Table of Contents | Previous | Next |
Chapter Two
CONTEXT FOR UNDERSTANDING FEDERAL AND STATE WELFARE POLICY REFORMS
In this chapter, we provide some context for understanding federal and state welfare policy reforms, beginning with a very brief historical overview of federal welfare policy. We then introduce the economic model that provides a framework for our synthesis of the literature. Finally, we discuss four broad categories of welfare reforms and use the economic model to discuss their expected effects on behavior.
2.1. A BRIEF HISTORY OF FEDERAL WELFARE POLICY5
Title IV-A of the 1935 Social Security Act established the Aid to Dependent Children (ADC) program, renamed Aid to Families with Dependent Children (AFDC) in a 1962 amendment to the Social Security Act. Where previously there had been no federal program of poor support, ADC/AFDC was a joint state-federal program:
designed to release from the wage earning role the person whose natural function is to give her children the physical and affectionate guardianship necessary . . .to] rear them into citizens capable of contributing to society.6
Initially, the program was designed to serve needy children under 16 with only one able-bodied parent at home. Over time, other family members–e.g., older children (1940), the mother or other caretaker (1950), the child of an unemployed parent and that parent (1961), an unborn child in the last trimester of pregnancy (1981)–became eligible for benefits.
For whatever eligibility rules were in effect, the federal government defined the basic program as an entitlement. All eligible individuals who applied had to be enrolled, sufficient funding had to be found to pay for the benefits, and recipients were entitled to certain due-process protections. The states set the standard of "need" and the payment level and operated the program with state (or sometimes county) employees. Funding was provided by both the state and federal governments in proportions that varied across the specific programs and across the states.
In 1960, the official program goals were revised to include work for recipients, and beginning in 1968, the Work Incentive (WIN) program was established to make work or training programs available for certain AFDC recipients. Those programs applied to mothers whose youngest child was age 6 or older and were for the most part small and ineffectual. However, in the early 1980s when welfare agencies received additional authority to operate their own work and training programs, these programs became the object of considerable policy innovation and careful evaluation. Those innovations and evaluations served as a research base for the 1988 Family Support Act (FSA).
The FSA reinforced this emphasis on self-sufficiency through the mutual efforts of the recipient and the government. On the work dimension, the FSA replaced WIN and other welfare-to-work programs with the Job Opportunities and Basic Skills Training (JOBS) program. JOBS increased funding and strengthened the participation mandate for welfare-to-work programs and included financial penalties for recipients who failed to comply. The age-of-youngest-child exemption was lowered from 6 to 3 (or as low as 1, at state option), reflecting, in part, increase in labor force participation among nonwelfare mothers with pre-school-age children. During this period, with the increased emphasis on self-sufficiency, the program rules also changed to provide some financial incentives for work by allowing working recipients to keep more of their earnings. In practice, implementation of the FSA and JOBS was crippled by the recession of the early 1990s. The combination of a worsened job market for welfare recipients and a cash crunch in many state governments resulted in an undersized and underfunded program.
Beginning in the late 1980s, the waiver authority granted to the Secretary of Health and Human Services under section 1115 of the Social Security Act was utilized to grant waivers to particular rules and regulations governing state implementation of AFDC. States could petition USDHHS to implement experimental, pilot, or demonstration projects they believed would result in a more effective welfare program. These experiments were required to be cost-neutral and to include a rigorous evaluation (usually random assignment). By the time of PRWORA’s passage, USDHHS had approved waivers for more than 40 states, many of them for statewide reforms (USDHHS, 2000, p. 259).
In August 1996, PRWORA passed and was signed into law, replacing AFDC with TANF. Beyond providing aid so that needy children could be cared for in their homes, the key objectives of TANF included ending the dependence of needy parents on government benefits by promoting job preparation, work, and marriage; preventing and reducing the incidence of out-of-wedlock pregnancies; and encouraging the formation and maintenance of two-parent families. In implementing these objectives, PRWORA eliminated many federal requirements on state welfare programs and allowed states nearly complete discretion in setting eligibility standards. The entitlement status of welfare was abolished, and federal funding of the AFDC and JOBS programs (as well as Emergency Assistance to Needy Families) was consolidated into a single TANF Block Grant. The TANF Block Grant was funded at 1994-95 spending levels. Thus, as caseloads and cash assistance declined, more funds were available for services that were to be provided to fewer cases.
