.g., Florida and Wisconsin). Most states that were minimal or moderate experimenters, in sharp contrast, entered long debates over what new reform policies they should adopt.
State responses were also influenced by their underlying philosophies about cash assistance, which were themselves related to the types of program models in place in 1996. For example, states with high benefit levels that limited waiver experimentation to broadening eligibility for two-parent families and increasing incentives to combine work and welfare stressed public responsibility for a strong financial safety net for families. Other states, mostly in the extensive experimentation category, stressed changes that increased personal responsibility, increased employment, and reduced welfare dependency. For example, four out of the six states in this category already had adopted tough sanctions to enforce participation in work activities and had adopted family caps hoping to reduce out-of-wedlock births. Historical attitudes to state-local sharing of policy authority (reflected in their administrative models) were another type of influence, which cut across categories.
In the remainder of this section, we discuss states' initial responses to TANF along four major dimensions:
- Who should receive cash assistance?
- Who should work and how much?
- How long should families receive cash assistance?
- What part should cash benefits play in the system?
Who Should Receive Cash Assistance?
TANF removed any federal presumption about which types of families with children should be eligible for cash assistance. Continuing the pattern begun during the waiver era--a pattern designed to remove the marriage barriers and work disincentives inherent in the preferential treatment of single-parent families--nine of our sample states now provide full parity for two-parent families in their TANF programs, determining eligibility in the same way as for single-parent families (table 8). Only Mississippi, a state in the extensive experimentation category, retained the AFDC eligibility policies. California, Washington, and Massachusetts retained some aspects of AFDC rules that limit eligibility for cash assistance to those two-parent families with some recent work history. With the exception of Mississippi, all states eliminated the AFDC rule that limited the number of hours an adult recipient in a two-parent family could work to 100 per month, presumably because this rule acted as a work disincentive for two-parent families, a group requiring a high work participation rate. California and Washington, however, still apply the 100-hour rule to applicants.
One important change regarding family eligibility for TANF relates to "diversion assistance payments." These payments are designed to provide short-term cash assistance to applicants to prevent their formal entry into the welfare system.24 Some states added diversion assistance payments to their policy arsenal before TANF (among our sample states, only Texas had waiver authority to implement this policy before TANF, starting in 1997). Many states are now using these payments in an effort to divert families away from welfare while still qualifying them for employment services, food stamps, child care, and Medicaid.25 Eight states in our sample now offer short-term cash assistance payments. While states that adopted diversion assistance payments fell into all experimentation categories, three out of four moderate experimenters adopted this policy. Diversion assistance policies in Colorado and California give counties more flexibility in setting welfare policy. But other states have adopted diversion as a statewide policy goal. Florida's WAGES program, for example, made reducing welfare rolls an explicit policy goal, and diversion assistance an explicit means to achieve it.26 Wisconsin's diversion assistance policy differs from the others; the lump-sum payment is a loan to assist with employment-related expenses and must be repaid to the state within 12 months.
Who Should Work and How Much?
Although states now have complete freedom in setting work requirements for their welfare caseload, the federal legislation gives them strong financial incentives to aggressively pursue strategies that move recipients into employment. At the time of our site visits, most states were either beginning to move toward work-first policies or, as in Michigan and Wisconsin, already had well-developed work-first policies. As mentioned earlier, work-first policies, under the philosophy that any job is a good job, emphasize work over education and training as a strategy for moving welfare recipients toward independence.27 These policies contrast sharply with the federal JOBS program, which required states to offer education and job training activities. States could move toward a work-first policy without waivers under AFDC by emphasizing employment activities over education. However, states that moved furthest in this direction sought and received approval for waivers so they could combine an emphasis on employment with the stronger sanctions and more limited work exemption policies and earnings disregards described earlier.
Table 9 shows four indicators of the intensity of state work policies: the generosity of the earned income disregard; the age of youngest child that exempts mothers from participating in work activities; formal, time-triggered work requirements; and the level of benefit sanction for failure to comply with participation requirements. Table 9 also indicates in bold type which policies represent TANF changes. A quick glance shows that, with the exception of Massachusetts, all sample states reported changing at least one aspect of their work requirements, clearly indicating that the federal financial incentives to meet TANF's participation requirements were having their intended effect in motivating states to make their welfare-to-work programs more employment focused.
