Changing safety net of last resort: downsizing general assistance for employable adults.

Anderson, Steven G. ; Halter, Anthony P. ; Gryzlak, Brian M. 等

The passage of the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996 (P.L. 104-193) has stimulated widespread concern about the viability of government income supports for low-income people. The act replaced the Aid to Families with Dependent Children (AFDC) program with state-designed and state-operated Temporary Assistance for Needy Families (TANF) programs and imposed strict work requirements and lifetime time limits on TANF receipt. It also tightened eligibility requirements for food stamp and Supplemental Security Income (SSI) recipients (Hagen, 1999).

Collectively, these changes have weakened the income supports for poor populations traditionally served by social workers (Hagen, 1999; Schneider & Netting, 1999). In addition, the devolution of program responsibility from the federal government to state and local levels raises questions about variations in services availability in a decentralized program environment (Peterson, 1995; Schneider & Netting). The decline of caseloads by over 40 percent since TANF was implemented, as well as wide differences in the percentage declines among states, has exacerbated these concerns (U.S. Department of Health and Human Services, 1999a).

General assistance (GA) programs represent an interesting precursor to the decentralized TANF program model. GA serves as the cash program of last resort for people who are poor (Lay, Lazere, Greenstein, & Gold, 1993). It consists of state and local welfare programs that provide cash and other benefits to individuals who are ineligible for TANF or SSI benefits (Piven & Cloward, 1993). Although GA recipients sometimes have been stereotyped as young, urban African American males, empirical research has shown considerable gender, racial, and age diversity in the GA population (Halter, 1996; Henly & Danziger, 1996). Furthermore, state GA programs serve a mix of families, disabled people, and individuals considered employable (Gallagher, Uccello, Pierce, & Reidy, 1999). Although national data on state GA caseload changes over time have not been published, program cutbacks in the early 1990s resulted in major GA caseload reductions in several states (Sylvester, 1992).

Because GA receives no federal funds, each state has complete discretion to design its program. This services model parallels TANF in its reliance on decentralized government decision making. GA programs, thus, can provide lessons about both services variability and common program features that have arisen in a decentralized income support system.

The adequacy of GA programs may critically affect the well-being of people who lose TANF and SSI eligibility (Uccello & Gallagher, 1997). Research has demonstrated that homelessness increases in the absence of GA programs (Burt, 1992; Coulton, Crowell, & Verma, 1992). Homelessness further diminishes employability (Halter, 1992); thus, prospects for self-sufficiency are compromised when GA programs are cut. In addition, because GA eligibility often is tied to eligibility for indigent medical care programs, the availability of GA may increase access to limited medical care for the poorest segments of the population (Gallagher et al., 1999). The importance of such program linkages for determining medical services eligibility has been shown by large declines in Medicaid caseloads after Medicaid eligibility was decoupled from TANF (Ellwood, 1999).

This study describes the characteristics of current State GA programs across important program dimensions: caseload size, geographic and target group coverage, work requirements, time limits, and administrative arrangements. Unlike earlier GA studies, we also analyze how these programs changed from 1989 to 1998. This allows us to assess how decentralized GA programs changed during the period in which national welfare reform debates culminated with the PRWORA of 1996, as well as the most important effects of such changes. These changes have important implications for social work practice in decentralized services systems.

Pre-1990s Program Background

GA experienced steady long-term growth from 1945 into the 1980s, with occasional cyclical downturns (Figure 1). Total GA caseloads rose from 244,000 in 1945 to 1.1 million in 1984 and then declined to 916,000 by 1989 (Social Security Administration, 1998). The characteristics of GA cases also have evolved over time. In particular, the average GA case size declined from 2.75 people per case in 1960 to 1.25 people by 1980 and has decreased slightly since that time (Social Security Administration, 1998).

According to an early study of GA programs, the average person on GA was a single adult living alone, equally likely to be male or female, and between 30 and 50 years of age (Burke, 1981). Similarly, later research by Henly and Danziger (1996) found both gender diversity and large numbers of people in middle-age groups among the Michigan GA population. This latter study also reported a high incidence of chronic health problems among GA recipients.