During the year following passage of PRWORA, almost all states started the process of replacing their AFDC programs with their TANF program. Implementing those programs often stretched well into 1998, and refinements continue to be made. In most states, the changes have been substantial. Exploiting their new discretion and block-grant funding, most states changed their welfare programs beyond what was required by TANF (Gais et al., 2001). As a result, welfare policies that a decade earlier had varied across states primarily by the size of the welfare benefit now vary along dozens of dimensions.
2.2. KEY POLICY REFORMS UNDER WAIVERS AND TANF
Policy reforms under waivers, and even more so under TANF, change many different aspects of the basic AFDC/JOBS model. Here, we review the four changes that have received the most attention in the research literature: (1) the financial work incentives implicit in the benefit structure, (2) mandates to participate in work-related activities, (3) time limits, and (4) restrictions on living arrangements for minors and family caps. Many of these reforms were initially instituted under section 1115 waivers during the mid-1990s; some represent extensions of policies implemented under prior reform legislation.7
Our discussion begins with a brief review of the standard economic model of welfare programs. We then use this theory in our discussion of the likely effect of each of these reforms. Specifically, for each of these reforms, we briefly describe the policies that were in place at the beginning of the reform era. We then describe the range of policies that states adopted under waivers and TANF.
2.2.1. The Standard Economic Model
The economic model of the consumer’s choice of labor supply is useful for integrating the results of the studies. The model begins by noting that the typical consumer (potential welfare recipient) would prefer both more leisure and more income.8 More work would yield greater earnings, but at the expense of less leisure. As a result, the consumer faces a trade-off. Her decisions about whether to go on welfare and whether and how much to work will depend on her wage and the structure of the welfare system.9
This simple model predicted behavior fairly well in the pre-reform environment. Under AFDC, the key policy parameters were the size of the welfare grant and the benefit reduction rate, which is the implicit tax rate by which the recipient’s grant is reduced as her earnings rise. The standard economic model predicts that a larger grant should increase welfare use while reducing employment and hours of work. It also predicts that a lower benefit reduction rate should increase employment. These predictions have been largely borne out by a substantial body of systematic empirical evidence (Moffitt, 1992).
The model also predicts how consumers should respond to some of the recent reforms; the effects of other reforms can be predicted by extending the basic model. As we discuss the recent reforms, we note how the model predicts that they should affect behavior, or how the model can be modified to incorporate their effects. The model serves an important function by providing us with a framework for synthesizing the literature–that is, for distilling a whole from the parts represented by several dozen independently conducted studies.
2.2.2. Financial Work Incentives
The debate about financial work incentives dates back to the 1967 Amendments to the Social Security Act. Prior to 1967, welfare benefits were reduced by one dollar for each dollar that the recipient earned, so increased work did not result in increased income. This benefit structure discouraged work. The 1967 Amendments sought to encourage work by allowing recipients to keep the first $30 of their monthly earnings (referred to as a $30 "earnings disregard") and by reducing the benefit reduction rate for additional earnings to 67 percent (i.e., the recipient retained one-third of each additional dollar of earnings). Subsequent legislation–1981 Omnibus Budget Reconciliation Act (OBRA), 1984 Deficit Reduction Act (DEFRA), and 1988 FSA–further modified these formulas, but the incentives remained approximately constant.10 Beyond some minimal threshold, a welfare recipient who worked kept at most a third of her earnings. From OBRA 1981 until the eve of PRWORA, after the fourth month of work, the benefit reduction rate was 100 percent.
States experimented with financial work incentives under waivers and incorporated them into their TANF plans. Although a few states maintain the AFDC incentive schedule, most now have more generous schemes in place. At least one state (Connecticut) now allows a recipient to keep all of her earnings up to the federal poverty level without losing any benefits.
