Disability, Welfare Reform, and Supplemental Security Income

Summary

The enactment of the Personal Responsibility and Work Opportunity Reconciliation Act in 1996 replaced Aid for Families with Dependent Children with Temporary Assistance for Needy Families (TANF) and had important implications for persons with disabilities and for the Supplemental Security Income (SSI) program. Various surveys since its enactment show that between 32 percent and 44 percent of TANF recipients of report having impairments or chronic health problems. The difference in estimates, most of which are based on national surveys, is in part due to differing definitions of disability used in the surveys and possibly also to timing. Because persons with disabilities often have substantial barriers to employment, TANF work requirements and time limits potentially could have a more pronounced effect on them.

Because the Personal Responsibility and Work Opportunity Reconciliation Act is intended to encourage through incentives and penalties most welfare recipients to work, two key outcomes reviewed are employment and sanctions for noncompliance with program requirements. Not surprisingly, surveys have consistently found that TANF recipients with disabilities were substantially less likely to be working than those without disabilities. Additionally, although most states have formal policies to exempt persons with disabilities from various requirements, studies of results in the states have consistently found that families with reported disabilities or health problems were sanctioned at a higher rate than were other families.

Welfare reform has increased the financial incentives for state governments to promote movement by recipients of assistance from Temporary Assistance for Needy Families to the Supplemental Security Income program. Trends in TANF benefits show an increase in the gain for recipients from doing so. Nationally, between 1996 and 2002, the average gain for a family of three from transferring one child to the SSI program grew by 3.8 percent (in constant dollars). Overlap and interaction between the two programs is substantial: in 2002, 17.1 percent of all TANF cases included an adult or a child recipient of the SSI benefits. One-third of the children receiving SSI benefits in December 2002, whose awards had been granted in the previous 24 months, had come from TANF families. Thirty percent of recent adult women awardees also came from families with links to TANF or Aid to Families with Dependent Children.

Introduction

The Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA) replaced welfare as the country then knew it—Aid to Families with Dependent Children (AFDC)—with Temporary Assistance for Needy Families (TANF).¹ The law continued and extended an upswell of state initiatives, with dramatic results. From 1996 to 2002, the total number of welfare recipients in the nation declined by 58 percent (DHHS 2003b, II–5).

In recent years, issues related to disability among TANF recipients have received increased attention. TANF, by requiring increased work effort by assistance recipients, focused attention on barriers to work, including physical and mental impairments. Estimates of the extent of disabilities among families receiving TANF assistance vary but are commonly high. For example, the General Accounting Office (GAO) estimates that the rate of disability among adult TANF recipients aged 18 to 64 in 1999 was 44 percent—three times the prevalence of disability among adults who were not recipients (GAO 2001a, 9).

Disability among people receiving TANF has ramifications for Supplemental Security Income (SSI), the nation's comprehensive means-tested program of income support for individuals with disabilities (adults and children) and the elderly. Administered by the Social Security Administration (SSA), SSI is the nation's largest welfare program. In 2003, federal TANF expenditures came to $16.5 billion, while SSI benefits totaled $31 billion, more than 80 percent of which went to people with disabilities. There is an overlap among the populations served by the SSI and TANF programs. A significant proportion of new SSI cases is composed of adults or children previously receiving TANF, and nationwide by 2002, more than one out of every six TANF families included an SSI recipient.

After a brief review of the TANF and SSI programs, this article addresses three questions:

• What is the prevalence of impairments in the TANF population?
• How has welfare reform affected access for people with disabilities to social assistance and their incentives for employment and rehabilitation?
• What are the connections, in the administration of the programs and in the experience of the clients, between TANF and SSI for people with disabilities?

The Two Programs

Supplemental Security Income and Temporary Assistance for Needy Families differ in a number of ways. The administration of SSI is uniform nationally; the only significant state-to-state variation that occurs is due to the supplementation that some states add to the federal benefit. By design, AFDC eligibility standards and benefit payments varied across states. PRWORA substantially increased states' discretion in designing and operating their social assistance programs for families with children, but it also required the states to impose stricter obligations on welfare recipients. As a result the divergence in the characters of SSI and AFDC–TANF has increased. Nevertheless, there is substantial overlap in the populations served.

Temporary Assistance for Needy Families

State TANF programs provide cash assistance to needy families with children. Each state has a TANF program (as do the District of Columbia, Guam, Puerto Rico, and the U.S. Virgin Islands). Rules are set by states, subject to restrictions established in PRWORA and incorporated into the Social Security Act. States have always enjoyed considerable latitude in setting standards for eligibility, in setting benefit amounts, and in establishing the degree of emphasis placed on work and efforts to prepare for work as an obligation of the recipients. PRWORA expanded the range of state options with respect to eligibility and obligation but also raised requirements that were imposed with respect to state program content and the duration of federal assistance for individual families.