Earnings Disregards
Nine states adopted new earned income disregards, with eight of them moving toward more generous disregards, continuing the trend started under
waiver policies. Only two states retained the AFDC rules (Colorado and Texas). Colorado, a state in the minimal experimentation category, operated a fairly traditional AFDC program before TANF, and its JOBS program (called New Directions) focused on long-term self-sufficiency through extensive education and training. Colorado's waiver in five counties, Colorado Personal Responsibility and Employment Program (CPREP), emphasized skill development over immediate employment, but it also sought to increase employment through more generous earned income disregards, stricter sanctions for noncompliance, and a two-year limit on education and training activities. In contrast, Colorado's TANF approach moves to a work-first philosophy using tougher work requirements and sanctions rather than incentives. Texas, on the other hand, a state in the extensive experimentation category, had already adopted strong initiatives to move a greater share of its caseload into employment in 1996, and its relatively low benefit level limited the number of families eligible for assistance.
Wisconsin eliminated the earned income disregards, intending to move families away from welfare dependency when it implemented its W2 program in September 1997. A family with earnings from an unsubsidized job receives only those earnings plus any other non-TANF federal and state income supports for which it qualifies.
Table 9 also shows the maximum amount recipients can earn before losing eligibility for cash assistance, which depends on the amount a recipient can earn before the benefit is reduced (the disregard), the benefit reduction rate that applies to earnings above the disregard, and the maximum benefit level. Most states in the moderate experimentation category stand out in their generous earnings limits (and maximum benefit levels), which allow recipients in three-person families to earn more than $1,000 a month before losing eligibility for cash assistance. The exception is Michigan (the state in this category with the most advanced work-first policy before TANF), which retained its pre-TANF earned income disregard policy. Earnings limits are considerably lower in all other sample states except Massachusetts and New York, which are high-benefit states.
Exemptions for Age of Youngest Child
Under waivers, most states exempted mothers of very young children from work activities, but the federal TANF work participation requirements forced states to review these policies. Under the pre-TANF federal policy rules, states could require mothers to participate unless they had a child under age one (the federal work participation requirement exempted mothers with a child under age three), and some states had waivers allowing them to require mothers with younger children to participate (10 states had these waivers, including two of our sample states, Florida and Michigan28).
Some states spent a considerable amount of time debating work exemption policies, addressing such issues as treating welfare mothers and other working mothers equitably and the effects of lowering the exemption age on the state's child care policy. The age of youngest child exemption stimulated heated debate in many states. To be effective, a policy that requires mothers of infants to work also requires a commitment to financing infant care. Infant care is more expensive than care for older children and in short supply regardless of price in many states.
All but one of the sample states (Massachusetts) lowered their age-of-youngest-child exemption policy. California, Florida, Michigan, New Jersey, and Wisconsin adopted exemptions more restrictive than the federal minimum (12 months).29 Before TANF, two sample states had such stringent limits: Michigan (a statewide policy) and Florida (a pilot). The youngest child exemption was such a contentious issue in California that it delayed passage of its welfare reform legislation into the final hours of a legislative session that had already been extended. The governor proposed a three-month exemption. Perhaps not surprisingly, given the state's fairly traditional approach to cash assistance and its high benefits, the legislature and many of the county governments forced the governor to accept a compromise of six months (allowing counties some flexibility to reduce the exemption for care of a child to 12 weeks or increase it to 12 months on a case-by-case basis).
All the other sample states except two instituted an exemption until the child's first birthday. Massachusetts and Texas--both extensive experimentation states--extended the exemption to six and four years, respectively. The Massachusetts decision was driven both by a desire to allow mothers to stay home until their children were ready for elementary school and a concern over the availability and cost of child care for young children. The Texas decision was driven by concern about the child care costs that a more stringent exemption would entail.
Time-Triggered Work Requirements
TANF requires that nonexempt recipients of federally funded cash assistance engage in activities that focus on work or work preparation after two years on welfare (or sooner at state option). While most states have indicated that they will conform their programs to the federal requirement, some have adopted policies that require participation only in employment or unpaid work after a set period of time. Massachusetts had a formal pre-TANF time trigger, requiring employment (subsidized or unsubsidized) or community service after 60 days, and California required 100 hours of community service per month for recipients participating in Greater Avenues for Independence (GAIN, California's JOBS program) for 22 out of 24 months.30 Two states in our sample implemented new formal work triggers as part of their TANF programs, and Massachusetts retained its pre-TANF trigger. California now requires unsubsidized employment or community service after 18 months on welfare (two years for those on welfare when the new policy was implemented). Wisconsin's TANF approach moves clients into unsubsidized employment, trial jobs, or community service as soon as they begin receiving welfare as bridges to regular paid employment. The other sample states have not adopted explicit time triggers requiring employment or community service, relying instead on work activity participation requirements to move clients into jobs.