The economic model discussed above predicts that introducing a financial incentive may have complex effects on behavior. Moreover, the effects may vary between consumers receiving welfare and consumers who were originally income-ineligible prior to the introduction of the incentive. The effect of the financial incentive on welfare recipients is relevant for interpreting the results from the random assignment experiments. Its effect on both groups is relevant for interpreting the results from econometric studies. We discuss these two effects in turn.
The financial incentive increases the return to an hour of work. As a result, it should encourage welfare recipients to begin working. Its effect on the labor supply and earnings of recipients who were already working prior to the policy change is ambiguous. In addition to increasing the return to an hour of work, introducing the financial incentive raises the recipient’s take-home income for a given number of hours of work. The effect on the return to work, known as the "substitution effect" in the economics literature, provides an incentive to increase work hours. The effect on take-home income, known as the "income effect," provides an incentive to decrease work hours. Because the substitution and income effects work in opposite directions, the net effect of a financial incentive on labor supply and earnings of welfare recipients is ambiguous (Blank, Card, and Robins, 2000). Most observers would expect the substitution effect to dominate for low-income families, which would imply that stronger financial work incentives should increase labor supply and earnings among welfare recipients. However, we find some cases where earnings appear to decrease, suggesting that the income effect may dominate in some cases.
Beyond its effects on work, the financial incentive should increase the welfare use of welfare recipients. Because the financial incentive allows the recipient to keep more earnings while remaining on aid, it provides an incentive to remain on aid longer. It is important to note that the incentive to remain on welfare arises from an incentive to combine work and welfare; financial work incentives should decrease the fraction of the caseload that receives aid without working.
The effects that the financial incentive may have on originally ineligible workers will depend on how the incentive is implemented. Expanding the financial incentive increases the level of earnings at which a working family becomes ineligible for aid. If this new eligibility threshold is applied to both new welfare applicants and ongoing recipients, then the increase in the eligibility threshold will make some previously ineligible families newly eligible to receive welfare. Such families may reduce their labor supply and start receiving benefits.11 If the new eligibility threshold is applied only to ongoing recipients, as is the case in many states, then previously ineligible workers would remain ineligible.
In summary, the model predicts that, among welfare recipients, a financial incentive should increase employment and increase welfare use by raising the fraction of recipients who combine work and welfare. Its predicted effects on hours of work, earnings, and income are ambiguous because of opposing income and substitution effects. If the new eligibility threshold is applied to new applicants, then the increase in welfare use should be greater than it would be if the new eligibility threshold applied only to ongoing recipients. Applying the new threshold to new applicants would also make the program more likely to reduce labor supply and earnings.
Finally, there may also be a reporting effect, sometimes referred to as "smoke-out." Financial work incentives may cause some recipients to report some previously unreported earnings in order to reduce the risk of being caught cheating. Work requirements, which are discussed in the next subsection, may also have such effects. However, smoke-out does not reflect a change in work effort, but merely a change in reporting. This implies that some of the recent improvement in employment and earnings may be apparent rather than real. Despite its potential importance in interpreting recent labor market trends, we are unaware of any evidence on the magnitude of the smoke-out effect.
2.2.3. Mandatory Work-Related Activities and Sanctions for Noncompliance
While work requirements date back to the WIN program implemented in 1968 and the Community Work Experience Programs (CWEP) created under OBRA 1981, the baseline for recent work-related activity mandates was the 1988 FSA’s JOBS program. JOBS was intended to be a mandatory program, exempting primarily single parents with children under age three.12 Nonexempt welfare recipients were required to participate in work-related activities, which often involved basic skills programs. Those who failed to participate were subject to sanctions, which involved forfeiting the adult’s portion of the AFDC benefit. However, less than half of the welfare caseload was required to participate in JOBS, and actual participation rates were much lower.
Because JOBS’ work mandates were widely perceived as too weak, many states strengthened their work-related activity mandates under waivers. Common modifications include higher hours requirements, more restrictive definitions of permissible work-related activities, and lower age-of-youngest-child-exemptions. Many states reoriented their welfare to work programs as well, emphasizing job search and employment (so-called "work-first" programs) over basic skills and education.