Several states began restructuring GA programs in the 1980s, and some of these changes foreshadowed the pre-TANF debates on welfare reform. For example, state program changes included time-limited cash assistance for individuals classified as employable, program elimination or benefit reductions for some recipient categories, and redefining work and other eligibility requirements (Ginsberg, 1998). Much like the benefits for AFDC, the real value of GA benefits also declined during this period. Between 1982 and 1989, the GA benefit for one person declined in 28 programs and increased in only seven.

A central controversy in these early GA program changes concerned. Whether GA recipients were actually employable. Critics contended that GA recipients were able to work and should be subject to stringent work requirements (Commonwealth of Pennsylvania, 1982). Others countered that many recipients had disabilities sufficient to exempt them from work (Halter, 1996). This debate was most dramatically illustrated in Pennsylvania, which in 1982 established two categories of assistance for GA: those chronically in need and those transitionally in need. Individuals with chronic needs were deemed unemployable, and therefore entitled to cash assistance. Individuals who needed only temporary assistance, on the other hand, were classified as employable and could receive only 90 days of GA cash benefits during a 12-month period (Halter, 1989; Katz, 1986). These changes set a tone for changes in other state GA programs.

Method

The decentralization of program authority to state and local governments complicates the analysis of GA programs, so that only broad classifications are possible. To measure program changes over the 1989 to 1998 period, we used secondary data analysis to aggregate information from four reports that provided state GA program descriptions at single points in time (Gallagher et al., 1999; Lewin & Bell Associates, 1990; Nichols, Dunlap, & Gallagher, 1992; Uccello, McCallum, & Gallagher, 1996). This allowed us to compare program dimensions across four points in time in the 10-year period--1989, 1992, 1996, and 1998.

The four reports provided extensive descriptions of GA program dimensions for each state. From the program dimensions described consistently in all four reports, we selected dimensions that have received attention in welfare reform debates and in evaluations of social welfare programs more generally (Gilbert & Terrell, 1998). These included geographic coverage, target group coverage, work requirements, time limits, and administrative arrangements. We used unpublished data files provided by the Social Security Administration to develop yearly state-by-state GA caseload trends over the period, which in turn were aggregated into national totals (U.S. Department of Health and Human Services, 1999b).

For each of the program dimensions selected, we developed classification schemes. Administrative arrangements were classified by whether state or local government units administered the program or whether they shared administrative responsibility. Geographic program coverage was classified by whether the state had statewide uniform coverage, statewide mandated coverage with local variations in programs, discretionary local programs, or no programs. States also were classified by whether or not they had coverage for three target groups: people considered disabled or unemployable, employable adults with children, and employable adults without children. Finally, for states having GA programs for employable adults, the existence of time limits and work requirements for recipients was determined.

Applying these classification schemes we categorized all states on each program dimension for each of the four years. This resulted in state tables for each program dimension, which showed how each state was classified in a given year and how these classifications changed over the four time points (1989, 1992, 1996, and 1998). The classifications from these state tables, in turn, were aggregated to provide national summary information.

Several methodological issues in this approach merit elaboration. First, although of obvious importance, we excluded benefit levels in this analysis because benefit levels vary considerably, depending on factors such as eligibility category, type of residence, or case size. This makes cross-state analysis of benefit level changes over time difficult. However, earlier analysis by Uccello and Gallagher (1997) indicated that GA benefit maximums for single individuals averaged only 40 percent of the poverty level in 1996.

Second, most states do not have uniform, statewide GA programs. Consequently, decisions had to be made concerning the unit of analysis for those not having uniform coverage. For example, some states administer and establish program rules for some counties but allow other counties to administer their own programs and set program rules. In these states, we used the portion of the state with state-administered coverage as the unit of analysis. For states that allow complete local discretion in establishing program criteria, information was derived from the largest county in the state. The geographic level of detail needed to support these analytic decisions was available consistently in the four reports used in this secondary data analysis.

Finally, GA was defined as ongoing cash assistance funded by state or local governments for longer than one month for poor people not eligible for TANF or SSI. We excluded one-time or sporadic payments, such as those provided through emergency needs programs, distributions of food or other commodities, and medical assistance. Although this definition resulted in some understatement of the total benefits that destitute people may be able to obtain, it provides a more accurate picture of continuous income supports.