Sanctions
Under AFDC/JOBS policy, individuals failing to comply with work participation requirements lost benefits for the adult in the unit until compliance; the second failure drew a sanction that lasted for at least three months or until compliance (whichever was longer); and the third and subsequent failures drew at least a six-month sanction. Federal food stamp benefits increased as sanctioned families' AFDC cash benefits decreased, thereby weakening the financial effect of the sanction. State and local offices in our sample states, particularly those with relatively low benefits, often reported that these sanctions were ineffective because the sanction amount was so low that it did not motivate recipients to comply with new work requirements. Many respondents reported that retaining cash assistance for the children and food stamps outweighed the loss of the adult AFDC allotment.
Federal TANF rules eliminated the food stamp adjustment for sanctioned families and placed no restrictions on states' ability to design their sanction policies, including the length and amount of the sanctions. Table 9 characterizes TANF sanction intensity as low for states that never impose a full family sanction (that is, loss of the entire TANF benefit), medium for states that impose a full sanction only after repeated warnings, and high for states that impose the full sanction after the first compliance failure. While the sanction intensity increases in all states as the number of compliance failures increases--either by increasing the length of time the sanction lasts before benefits are restored or increasing the amount of benefit reduction31--the use of full family sanctions seems to serve as the best indicator of the intensity of the sanction. Families who can lose all cash assistance because of sanctions face the stiffest penalties and thus the strongest motivation to comply.
Three states in our sample--Florida, Mississippi, and Wisconsin (all extensive experimenters before TANF)--are at the high end. In contrast, three out of four of the states in the moderate experimentation category and New York in the minimal experimentation category--all states with relatively high benefits--use low-intensity sanctions, indicating their more traditional philosophy of maintaining a relatively strong cash assistance safety net for families with children. The exception is Michigan, which had moved toward a strong employment-focused program before TANF. Michigan used its flexibility under TANF to increase the intensity of its sanctions and now imposes a full family sanction after repeated failures to comply with work participation requirements. Michigan is also one of two states in the country that includes the food stamp benefit in its sanction (the other is Florida), making its sanction policy stronger than other states in the medium-intensity category.32
Sanction intensity, of course, does not indicate the level of enforcement, which is critical to its effectiveness. Respondents in our site visits reported wide differences in the rigor of sanction implementation across states (even among states whose formal sanction policies were identical) and across counties within states. The full effects of TANF sanction policies on the caseload will not be fully understood without new data from the states on how often sanctions are imposed.
How Long Should Families Receive Cash Assistance?
States' responses to the new federal time limit requirements take many forms that can only be fully understood in the context of a range of time-limit exemption policies. Table 10 shows three important aspects of time limits--intermediate or "tiered" time limits, lifetime limits, and exemptions to lifetime limits--and a summary intensity measure. Intermediate limits, which are allowed under TANF because states can use lifetime limits shorter than five years, place a maximum on any continuous period of assistance to force recipients into employment or other private support systems quickly but to provide for support if there should be a future need. Lifetime limits are self-explanatory. Exemptions are listed as narrow for states that only permit exemptions for reasons of disability;33 broad for states that exempt individuals who cannot find employment because of a weak labor market, general hardship, or "personal barriers"; and medium for states with policies falling between these two extremes.
Time limits were clearly an area of major debate and change across the states. As table 10 shows, nearly all the time-limit policies in our sample states were adopted in response to the new federal legislation. Texas and Washington are the only states that implemented new, statewide time-limit policies in 1996 through federal waiver approval.34 Texas had adopted a tiered time limit that was being implemented gradually beginning in June 1996, and Washington had a 10 percent benefit reduction each year until benefit exhaustion (beginning after an adult was on welfare for four years). But even these states had to make changes to conform to the new federal TANF rules. Texas retained its tiered policy but also will be bound by the five-year lifetime limit in federal law, and Washington changed its policy to conform to the federal maximum. As described below, Florida and Wisconsin, states with time-limit pilots before TANF, were able to move toward the time-limit policies previously prohibited under waivers.