TANF accelerated and in some cases required such changes. Under TANF, states were required to increase the fraction of their caseload participating in work-related activities. Furthermore, TANF limited the extent to which education and training could be used to satisfy this requirement. In practice, the sharp caseload decline and TANF’s caseload reduction credit–which implicitly treated any fall in the caseload not related to stricter eligibility rules as participation–rendered these aggregate participation rate requirements largely meaningless.
The effects of a perfectly enforced work-related activity mandate can be predicted by extending the standard economic model to incorporate the requirement that welfare recipients spend the prescribed amount of time engaging in permissible work-related activities. The mandate is obviously predicted to increase participation in such activities, and to the extent that the permissible activities include working, they may increase employment as well. Work-related activity mandates should also decrease welfare use. Given most states’ benefit structures, an increase in work will raise earnings. Especially in low break-even states, work will yield income high enough to make a family income ineligible for welfare. Furthermore, earnings near that level may induce a family to leave welfare–the small payment is not worth the effort and stigma of remaining on welfare. In addition, by reducing leisure–and the utility derived from that leisure, work-related activity mandates may induce some recipients to find a job and others to leave welfare for other sources of support (e.g., family or friends).
In the more realistic case of imperfectly enforced work mandates, these effects may be muted. The extent to which they are muted will depend on the extent of enforcement. The less strictly the mandate is enforced, the less pressure recipients will feel to participate in the work-related activities and the less likely they will be to leave the welfare rolls. In the real world, enforcement is likely to be important in determining the effects of mandated work-related activities.
Enforcement can take various forms. Monitoring is one method that has been used to achieve compliance with the mandates. Sanctions are another. Most states have raised sanctions for noncompliance, both for first and subsequent violations. Some states now have full-family sanctions for initial violations, and, in some states, third violations can lead to lifetime case closures.
Sanctions can also be incorporated into the standard model using insights from deterrence theory, which has long been used to study street crime, insider trading, and lesser forms of rule-breaking such as shirking on the job (Becker 1968; DeMarzo, Fishman, and Hagerty, 1998; Dickens et al., 1989). In deterrence theory, the consumer decides whether to comply with a rule as a function of two things: the likelihood of being detected if she fails to comply, and the penalty associated with detection. The model predicts that compliance will rise as the detection risk rises and as the penalty for noncompliance increases. In the welfare context, we would expect higher detection risk and higher penalties for noncompliance to increase participation in work-related activities, increase employment, and decrease welfare use. Of course, some families who fail to comply will be detected and sanctioned, which will further decrease the caseload.
2.2.4. Time Limits
Time limits, which reduce or eliminate a recipient’s benefits after a specified amount of time on aid, are among the most radical of the reforms instituted under TANF.13 AFDC was an entitlement under which families were to receive aid for as long as they remained otherwise eligible. Prior to TANF, however, about a dozen states implemented waivers to time limit cash assistance. Consistent with this trend, with a few exceptions, TANF mandated a 60-month maximum time limit for adult receipt of federally funded benefits.
Although about half of the states have imposed the federal 60-month limit, many have adopted shorter limits. Some specify an intermittent limit, for example, allowing for only 24 months on aid within any 60-month period. Some have both intermittent and lifetime limits. Michigan and Vermont effectively have no time limit, having chosen to fund aid receipt in excess of the federal limit from state-only funds.14 States also vary as to their exemption and extension policies.
Time limits may have complex effects on both welfare use and other dimensions of behavior. Like financial work incentives, they have mechanical effects, reducing welfare use and caseloads as families reach the time limit. They may have behavioral effects as well, leading families to exit the welfare rolls before they actually exhaust their benefits to preserve ("bank") their months of eligibility for future use.
Grogger and Michalopoulos (1999) have extended the standard economic model to include time limits. Their model predicts that time limits should have the greatest behavioral effects among the families with the youngest children. The reason is that these families have the longest time before their youngest child turns eighteen, making them ineligible for aid. Thus, they have the longest time over which to spread their benefits, giving them the greatest incentive to save some for the future.