General Assistance Caseload Trends

The data presented in this section for GA caseload trends are from annual reports provided to the federal government by the states and, thus, are subject to some state variations in interpreting GA programs. Because these data no longer are being collected, caseload data are available only through 1997.

General assistance caseloads dropped by 366,000 (40 percent) between 1989 and 1997, from about 916,000 to 550,000 (Figure 1). Only seven state caseloads increased during this period, with California the only large state among these (U.S. Department of Health and Human Services, 1999b). The caseloads declined after 1994 when the economy was strong, suggesting that economic factors are important in explaining these caseload reductions. However, large drops in state programs over brief time periods underscore the importance of program changes.

Caseload reductions in four states that dramatically curtailed eligibility for employable people accounted for most of the aggregate reduction. The combined caseloads of Illinois, Michigan, Ohio, and Pennsylvania decreased by about 345,000 people between 1989 and 1997. Each of these states experienced precipitous caseload declines over one- or two-year periods during the 1990s, with the greatest changes typically in the 1991 to 1993 period, during which states were experiencing fiscal crises (Sylvester, 1992). For example, Michigan's caseload declined from more than 120,000 in 1990 to about 17,000 in 1992, when the state eliminated GA for single adults considered employable.

With these major contractions of programs in the large Midwestern states during the 1990s, the remaining states with large GA caseloads in 1997 were New York (197,866), California (135,244), and Pennsylvania (65,875). These three states accounted for nearly 73 percent of all GA cases in 1997. No other state had a caseload of more than 19,000, and only eight other states had caseloads of at least 10,000 (IN, MA, MD, MI, MN, NJ, OH, and WA). California, New York, and Pennsylvania plus these eight states accounted for nearly 93 percent of the GA caseload in 1997.

State Comparisons across Program Dimensions

Coverage under GA can be examined from both geographic and target group perspectives. A beginning question is whether a state has any form of GA program, and, if so, whether coverage is available throughout the state. For states having programs, different eligibility criteria may result in differences in the characteristics of people covered. States also vary in the government arrangements used to administer GA programs.

Geographic Coverage

Ten states had no general assistance program in 1998, 33 states had programs with some level of statewide geographic coverage, and nine states had programs only in some localities (Table 1). All of the states without programs in 1998 were either in the south or had relatively small populations (AL, AR, LA, MS, OK, SC, TN, TX, WV, and WY)--as were most states that allowed but did not require localities to provide programs. Over the period studied, very little change occurred in this overall geographic pattern of coverage.

Although most states had some level of statewide coverage in each year studied, a subset of these states mandated programs but allowed local rule setting and thus did not have uniform GA programs. For example, although Indiana requires that each township provide "poor relief" for its citizens, the more than 1,000 townships have total discretion in establishing eligibility rules and program administration. Similarly, California allows individual counties to set benefit levels.

A complete elaboration of the relationships between TANF and GA programs is beyond the scope of this study. However, a review of GA program availability in states experiencing the most rapid TANF caseload reductions suggests reasons for concern. Of the 10 states reporting the highest percentage reductions since TANF implementation (CO, FL, GA, ID, LA, MS, SC, WI, WV, and WY), none had statewide GA programs for employable people and only Colorado had such coverage in some counties. Five of these states, which experienced TANF caseload reductions of 56 percent to 86 percent between August 1996 and June 1999, had no GA programs for any groups.

Changes in the Employable GA Population

State GA programs differ substantially in the target groups they serve. In 1998, 40 states provided some GA coverage for people considered disabled or unemployable, 17 assisted employable adults with children, and 14 assisted employable adults without children (Table 1). The most noticeable trend in groups covered was the dramatic decline in programs for people considered employable. Although the number of states offering coverage for disabled or unemployable people remained stable between 1989 and 1998, the number of states providing GA for employable adults with children decreased from 30 to 17. Similarly, state programs for employable adults without children declined from 27 to 14 during this period.