States in our sample are distributed fairly evenly among low-, medium-, and high-intensity time limits. While all three states with high-intensity time limits (Florida, Texas, and Wisconsin) fall into the extensive experimentation category, two other states in this category adopted medium- (Mississippi) or low- (New Jersey) intensity time limits.35 The two states in this category that had time-limit pilot programs in place before TANF--Florida and Wisconsin--moved toward high-intensity time limits with intermediate limits, narrow exemptions, and, in the case of Florida, a four-year lifetime limit. Florida's high-intensity time-limit policy, which is consistent with the strong dependency-reduction theme of its WAGES program, is the only state in our sample that adopted a time limit shorter than the federal maximum.36 The new Wisconsin policy still retains a five-year lifetime limit, but it also includes two-year time limits for three components of its W-2 program (trial jobs, community service jobs, and transitions). Wisconsin has no formal exemptions to its time limit, based on a belief that it will be able to move nearly all welfare recipients into self-sufficiency within five years.
In contrast, most of the states with low-intensity time limits, consistent with maintaining more of the entitlement nature of cash assistance, were the high-benefit states falling into the minimal and moderate experimentation categories. California, New York, and Michigan adopted the most generous time limit policies in our sample. California guarantees cash assistance to children (given county verification that employment is not an option). New York offers assistance in the form of vouchers for the entire family.37 California also adopted broad time-limit exemption policies. Michigan rejected time limits as a means of enforcing its intensive work-first policy. It had been gradually reforming welfare since 1992, and its policies reflected a strong conviction that requiring work for nearly all the caseload, along with rigorous enforcement of its strong sanctions, would be enough to move able-bodied recipients into the labor market. Respondents in Michigan reported that the state intended to move families who exhaust the federal time limit into a state-funded program, although no plans were yet under way to design such a program.
New Jersey, a state with extensive experimentation under way before TANF, debated its time-limit policy at length, with some arguing that time limits would penalize families complying with work requirements who were not yet earning enough to leave welfare. Ultimately, an extension policy was added to its five-year lifetime time limit that will allow some recipients up to 12 months in additional benefits, a compromise that allows flexibility for those who have a difficult time moving to self-sufficiency. New Jersey also adopted a broad exemptions policy, providing exemptions for those who face personal barriers to employment or poor labor market conditions, in addition to the more common exemptions for disabled persons or for victims of domestic violence.
What Part Should Cash Benefits Play in the System?
While states were free to set their own benefit levels before TANF, the new legislation gave states more flexibility over payment policies.38 For example, states are free to adopt a family cap, a policy some states with waivers used before TANF to eliminate or reduce any increase in benefits when a child was born to a family on welfare. The new legislation also offered states an incentive to cut benefits, since the block grant allocations were fixed over five years, and states may use savings from what they spent on welfare for other purposes. Countering this incentive, however, was the "windfall" that most states received from the federal block grant.
Only two states in our sample changed basic benefit levels as part of their new TANF policies (table 11). California decreased benefits, but this was a continuation of a trend toward reducing welfare costs established long before TANF. Wisconsin adopted a unique payment policy in which all families qualify for a flat benefit, resulting in slightly increased benefits for smaller families and decreased benefits for larger families.
California and Florida adopted new family caps under TANF, bringing the total number of sample states with caps to six.39 (Wisconsin had a pre-TANF family cap policy, and its new flat-benefit policy has the same effect as a family cap.) All but one of the states with family caps fell into the extensive experimentation category. California, a state with a more traditional program before TANF, adopted a family cap policy after the contentious debate that characterized other facets of its welfare reform legislation. The legislature fought the governor on the family cap, but unlike its record on time limits and child exemption policy, it failed to achieve a compromise.
Implications of States' Early TANF Policies
One intent of the federal welfare reform legislation was to allow states more freedom to run programs that provide assistance to families with children. In some ways, however, the new legislation gives states less freedom to design cash assistance programs than they had during the waiver era, because the legislation mandates important features of states' welfare programs and sets up financial incentives to direct reform in other ways. The federal government, for example, mandated a maximum five-year termination time limit (allowing an exemption for 20 percent of states' current caseloads), forced states to move recipients into work activities at the end of two years, and established financial penalties for states not meeting federally defined work participation rates for adults receiving assistance that were significantly higher than those imposed under JOBS.