Eligibility and Benefits. The operation of the Temporary Assistance for Needy Families program is the responsibility of state social services departments or, in some cases, state employment services departments or both. Eligibility requirements for TANF are several: adults who are seeking aid must be caring for children, have few resources and little or no income, and, in some states, be willing to participate immediately in searching for work. Although the child requirement is universal, the resources, income, and job search requirements vary. Because establishing eligibility can take up to a month, many states provide emergency assistance for families in immediate need. Once eligibility is established, monthly payments begin. The variation in state benefit levels is substantial. In 2003, the median state TANF benefit for a family of three was $426 per month. The lowest benefit ($164) was paid in Alabama; aside from Alaska, the highest benefit ($758 to families exempted from work requirements) was paid by California. This variation is partially offset by the Food Stamp program. The lower the TANF benefit, the greater the amounts of food stamps for which recipient families are eligible. In addition, individuals who are eligible for TANF are generally also eligible for Medicaid.

Promoting Work and Addressing Impairments. One of the objectives of TANF is to promote work. Federal rules include time limits for receipt of benefits, require states to achieve certain minimum rates of participation by recipients in work or work-related activities, require penalties for recipients who do not meet work and employment-preparation obligations, and impose fiscal penalties on states that do not meet participation requirements. These requirements signal a congressional presumption that most adults receiving TANF benefits can work and should be expected to do so.

Families seeking TANF benefits often report problems that preclude work outside the home or that make work exceptionally difficult. Some of these barriers, such as a lack of child care or inadequate transportation, are environmental. Others reflect a lack of skills—illiteracy, for example. But barriers related to a mental or physical incapacity for work or caretaking responsibilities created by children's disabilities are also common. To achieve TANF objectives and work requirements, welfare agencies need to address barriers, including those related to disability. Procedures for responding to these barriers vary from state to state and often from place to place within states (Wilkins 2003).

Supplemental Security Income

The Supplemental Security Income program also serves poor elderly individuals, but the focus of this article is on nonelderly individuals, including children with disabilities.

Eligibility and Benefits. The SSI program provides a nationally uniform maximum benefit, known as the federal benefit rate, which is adjusted annually for inflation. The monthly federal benefit rate in 2004 is $564 for a single individual and $846 for a couple. SSI is intended to be a resource of last resort. Accordingly, payments are reduced if an individual or a couple has earnings or other income and depend as well on a person's living arrangements. In about half of the states, the federal SSI benefit is augmented by a state supplemental payment. SSI beneficiaries are also immediately eligible for Medicaid in most states and, if they live independently, for food stamps.²

To be eligible, SSI disability applicants must pass a financial and a disability test. Financial eligibility requires that net income (whether from work or other sources) be less than the current federal benefit rate. Certain income exclusions are applied to the calculation of net income. The SSI rules exclude the first $20 of unearned income, the first $65 of earned income (or up to $85 of earned income if the applicant or recipient does not have $20 of unearned income), and half of any additional earnings. Generally, resources cannot exceed $2,000 for an individual and $3,000 for a couple, but equity in a home and most automobiles as well as a number of other resources are not counted. There is a complex set of rules regarding how resources other than cash are counted. The financial eligibility requirements for children generally pertain to the parents whose income is partially deemed to the child, but no TANF income is deemed to children.

The SSI disability test (for individuals aged 18 and older) is the same test used for Social Security Disability Insurance and is quite stringent.³ It requires that the applicant either be blind or have a physical or mental impairment that prevents engaging in any substantial gainful activity and that has lasted or is expected to last 12 months or to result in death. Substantial gainful activity is generally defined in terms of specific earnings thresholds and is currently (in 2004) set at $810 or more per month.⁴ The threshold of substantial gainful activity is automatically adjusted each year for changes in the average wage. Although TANF recipients may be affected by substance abuse, in 1996, Congress prohibited considering addiction or alcoholism as a significant factor in determining a person's disability status for the purposes of SSI eligibility.

The current law provides that a child must have a medically determinable impairment (or a combination of impairments) that results in "marked and severe functional limitations." This standard was introduced by PRWORA and is more stringent than the test that was previously applied.⁵ PRWORA also required the Social Security Administration to reassess, using the new standard, the disability status of children who might have been found disabled under former policies that the new law removed. Therefore, SSA redetermined the eligibility of approximately 288,000 of the almost 1 million children who were on SSI at the time of the enactment of PRWORA. As of April 2002, approximately 101,000 had been found to be no longer eligible under the new standard (Committee on Ways and Means 2004, 3–43). The legislation also required that child cases in the SSI program be reviewed against the adult disability criteria when children reach the age of 18. As of August 1999, approximately 45 percent of child SSI recipients who attained the age of 18 were found to be no longer disabled upon initial redetermination under the adult standard, as required under PRWORA (Rogowski and others 2002, xiv–xv).⁶

The difficulty and time that it can take to meet the eligibility standards of SSI affect the relationship of that program to TANF. Just 41 percent of adult SSI applications and 43 percent of SSI applications made on behalf of children in 2000 were ultimately approved.⁷ For adults, roughly 30 percent of the claims that are eventually approved go through the appeals process; about 10 percent of child awards are the result of appeals.⁸ The multiple steps of the application and appeals processes can be quite time consuming. On average it takes more than 3 months for initial disability claims to be processed, and appeals take almost a year; some successful efforts take much longer.⁹