2.2.5. Minor Residence Requirements and Family Caps
PRWORA did not focus only on work. PRWORA specifically called attention to family structure, with explicit goals of reducing the incidence of out-of-wedlock pregnancies and encouraging the formation and maintenance of two-parent families. It sought to solve the problem created by the perception that AFDC encouraged young women to have children out of wedlock and discouraged them from marrying.
Beginning in the waiver period and continuing with increased intensity into the TANF period, states took several actions to address these concerns. First, the FSA gave states the option of requiring minors to live with an adult (usually a parent, but also with a guardian, or in a supervised setting) in order to received benefits. These provisions are commonly referred to as "minor residence rules." By the time of PRWORA’s passage, 13 states had adopted such provisions as part of their state TANF plans and another 11 states had adopted these provisions under waivers. PRWORA mandated such provisions for each state’s TANF program.
Similarly, under AFDC, welfare payments increased with family size. It was claimed that this provision gave welfare mothers an incentive to have more children. Under a "family cap," the welfare benefit does not increase with the birth of a new child who is conceived while the mother is receiving cash assistance. Beginning with New Jersey in 1992, 14 states received waivers for such a provision. Under TANF, 16 states provided no increases in benefits with an additional child, 2 states provided only a partial increase, 3 states provided the increase in the form of a voucher, and 2 states had flat grants independent of the number of children for all families (regardless of whether a child was born while the mother was receiving welfare).
Although family caps and minor residence rules were the most common reforms to address concerns about family structure, they were not the only ones. Other reforms included changing the eligibility rules for two-parent families (what had been the AFDC-Unemployed Parent program) to lessen disincentives to marriage and closer links to family planning programs. However, these other reforms have received little study and, therefore, receive less attention in the synthesis that follows.
While the primary objective of these policies is to reduce teen childbearing and subsequent pregnancies among welfare participants, these policies may reduce welfare receipt by reducing nonmarital births (especially first births) and more generally by making welfare less attractive. Such policies may therefore reduce welfare entry, as well as increase welfare exits.
5This section draws on various editions of the Green Book prepared by the U.S. House of Representatives Committee on Ways and Means.(back)
6As quoted in Garfinkel and McLanahan (1986), Chapter 4.(back)
7A variety of other reforms were approved under waivers and have been incorporated into state TANF programs. These include removing restrictions on eligibility for two-parent families, increasing asset limits that determine eligibility, instituting parental responsibility requirements (e.g., child immunizations and other preventative health care, and child school attendance), and providing extended transitional child care and health insurance after welfare exit.(back)
8"Leisure" here is used to mean all activities other than work. It includes activities referred to in the vernacular as leisure, but also such activities as household work. In terms of the model, what is important is that the activities referred to as leisure be preferable to work, all else equal, and not be paid.(back)
9Throughout, we use the feminine pronoun in referring to welfare recipients. The overwhelming fraction of adult welfare recipients are female.(back)
10Among the changes with OBRA 1981 was a limit to work expenses of $75 per month plus actual child care costs up to $160 per child. In addition, the $30 plus one-third formula applied only to the first four months of work. DEFRA 1984 extended the $30 disregard to 12 months, but kept the one-third disregard at only four months. The 1988 FSA increased the limit on work expenses to $90 and the limit on child care to $175 per child over age two and $200 per child under age two.(back)
11Indeed, some families that remain just above the new eligibility threshold may reduce their hours (and earnings) in order to qualify for benefits.(back)
12Beyond mothers of children under age three and those working at least 30 hours a week, the other exemptions included the ill, the disabled, those over 60, those living in remote areas, those needed in the home, those in their last trimester of pregnancy, and those for whom guaranteed child care was not available.(back)
13Although the term "time limits" also has been used to refer to the deadlines by which recipients must satisfy work-related activity mandates, we reserve the term for what has also been referred to as "termination" or "benefit-reduction" time limits.(back)
14There is some ambiguity about whether and how New York will continue benefits after the time limit.(back)
| Table of Contents | Previous | Next |