The new federal legislation constrains states in another way as well. States can sidestep the federal program time limits only by providing non-TANF state funds to support recipients who become ineligible for federal TANF support. Because the block grant allocations for states, based on their AFDC spending, are fixed for five years, and because states' MOE requirements are set at only 80 percent of pre-TANF spending, states can support recipients who become ineligible for federal TANF support only as long as caseloads remain low. Even so, the contrast between TANF and the most recent waiver era--when states could experiment with most aspects of their AFDC programs--is not as stark as the rhetoric surrounding the new legislation seems to imply.
While states' new TANF policies were clearly designed to conform to federal parameters, their responses also reflected their previous approach to cash assistance. State choices since TANF definitely are increasing diversity in cash assistance policies across the nation. Some states are moving further toward policies that discourage the use of welfare. Others are maintaining large parts of the safety net available under AFDC. All states are moving toward a stronger focus on work, but they emphasize different approaches to accomplish this goal. It is clear that, even among states that had been experimenting with broad changes to AFDC before TANF, welfare programs are undergoing a great deal of change.
Below we assess the nature of this diversity in cash assistance policies across the states. We also discuss the importance of state and local policy implementation to the ultimate success of welfare reform and to defining the diversity in states' programs.
The Changing Nature of Cash Assistance in Our Sample States
Table 12 summarizes policy emphases in the sample states since TANF, with TANF policy changes in bold type. While somewhat fewer changes occurred in the states with extensive experimentation under way, nearly all states in our sample changed more than one major policy. Massachusetts provides the exception. The medium-intensity benefit time limit was its only new formal policy, and, in fact, the state had sought, but was denied, approval for this time limit as part of its 1995 waiver.
There is more conformity among the states for some policies than for others. Nine out of 13 states now treat two-parent families who need assistance the same as single-parent families, mostly eliminating all special work rules for two-parent eligibility. While 10 states had been experimenting with two-parent eligibility before TANF, only 3 states had eliminated all special rules applying to two parents. Eight out of 13 states have formal diversion assistance programs (compared with only one state before TANF). While we showed earlier that the amount of states' diversion assistance payments varies, the adoption of this policy indicates that most states in our sample value the ability to offer applicants short-term assistance to resolve need and prevent entry into welfare.
Our categories indicating the degree of pre-TANF departure from federal AFDC rules provide some insight into states' early TANF decisions, but these decisions also were clearly affected by the states' historic generosity and, in some cases, a desire to give local governments more control over cash assistance. With the exception of Michigan, states with relatively high benefits and minimal or moderate experimentation before TANF (New York, California, Minnesota, and Washington) tend to prefer approaches that encourage families to move off welfare through incentives rather than through penalties. These states have high-intensity earnings incentives and low-intensity sanction policies (under which families never lose their entire benefit as a result of a sanction), combined with low- to moderate-intensity time-limit policies. All four of these states adopted diversion programs, offering families cash incentives for staying off welfare. Michigan, a high-benefit state in the moderate experimentation category because it primarily focused its pre-TANF changes on work, did intensify its sanction policy, but it did not adopt a time-limit policy, trusting that its welfare-to-work policies would move most recipients into jobs.
States with the most extensive pre-TANF experimentation changed their policies less than other states. However, Florida, New Jersey, and Wisconsin shifted policies considerably. Florida and New Jersey moved to a stronger work-first model of welfare with even higher earned income disregards and fewer exemptions from work activities. However, New Jersey also maintained policies consistent with its relatively strong safety net by holding to moderately intensive sanctions and allowing two six-month extensions to its five-year time limit. In contrast, Florida adopted high-intensity sanctions and time limits to reduce families' dependence on welfare.
Texas's cash assistance policies stand out among those of the extensive experimenters because they combine a broad exemption for mothers of young children, low-intensity sanctions, and no family cap. On the other hand, Texas discourages dependence on cash assistance by combining a relatively low monthly benefit ($188 per month for a family of three with no other income, compared with the national average of $393 per month) with a high-intensity time limit (continuous benefit receipt limited to one, two, or three years depending on work readiness and a five-year lifetime limit). Once again, Wisconsin stood out as the leader in experimentation. As noted earlier, Wisconsin had applied for a waiver for its W-2 program during the federal debate over welfare reform. Once PRWORA passed, Wisconsin could implement these policies without federal waiver approval.