Efforts Toward Employment. The orientation toward work in the SSI program differs markedly from that in TANF. The decision that an applicant is incapable of significant employment is central to the SSI eligibility determination process. Unlike TANF policy in many states, SSI policy does not promote diversion of applicants to alternatives, nor does it require that recipients work.¹⁰ Somewhat paradoxically, once applicants have succeeded in gaining SSI benefits on the basis of their incapacity for substantial gainful activity, they are presented with several incentives for voluntary employment.¹¹ One incentive is the income-disregard policies already discussed. Another incentive is the provision of special cash benefits to recipients whose earnings exceed the threshold of substantial gainful activity and the provision of extended Medicaid eligibility for individuals whose earnings preclude SSI payments but who lack ability to pay for medical care.¹² The newest work incentive in the SSI program is the Ticket to Work program, which provides vouchers for voluntary rehabilitation services after the applicant has been awarded benefits. The SSI work incentives do not, however, seem to affect most recipients, as only about 8 percent of SSI recipients aged 18 to 65 work (SSA 2002c, 18), and few cases close for reasons related to employment.

Disabilities in the TANF Population

There is no national standard or requirement for administrative tabulation of disabilities among TANF recipients. As a result, the available information on disability prevalence is mostly derived from national household surveys. When possible, general survey information may be supplemented with information collected idiosyncratically for particular jurisdictions. The accumulation of evidence suggests that by the late 1990s between 35 percent and 44 percent of TANF families had disabilities or health-related limitations, even though most may not meet the SSI program definition of disability. These reports and others also find that, while both physical and mental impairments have been found prevalent in the TANF caseloads, mental impairments are more common (Danziger, Kalil, and Anderson 2000). Lennon, Blome, and English (2002) have summarized many such studies.

Generally, studies of disability among recipients of the Temporary Assistance to Needy Families program rely on the Survey of Income and Program Participation (SIPP), the Current Population Survey, or the National Survey of America's Families, which use a variety of disability definitions. Disability is typically gauged by respondents' self-reports of a serious health problem or of an impairment that serves as a barrier to employment. For purposes of reporting the prevalence of disability in the TANF population, this article uses a concept of disability that includes any physical or mental impairment that impedes, directly or indirectly, the ability of an adult to be self-supporting through work. Some conditions may have only minor effects on employability; others may substantially reduce potential earnings and impede movement to self-sufficiency but nevertheless be insufficiently severe to meet the SSI program standard of inability to engage in substantial gainful activity. Thus, the definition of disability used in the studies discussed below is much broader than the strict SSI disability standard.

Additionally, it is important to note that these studies report disability prevalence at different times. Because of the substantial turnover in TANF recipients and the dramatic decline in the caseload from 1994 through 2001, studies at different times cover different populations.

Prevalence

The results of selected studies of disability among TANF recipients appear in Table 1. The most comprehensive use of data from the 1996 SIPP panel is presented in the study by Lee and others (2004). The authors estimate that in 1997 slightly more than one-third (35.1 percent) of all single mothers who were TANF recipients reported a disability that satisfied the Census Bureau criteria established for SIPP use.¹³ This was 15 percentage points greater than the rate reported for other low-income single mothers, and almost three times the rate (12.1 percent) that was reported for families with higher incomes (Table 1). Disability rates are similar for married and unmarried TANF recipients, but the gap in the prevalence of disability is greater between married TANF recipients and their married low-income and nonrecipient counterparts.

Lee and others (2004) focused on 1997. The General Accounting Office study (2001a) analyzed the outcomes for the same panel of households in 1999 and found that 44 percent of all adult recipients reported impairments in 1999 that met the Census Bureau's standard. This percentage is higher than that reported for 1997 by Lee and others (35 percent), but the GAO study includes the relatively small number of adult male recipients, so it is not clear that the difference reflects actual change (Table 1). A subsequent GAO study (2002a) found that 37 percent of adult recipients aged 18 to 62 reported having impairments at both interviews (1997 and 1999) and compared this percentage with a prevalence of 12 percent in the non-TANF adult population (Table 1). This comparison is flawed, however, because the GAO definition of TANF recipient included persons who at any time during the interval from 1997 to 1999 lived in a household receiving TANF assistance, and the agency excluded altogether from enumeration adults who were disabled at only one of the two interviews. This exclusion amounted to 12 percent of all adults aged 18 to 62 in the sample (GAO 2002a, 23) and inflated the estimate of prevalence of long-term disability.¹⁴

A series of studies at the Urban Institute have employed data from the National Survey of America's Families to examine barriers to employment among TANF recipients (Zedlewski and Alderson 2001; Zedlewski 2003). The Urban Institute's tabulations refer only to barriers created by physical and mental health problems but not to those created by other physical impairments, so the studies underestimate the prevalence of all impairments. Nonetheless, the results are important because they are based on a consistent set of questions administered in the same way to a representative household sample at three points in time. The analysts of the National Survey of America's Families find that the proportion of adults in families currently receiving TANF for whom "very poor mental or physical health" was reported to interviewers was consistently in the range of 32 percent to 36 percent during the surveys of 1997, 1999, and 2002. The differences across years were not statistically significant (Table 1).