Some of these reductions were dramatic and accounted for much of the GA caseload decline. Illinois, Michigan, Ohio, and Pennsylvania all made program changes in the early 1990s that severely reduced or eliminated eligibility for individuals deemed employable. In some instances, these changes followed an incremental pattern in which program eligibility was first tightened and then eliminated. For example, Illinois began by instituting time limits on GA receipt and then implemented 6.8 percent benefit reductions in 1992 (Center on Social Welfare Policy and Law, 1992). By 1994 Illinois eliminated state funding for employable adults without children. Similarly, Michigan reduced benefits for single employable individuals in 1989 and then eliminated them from eligibility in 1991 (Sylvester, 1992).

Some recent reductions in GA eligibility resulted from state decisions to consolidate their GA family program components with TANF. Six states made such program consolidations during the 1996 to 1998 period (HI, KS, MI, MN, OH, and RI). Although this indicates that some people formerly covered by GA now receive coverage under TANF, these people also are now subject to the TANF time limits existing in those states. Consequently, the lack of income support coverage after TANF eligibility is exhausted remains an important issue.

Conditions of Benefit Receipt for GA Recipients

States also differed markedly regarding the conditions or limitations they attach to eligibility. Of particular interest, given current welfare reform initiatives, are work requirements and time limits. Normally, work requirements constitute either working for a nonprofit or public facility in return for cash assistance or participation in job training, job search, vocational or remedial education, or job readiness programs. More than three-fourths of the states that had GA programs for people considered employable in 1998 included work requirements as a condition of eligibility (Table 2). The number of states having GA programs with work requirements declined during the 10year study period. However, this decline resulted largely from the reduction in the number of states maintaining any GA programs for people considered employable.

The number of states with time limits increased from seven to 11 between 1989 and 1992 and then declined. Again, a principle reason for the decline in time limit policies since 1992 is that many states simply eliminated eligibility for people considered employable. In fact, the introduction of time limits sometimes preceded the elimination of assistance for employable individuals. For example, Connecticut, Idaho, Minnesota, Montana, Ohio, and Pennsylvania each initiated time limits before discontinuing eligibility for employable adults without children.

There is no consistency among states regarding the time limits, which typically are of two types. One approach is for a state to specify that people can only receive benefits for a certain number of months within a period, such as California's limit of five months of receipt in a year. A second option, similar to TANF, is to impose a lifetime limit on receipt of benefits. New Jersey, New York, and Utah imposed lifetime limits in 1998.

Finally, although conditions of receipt are generally more stringent for individuals considered employable, some states imposed eligibility restrictions for disabled or unemployable people. For example, California and Colorado require individuals to be tested for alcohol or drug abuse. This is consistent with a trend in SSI policy to restrict eligibility for people thought to have contributed to their disability through substance abuse (Karger & Stoesz, 1998).

Administration of GA Programs

Administrative responsibilities for GA programs vary considerably (Table 1). State offices assume all administrative responsibilities in some states, whereas local government offices are fully responsible in others. A third option is for states to include both state and local government units in administration. In this case, a state may give local government units the option of administering their own programs and otherwise provide state administration, or a state may distribute administrative functions between state and local authorities.

In 1998 states were evenly split among these three administrative options, with 16 states relying on state administration, 13 on state and local sharing of responsibility, and 11 on local administration. Over the 10-year period, there was a slight increase in the number of states exercising a shared state-local administrative option and a corresponding decrease in programs that relied totally on local administration.

Discussion

The findings demonstrate that, during a period when the federal safety net for low-income people was substantially reduced, the state-local GA system also was cut back and redefined. Social welfare cash programs traditionally have existed for three groups: families through AFDC and now TANF, elderly people and disabled individuals through SSI, and people not eligible for federally funded social welfare programs through GA. However, with the reduction of GA for employable adults in many states over the past 10 years, the remaining GA population increasingly consists of people with disabilities and unemployable individuals. This "new" GA population includes large numbers of disabled individuals who either are awaiting SSI eligibility decisions or have been denied SSI. Although GA thus continues as an important safety net program for people with disabilities, it offers little help for people affected by TANF and food stamp reductions.