The debate in three states in our sample--Colorado, California, and New York (all operating county-administered models of cash assistance)--was characterized at least in part by an historic desire to give counties a greater role in setting social welfare policy. Colorado moved furthest in this direction by allowing counties to determine diversion policies and work requirements. While New York's counties were not given control over any of the specific policies shown in table 12, their TANF plan provides performance awards based on job placements, and local districts can apply for waivers from state regulations.
It is important to note some optional features that most states in our sample did not adopt. Only two states adopted new family caps under TANF (and the three states that had adopted family caps under waivers retained them); only one state reduced its basic benefit. While the family cap generated considerable debate in the U.S. Congress, and several versions of the federal welfare reform bill proposed to mandate family caps for all states, many states did not believe that this policy was effective or necessary. The family cap separates the pre-TANF experimenters from the more traditional states (only one state in the pre-TANF minimal and moderate experimentation categories adopted a family cap, compared with five out of six extensive experimenters, including Wisconsin's new flat-benefit policy). This may be an example of the federal government taking its cues from the policies of the early experimenters with state welfare reform that were not reflecting widespread desires of states.
The Importance of Implementation
While assessing states' policy choices illustrates the variety of approaches to cash assistance, the ultimate key to success in these efforts will lie in reforming how services are actually delivered to clients. Failures of earlier welfare reform efforts have been attributed in no small part to overlooking the need for organizational reforms that articulate new goals clearly to caseworkers and clients, provide adequate resources to welfare bureaucracies, and build the organizational capacity to effect policy change efficiently.40 In addition, county welfare offices across different states and even within a single state may interpret and administer policies differently even when they appear identical.
In many states, new cash assistance policies represent a major departure from the past. Diversion assistance and lifetime time limits clearly present new implementation challenges. However, most states reported that forging a direct link between welfare and work within a traditional welfare office was still a significant challenge. The major product delivered by local welfare organizations for decades has been cash assistance, but now state and local welfare programs are expected to deliver economic self-sufficiency.
The clearest message we heard in our site visits to state and local welfare offices is that most will adopt new procedures or complete transitions begun under waivers to ensure that new work-focused programs will be successful. Most offices in our sample were still operating an "eligibility-compliance" model of welfare and recognized the need to reorganize, retrain eligibility workers, and establish new collaborations among welfare eligibility offices and providers of employment services to focus on employment. As we have seen, most states furthest along the road to welfare reform before 1996 were also making major changes in their service delivery systems to accommodate a more work-focused system. Typically these changes involved giving local offices more flexibility in designing policies to move clients into work activities, even in states with state-administered models of cash assistance. But many respondents were worried that the TANF time clocks for work participation and benefit termination were ticking away while these reforms were still getting under way, raising the specter that--even in a good economy--some state changes focused on service delivery might come too late for those whose clocks are ticking.
Notes
1. The sample includes California, Colorado, and Washington in the West; Michigan, Minnesota, and Wisconsin in the Midwest; Alabama, Mississippi, Texas, and Florida in the South; and Massachusetts, New Jersey, and New York in the East. Relative AFDC program sizes varied across the sample, ranging from a low of about 3 percent of the nonelderly population for Alabama and Colorado to almost 10 percent for California.
2. While the new federal law places more restrictions on what states can count as work activities for the caseload than were allowed under the former federal Job Opportunities and Basic Skills (JOBS) program, work activities can extend to job search, education, and training activities for short periods of time. TANF work activities can include unsubsidized employment, subsidized private or public employment, work experience, on-the-job training, job search and job readiness for up to six weeks, community service, vocational education for up to 12 months, provision of child care to TANF recipients, job skills training, education directly related to employment, and high school education or its equivalent. However, only the first nine activities count toward the first 20 hours of work activity participation per week.
3. The social welfare system of each of the sample states is summarized in more detail in a series of papers entitled Income Support and Social Services for Low-Income People.
4. We were not able to visit state offices in Mississippi.
5. This material is summarized from L. Jerome Gallagher, Megan Gallagher, Kevin Perese, Susan Schreiber, and Keith Watson, One Year after federal Welfare Reform: A Description of State Temporary Assistance for Needy Families (TANF) Decisions as of October 1997 (Washington, DC: The Urban Institute, June 1998).
6. Other papers in this series include Sharon Long, Gretchen Kirby, Robin Kurka, and Shelley Waters Boots, Child Care Assistance under Welfare Reform: Early Responses by the States (Washington, DC: The Urban Institute, 1998), and forthcoming assessments of states' education and training services and child welfare services, and the organization of the social safety net in the states. All of these papers provide an