Studies that are focused on more narrowly defined recipient populations, especially those drawn from inner-city areas of intense poverty, generally find even higher rates of disability than those uncovered in general surveys of TANF recipients (Table 1) (compare with Danziger, Kalil, and Anderson 2000; see also Cherlin and colleagues 2002). As part of the Project on Devolution and Urban Change, MDRC (formerly known as the Manpower Demonstration Research Corporation) conducted in-home survey interviews and in-depth ethnographic interviews with current or former welfare recipients in Cleveland (Cuyahoga County), Los Angeles (Los Angeles County), Miami (Dade County), and Philadelphia in 1998 and 1999 (Polit, London, and Martinez 2001). The results are stark: "Compared with national samples, women in the Urban Change survey sample had substantially higher rates of personal health and mental health problems and children's health problems. . . . The ethnographic data suggest that the survey data do not fully capture the severity of the health-related hardships the families faced" (Polit, London, and Martinez 2001, ES-2 [emphasis in original]).

Analyses of children's disabilities using the SIPP, the National Survey of America's Families, and other data sets uncover substantial prevalence of disabilities among children in TANF families. Lee and others found that in 1997 about one in five single mothers receiving TANF benefits had a child (or children) with a disability that met the Census Bureau standard (Table 1), and about one in eight had a child with a severe disability. Studies that focused on health reveal similar issues: Using the 1998 National Health Interview Survey, Wise and others (2002) found that approximately one-fourth of all children among TANF recipient families had a chronic illness. This proportion was significantly higher than that calculated for low-income families who were not TANF recipients (Table 1).

We conclude that the prevalence of disability among TANF recipients is substantial; a conservative estimate is that, by the late 1990s, one-third of all adult recipients would report in a survey a disability that was sufficiently severe to satisfy the SIPP criteria. Similarly, at least one out of four families receiving TANF benefits included a child with a significant health or physical disability. Comparison of studies is complicated by variations in definitions and samples as well as methodological problems, but the result is robust.

Changes Over Time

By definition, point-in-time estimates of the prevalence of disability among TANF recipients miss trends. Between 1994 and 2002, the number of families in the AFDC or TANF programs fell by almost 60 percent. Historically, AFDC termination rates were lower for families with disabilities than they were for others (see, for example, Bane and Ellwood 1994, 44). If the contraction was accomplished by moving the most employable into jobs, then the prevalence of barriers among the families still on the rolls should have increased.

Currently the only methodologically consistent estimates of the prevalence of disabilities among adult TANF recipients at multiple points in time are provided by the National Survey of America's Families from 1997, 1999, and 2002 (Table 1). These surveys reveal no significant change in the proportion of adult TANF recipients who report poor physical health or are evaluated as having very poor mental health.

This outcome is surprising. It implies that, whatever the process is that has led to caseload contraction (reduced rates of entry, increased rates of case closure, or both), the combined consequences for families with disabilities has been roughly the same as for those without, and this has kept the share of the caseload attributable to such families constant. In light of the historical record of lower termination rates for families with disabilities, this stability implies that welfare reform has accelerated their departure relative to the rates observed for other families or has reduced the likelihood that they come onto welfare at all—otherwise, the proportion would change. Just what is in fact happening is an important matter for research.

Changed Environment of Welfare

When compared with the operation of AFDC, the 1996 welfare legislation altered the terms of the federal and state fiscal relationship, expanded the range of discretion in program design, and imposed new requirements for program operation. Each of these changes potentially affected families with disabilities, either directly or indirectly, by influencing the way states treat such cases. This section examines how the structure of welfare changed in ways that have potential consequences for people with disabilities and discusses changes that specifically affect people with disabilities.

Fiscal Relationship

Aid to Families with Dependent Children was a matching-grant program in which the federal government paid at least half of all costs (the actual share depended on state per capita income), and the federal contribution increased automatically when state expenditures increased. Assistance was, from the states' perspective, cheap in comparison with aid programs funded from state revenues alone. Temporary Assistance for Needy Families is a fixed block grant to the states, the total amount of which is somewhat greater than federal spending in the last year of the old system. However, under TANF, whether state expenditures increase or decline, federal aid does not change.

Although the Personal Responsibility and Work Opportunity Reconciliation Act increased state resources, it created a powerful incentive to reduce outlays. Unlike the situation with a matching grant, under the new regime every dollar saved at the state level released a dollar for other uses, including tax reduction. This incentive was in part offset by a maintenance-of-effort requirement that penalized states for reducing their own social assistance outlays by more than 25 percent of levels established in 1994. However, under the law's provisions states could shift expenditures from payment of benefits to funding of other supportive services, not only for current TANF recipients but for persons considered at risk of applying.

The change to the block grant formula increased the incentives for states to encourage movement by adults and children to the SSI program. In states with the minimum federal AFDC matching rate, the payoff to such transitions under the block grant was twice as great as had been the case under AFDC—a $1 reduction in benefits paid now saves the state $1, not $0.50. For states with the highest match (roughly 75 percent federal funds), the switch multiplied the payoff by a factor of 4.¹⁵

Participation and Activity Requirements

The Personal Responsibility and Work Opportunity Reconciliation Act called for states to achieve certain levels of recipient involvement in work activities. This requirement was cast in terms of a specified minimum percentage of all adults who were participating at least a minimum number of hours each week in a set of work and work-related activities designated by the law. Both the definition of participation (percentage of adults) and the required participation rate (hours per week) were to escalate over time. In the first year of operation of the TANF program (fiscal year 1997), at least 25 percent of adults were to be participating in approved categories of activities for at least 20 hours per week. By the last year, 2002, the standard was 50 percent of adults engaged at least 30 hours per week. Higher minimum hour and participation rates were set for two-parent families. States failing to meet these standards faced a financial penalty. Unlike AFDC, PRWORA included no blanket exemptions from the participation requirements for persons with disabilities.

At the time PRWORA passed, many states had already obtained federal waivers to increase work requirements and sanctions for noncompliance. However, actual participation of recipients in work activities fostered by these demonstrations was generally much lower than that required by PRWORA, and virtually all state demonstrations allowed substantial exemptions for persons with disabilities and for persons caring for people with disabilities.

When compared with levels of participation in welfare-to-work programs operated before PRWORA, the target participation standards set by the new legislation were exceptionally ambitious. However, the legislation created an "out": states were allowed to reduce annual targets for participation percentage-for-percentage by any caseload reductions achieved between 1995 and each of the 6 years of the PRWORA horizon (1997–2002). (The caseload reduction credit excluded changes in caseload brought about by any alteration in federal or state law.) Moreover, states were permitted to count recipients who were working in unsubsidized jobs as meeting the participation requirements. Thus, while the participation and activity standards created some incentives for active creation of employment and training activities for TANF recipients, pressure to do so would be diminished insofar as the caseload contracted and some recipients combined welfare and unsubsidized work.

Time Limits

The new law placed a time limit of 5 years on the use of federal funds for cash assistance for adults. However, the law allowed states to exempt 20 percent of the caseload from the federal funding time limit and to use state-funded separate state programs to avoid the time limit in some other cases. Money spent on such programs counts toward satisfying the federal maintenance-of-effort requirement.

Adoption of time limits was a major development in state regulations post-PRWORA and one with particular importance for people with significant barriers to employment. Although most states took a cue from PRWORA's funding restrictions and set time limits at 60 months, many developed alternatives that were either more or less generous.¹⁶ The less generous alternatives generally involve a shorter time limit (for example, only 21 months in Connecticut); those that were more generous either continued families as child-only cases after the 60^th month or shifted them to separate state programs. Two states, Michigan and Vermont, did not impose any time limits at all.

Accommodating People with Disabilities

Many states make efforts in their policies and practices to accommodate people with disabilities. These efforts generally fall into two categories: providing additional assistance or exempting them from requirements. To do either, states must first screen applicants and recipients for disabilities.

Screening. Most county TANF administrators report that they conduct some sort of screening and assessment for disabilities or impairments but that the methods that are used and the point in the TANF process when they are applied vary (Kramer 2001; Thompson, Van Ness, and O'Brien 2001). Although it is clear that some agencies are careful to identify and respond to disability problems among TANF applicants and recipients (for examples, see Sherman 2003), no surveys have been conducted that are suitable for investigating how common are screening and case management for families with disabilities across all local TANF offices.

A study of procedures, very general in scope, was produced in 2001 by the General Accounting Office (GAO 2001a), which surveyed 600 county welfare administrators from 100 of the largest counties in the country plus 500 randomly selected counties. The GAO found that "about half" of the surveyed county administrators could not tell how many cases within their jurisdictions had disability issues or what the problems were (GAO 2001a, 21). An earlier GAO study of nine states found that such data were available for only two (GAO 2001b). This information suggests that procedures for evaluating disabilities may not be the focus of executive management. The data from the GAO survey of county administrators are not weighted by population or caseload size for the counties involved, so it is not possible to judge whether the lack of information was primarily a feature of small county operations or was common even in major jurisdictions.

Accommodation Strategies. Many TANF agencies use contractors to conduct disability assessments as well as to provide supportive services. One common element among the strategies that have been considered successful is intensive, individualized case management often made possible by the linkages with other providers. Several reports cite the benefits of case management and other approaches (Kramer 1999, 2001; Thompson and others 2000; GAO 2001a).

Case studies completed by the Urban Institute in 2001 focused on county administrators recognized as having innovative approaches to serving disabled TANF recipients (Thompson, Van Ness, and O'Brien 2001). County social services agencies with such leadership maintained the focus on moving people into work, and only two of the six sites visited were located in states that exempt disabled recipients from work requirements. However, staff in such agencies were found to have wide discretion to place recipients in activities that do not count toward the state's work participation rate (such as a mental health program). The report suggests that proposals to change the level of available TANF funds or to change the work participation rules should be approached with caution so that the flexibility used to serve disabled recipients is not lost.

Exemptions. The consequences of TANF work and time-limit policies for families with disabilities depend upon state policies with respect to accommodation and exemption.¹⁷ Accommodation cannot be evaluated only by looking at regulations, but regulations do provide information on the extent to which the presence of disabilities leads to exemption from work requirements and time limits. The tabulation of exemption procedures that are included in state TANF manuals (see Rowe and Russell 2004) indicates that by 2002, 33 states formally exempted heads of TANF families from work requirements if they are ill or incapacitated, and 34 states exempted heads if they are caring for an ill or incapacitated person.

The GAO survey (GAO 2001a) asked administrators about their practice of granting exemptions from work requirements and time limits. Thirty-six percent reported that they exempt TANF recipients with impairments from work requirements and from state time limits, but 27 percent reported exempting such recipients only from the work requirements. About one-third reported exempting recipients caring for a disabled child from the time limit and the work requirement. Twenty-eight percent exempt such caregivers only from the work requirement.

The significance of exemptions from work requirements is diminished, of course, if the state has a time limit on assistance and if exemption from work requirements does not stop the clock by exempting the recipient from the time limit. States generally have latitude to help large numbers of recipients, including recipients with disabilities, avoid being subject to time limits. Under TANF rules, states are allowed to continue the use of federal funds after 5 years for up to 20 percent of all cases that are the result of various hardships including disabilities. By 2002, 39 percent of all TANF families nationwide were child-only cases for which the time limit on federal funding is inapplicable (DHHS 2003a, xii). Since the allowable exclusion is based on all cases, states could potentially (and on average) exempt one-third of all cases with adults from time limits on the grounds of disability and other hardships (20 percent of all cases is one-third of the 61 percent that were not child only). In 2002, 30 states had formal extension provisions for time limits for cases in which adults are disabled or caring for a disabled family member (Bloom and others 2002, 16). Regulations in all but one of the remaining states provided for extensions on a case-by-case basis (Bloom and others 2002, Table A.7). Additionally, four states specifically included persons with disabilities in the separate programs, which are not under federal time limits (GAO 2002b).

Effects on Employment and Receipt of Benefits

The Personal Responsibility and Work Opportunity Reconciliation Act is intended to promote the employment of welfare recipients through the application of incentives and penalties. After the previous discussion of key policy changes of particular importance to persons with disabilities, this section looks at two key outcomes: employment and sanctions. How do TANF recipients with disabilities fare compared with those without disabilities?

Employment of Disabled TANF Recipients

Most surveys have too few observations on TANF recipients with disabilities to support comparative analysis of characteristics of families that report disabilities and those of families that do not. Nonetheless, it is common to find that rates of employment are lower among TANF recipients reporting disabilities than among those not reporting disabilities. Kim (2000) used the Current Population Survey data from 1998 to examine employment outcomes at an early stage of welfare reform. She found that a disabled TANF recipient was 21.1 percent less likely than a nondisabled recipient to be employed, holding other factors constant. GAO's study (2001a, 11) examined work among those currently on the rolls. The study found that "20 percent of TANF recipients with impairments were working full- or part-time in 1999, compared with 44 percent of TANF recipients reporting no impairments [who were working]." The effect of mental health on employment may be particularly strong, which is indicated by Jayakody and Stauffer's (2000) regression analysis showing that poor single mothers are 25 percent less likely to be employed if they suffer from a psychiatric disorder.

State-level findings are similar. For example, Danziger and others (1999) used 1997 survey data to examine work outcomes of TANF recipients in Michigan, a state with very generous financial incentives for work and no time limit. They find statistically significant differences between those with health barriers and those with none. Of those not suffering from a major depressive disorder, 61.2 percent were working at least 20 hours per week, while the percentage for those with major depressive disorders was only 48 percent. Those mothers with no health problems worked at least 20 hours per week in 61.7 percent of cases, while the percentage for those mothers with health problems was only 37 percent. Further, the likelihood of working was even much less for those affected by multiple barriers to work. Likewise, Danziger, Kalil, and Anderson's (2000) study of recipients with barriers to work found those with physical and mental impairments were even less likely to work if their impairments were compounded with other barriers. Similar findings were made by Rangarajan and Johnson's (2002) survey of recipients at 40 months after entering New Jersey's TANF program. It must be noted, however, that even for persons with disabilities who have never had a TANF connection, employment results have been dismal during the past decade (Burkhauser, Daly, and Houtenville 2001).

Sanctions

The emphasis on the obligation of recipients to work has been a keynote of state welfare reforms. The obligation has been enforced by reducing or eliminating benefits for families with adults who do not comply with program requirements. Disabilities may raise the likelihood that families fail to meet various requirements for sustaining TANF eligibility once it is achieved. Such failures generally result in sanctions that reduce benefits and can ultimately result in the loss of eligibility for the entire family in some states, until the family is again in compliance with program rules. Thus, the incidence of sanctions and the circumstances of noncompliance with work requirements are of special interest for those interested in the role of TANF in supporting families with disabilities.

According to the GAO survey of county welfare administrators, the majority (about 60 percent) attempt to notify and warn a recipient household at least twice that they are not complying with TANF rules and face a sanction (GAO 2001a). In the states with time limits shorter than the federal 5-year funding maximum, almost 75 percent of counties reported that recipients who are nearing the time limit are evaluated for disabilities or impairments. The GAO also found that about two-thirds of the counties surveyed reported that recipients not conforming to work requirements are evaluated for disabilities or impairments that might prevent them from working (GAO 2001a).

Although there is some evidence that the rates of sanctioning have declined since the 1990s, sanctions are regularly applied in most states. In 2000, the General Accounting Office reported on the application of sanctions in an average month in 1998 across several states (GAO 2000). In the states providing data on all sanctions, 5.1 percent of TANF families were subject to sanctions in an average month in 1998. In those states providing data on full-family sanctions (which are a complete loss of TANF benefits for a household, at least temporarily), only 0.9 percent of TANF families were subject to full-family sanctions during an average month in that year (GAO 2000, 29). More than half of the full-family sanctions were due to a failure to comply with work requirements.

Despite indications of administrative attention to the problem of disability-related noncompliance, there is evidence that households facing disabilities or impairments are sanctioned more frequently than others. Goldberg and Schott (2000) cite several studies showing that health problems and disabilities are a major cause of the noncompliance that results in TANF sanctions. A Utah study, for example, found that more than one-third of the state's sanctions resulted from physical health problems that prevented recipients from working, while one-fifth resulted from mental health problems. More recently, Goldberg (2002) reviewed data that suggest that certain states are disproportionately sanctioning recipients who have disabilities or face other barriers.

Other research supports the above findings cited. For example, the GAO (2000) has cited the Utah report as well as other studies of sanctioned families conducted early in the welfare reform period in Iowa, Michigan, and Minnesota. Each found greater prevalence of health problems in sanctioned families than in the caseloads in general. In 2001, the ongoing research by Cherlin and others (2001) resulted in a brief on sanctions in three cities that showed that more recipients who were sanctioned were in poor or fair health than were those not sanctioned.

A report issued by MDRC in 2001 (Polit, London, and Martinez 2001) that used surveys of more than 3,700 women and 171 in-depth interviews in four major urban areas found that the likelihood of sanctions seemed to increase with the number of health barriers that the family faced. Only 24.8 percent of those with no such barriers were sanctioned in the previous 12 months compared with 30.1 percent with one or two barriers and 33.6 percent with more than two barriers. These differences are statistically significant. It should be noted, however, that this MDRC study includes within its definition of health barriers several factors that are not considered health impairments in the other studies discussed and that are beyond the scope of this article.

Although these comparisons present many problems, it appears that despite the various accommodations made for welfare families affected by disabilities or health-related impairments in some states, recipients with disabilities are, overall, disproportionately sanctioned for noncompliance with TANF rules. This may imply that sanctions do not always hinge on a recipient's willingness to work, as Congress intended, but rather on a recipient's ability to do so. However, there may be an identification problem with this research insofar as disability is generally self-reported. It is possible that some recipients who fail to comply with an agency's obligations may feel, when queried, a social obligation to ascribe such failure to factors beyond their control, such as poor health or impairments.

Connections Between TANF and SSI

The changes in welfare policy in 1996 also changed the policy environment of the Supplemental Security Income program. From the beginning of the SSI program in 1972, SSI and Aid to Families with Dependent Children were linked by an overlap in the populations served and by the consequences of SSI receipt for AFDC eligibility and for state AFDC costs. With the SSI program, the federal government assumed financial responsibility for the needs of low-income disabled children. At the time of transfer, many of these children were already in families receiving support through AFDC. Simultaneously, some parents in AFDC families became themselves eligible for SSI benefits.

The financial connection arises from the difference in the financial role of the federal government in SSI and in AFDC and now TANF. AFDC was funded under a matching grant system in which state and federal governments shared responsibility for administrative and benefit costs. For all states, administrative costs were split 50/50; the federal contribution to benefits varied with state per capita income and ranged from 50 percent in the states with the highest incomes up to almost 90 percent in the poorest states. Regardless, in virtually all states, a move from AFDC to SSI saved states money, because the core SSI benefit is 100 percent federally funded and states were not obligated to increase their total contributions toward SSI supplements when the number of recipients increased. States promoted transfer and in some instances paid organizations to assist AFDC recipients in establishing SSI eligibility. This remains the case under TANF in at least some states. Practices in five states are described in Sherman (2003): for example, two states use third-party contractors to provide legal representation and assistance in the SSI (and Social Security Disability Insurance) application and appeals processes.

Similarly, recipients had a fiscal incentive to transfer from AFDC to SSI as well. The change could substantially increase benefits for the families involved. In general, SSI benefits exceeded amounts paid under AFDC. Between 1996 (in AFDC) and 2003 (in TANF), 7 states reduced the maximum nominal cash benefit available to families, 25 held cash benefits constant, and 19 increased benefits. In none of the 19 states did the increase in benefits match inflation. At the same time, the federal benefit rate for a single adult increased from $470 to $552, an increase of 17 percent. In Table 2, the change in the gain in constant dollars for a family of three by moving a child from AFDC–TANF to SSI is illustrated for the period of 1996 to 2002. Obviously, the incentives for making the transfer have grown.

However, it also made sense for poor families to apply for AFDC first and then, if possible, move family members to the SSI program. Aside from covering family members ineligible for SSI, the great advantage of AFDC for persons with disabilities was that eligibility could typically be determined much more quickly than what commonly occurred in SSI. Moreover, two-parent families made poor by a parent's disability were granted AFDC benefits under a special provision that was more lenient than the SSI rules.¹⁸

Although there is no empirical evidence on this point, it is likely that the influence of the relative generosity of the SSI program went beyond simply motivating applications. It may have also reduced the incentive that adult AFDC recipients had for undertaking efforts to work or rehabilitate while awaiting a decision on SSI eligibility, since applicants could fear that work in any job could be interpreted as evidence of an applicant's ability to sustain substantial gainful activity.

There is considerable anecdotal and empirical evidence of interaction between demand by families for SSI benefits and demand for AFDC. Garrett and Glied (2000) find that, at the beginning of the 1990s, the rate at which children in a state received SSI was inversely related to the size of state AFDC benefits and positively related to the amount of state supplementation of the federal benefit rate, even holding measures of family income constant. Similar results are reported by Kubick (1999). The implication is that when TANF benefits and requirements change relative to those of SSI, the demand for SSI will change as well.

TANF Overlap with SSI

Families with children can receive both SSI and TANF benefits, a situation that was true under AFDC as well. When the SSI beneficiary is a child, the family budget unit is generally constructed as if the child were not present. Indeed, the family could be eligible for TANF payments on behalf of the parents even if all the children were SSI recipients. When the SSI beneficiary is the parent, that adult is typically excluded from the family budget unit, and TANF payments are made on behalf of the child or children.

According to state data provided to the Department of Health and Human Services, by 2002, on average 13.5 percent of TANF cases were for households that also included a parent or other caretaker receiving SSI benefits, and 4.7 percent were living with at least one child SSI recipient (Table 3).¹⁹ The overlap between these groups is small: 17.1 percent of TANF cases—more than 1 in 6—were living with either an adult or a child SSI recipient (Table 4). Between fiscal years 2000 and 2002, the caseload fell, but the prevalence of SSI receipt in the households that included TANF recipients rose. On average for each month of fiscal year 2001, an estimated 100,000 TANF recipient families included at least one child on whose behalf the parent or caretaker was receiving SSI benefits (Table 4). In December 2001, 881,836 children were SSI recipients (SSA 2002b, 33). Thus, in any month in that year, about 11.3 percent of all SSI child recipients were in TANF families.²⁰

TANF to SSI Case Progression

A different perspective on the connection between the TANF and SSI programs is provided by examining the history of new SSI recipients. When adults or children apply for SSI, information on current sources of income is recorded, including information on receipt of means-tested benefits such as TANF (and AFDC before 1997). Each year the Social Security Administration creates a 10 percent sample of cases from all SSI cases in "active-pay" status. For children, the administrative data extracted include information on whether or not the recipients were receiving TANF or AFDC at the time of the application that commenced the current spell or at the time of any prior application on their behalf. For adults, an association with TANF or AFDC is also established if they lived at any time with a spouse or child who was receiving TANF or AFDC at the point of application. Although ideally the data would be confined only to the recipient's status at the point of the most recent application, it is unlikely that the difference between the ideal and what we have is important, especially for children.

Table 5 presents the results of looking at the connection of SSI with TANF from the perspective of SSI. Similar to the tabulations done with TANF data, these results reveal a significant connection between the TANF and SSI programs. Almost one-third of the women who recently applied for SSI (that is, within the past 2 years) and who were receiving SSI benefits in December 2002 had some association with AFDC or TANF. More reliably (because there is less chance that their current receipt of SSI benefits is not their first), slightly more than one-third of all children similarly defined had been at the time of an SSI application in families receiving TANF.

Conclusion

Despite differences of disability definitions and sampled populations, most analyses imply that at least one-third of all TANF recipient households include an adult or child with a disability. TANF outcomes for families with disabilities often appear worse than those for other TANF recipients. Sanctioning rates of families with disabilities exceed those for families without disabilities, and continuing poverty is more common among cases that close. Although most states have policies that seek to accommodate TANF recipients with disabilities, outcomes in terms of sanctions applied or employment obtained show that much still needs to be done.

There is considerable overlap between the TANF and SSI populations. About 17 percent of TANF households include someone receiving SSI benefits, and roughly one-third of nonelderly women and children who begin receiving SSI benefits were at the time of application in families receiving TANF. In effect, TANF serves as a second-order safety net for a portion of the disabled population whose disabilities are not severe enough to meet SSI standards or who are SSI recipients-in-waiting. Incentives for moving people from TANF to the SSI program are increasing.

The interests of families with children with disabilities are best served either by applying for SSI benefits if the disability appears sufficiently severe or, if not, by providing other supports. For adults with disabilities, TANF agencies need to perform a sort of triage. If the disability equals or exceeds the standard of substantial gainful activity, the objective should be to establish the applicant's eligibility and, once benefits begin, to adjust TANF strategy for other members of the family accordingly. If the disability clearly does not preclude work, the challenge is to assemble the services needed to support it. In between the sure-fire SSI and non-SSI cases falls a group of applicants and recipients for whom SSI eligibility is uncertain and the messages of the two programs conflict. Forgoing efforts at diversion and employment while awaiting an SSI award mutes the TANF work message, prolongs separation from the work force, and depletes years of eligibility permitted under TANF time limits in some states. The consequences are surely negative for those who are ultimately determined not to be SSI eligible, but they may also hamper the effectiveness of the work incentives now present in SSI. However, promoting diversion and work may lead to sanctions, hardship, and, in some cases, disability exacerbation. More needs to be learned about the procedures for conducting this process in ways that benefit the disabled, while encouraging and supporting employment for those with the potential to work. Experimentation with alternatives could benefit Social Security Administration policy and recipients of Temporary Assistance for Needy Families alike.