The tighter restrictions on GA eligibility for people considered employable are consistent with TANF policy. In fact, some may view the implementation of more stringent work requirements and time limits under TANF as an expansion of welfare policies first tested under GA programs in the late 1980s and early 1990s. Because states have total discretion in the operation of GA programs, it was easier to institute program reductions in GA than in AFDC. Given that these policies are controversial, it probably was politically desirable for state officials to implement them first with a program heavily populated with single adults, who traditionally have generated less political support than families with children (Trattner, 1999).

The variation in GA programs shown in this analysis raises serious concerns about the consistency of support likely to be available to people who are poor in the devolved services systems emerging under TANF. State GA programs vary widely not only in basic coverage but also in their conditions of eligibility. Consequently, the extent to which GA serves as a safety net for people not eligible for other programs depends on the state, and in many cases the locality, in which they reside. This raises questions of equity and presents the possibility that potential recipients will be confused about eligibility requirements as they move between jurisdictions. Decentralization under TANF likely will result in substantial similar equity issues and program confusion. Of particular concern are states without GA programs that have experienced rapid TANF caseload decreases.

The reductions in GA eligibility for people considered employable brings to greater prominence the question of how employability is determined. Although we have followed the descriptions of states in classifying programs as serving unemployable and employable individuals, the specific criteria used to define employability are open to debate. Historically, the definition has centered on whether one is "able-bodied" (Axinn & Levin, 1997; Katz, 1986; Trattner, 1999), but this determination often is not clear-cut. More fundamentally, the use of able-bodied criteria in the postindustrial economy is suspect. Jobs for which low-skilled people must compete increasingly involve substantial cognitive skills (Holzer, 1996). Thus, using able-bodied criteria is likely to screen many people with poor work prospects from GA eligibility. The importance of developing realistic employability definitions extends to TANF as well, given that up to 20 percent of the caseload may be exempt from TANF time limits.

The limited empirical evidence available from states that have restricted eligibility also casts doubt on the actual employability of people denied benefits. Danziger and Lehrman (1996) found that fewer than 25 percent of employable people discontinued from GA in Michigan maintained employment, and that among those who were working, few were able to elevate themselves above the poverty line. This reinforces a Pennsylvania study (Halter, 1989), which found that few individuals who left GA were able to find and keep jobs.

Implications for Social Work

Social work is distinguished by its long history of services provision and advocacy on behalf of people who are poor. The NASW Code of Ethics (2000) reinforces this orientation by calling on social workers to "advocate for changes in policy and legislation to improve social conditions in order to meet basic human needs and promote social justice" (p. 27). During a period in which income-support services are being transformed and reduced, social workers must play a critical role in safeguarding the interests of people who are traditionally served by GA, TANF, and other low-income support programs. This requires both improvements in general advocacy capabilities in the devolved policy environment and the articulation of specific program issues.

As other analysts have noted (Hagen, 1999; Schneider & Netting, 1999), the devolution of social services requires social workers to rethink advocacy and analysis strategies for assisting people who are poor. In the new decentralized services systems, advocacy infrastructures must be strengthened to monitor state and local programs and lobby for policy and program changes. For example, Schneider and Netting have argued for shifting NASW advocacy efforts from the national organization to the 50 state chapters, as well as investing more in the state-level presence of other social welfare advocacy agencies. State-level linkages between these social services organizations also are needed to coordinate advocacy efforts.

Similarly, social work educators and researchers should target more attention on state and local programming for low-income people. This may include disseminating research results to state legislators and program personnel, monitoring program policies and outcomes, and training students in program evaluation and advocacy methods. The establishment of the Influencing State Policy organization, which assists social work faculty and students in influencing state policy development, is a useful start in this direction.

Developing these advocacy infrastructures is a prerequisite for effective lobbying and program monitoring on behalf of people who are poor in the new decentralized program environment. Our findings suggest several substantive issues on which such state-level advocacy and further research should focus. First, social workers should continue to advocate for income assistance for people who are poor and without other means of support. The changes in GA programs over the past 10 years have further diminished the income safety net, just as time limits and work requirements under TANF promise to do so for broader populations. Social work advocates need to keep legislators informed about the effects of proposed program changes and the characteristics of the people affected, because decision making on program changes often occurs with only limited or stereotypic information.

Employment and training programs for people who have been eliminated from GA and other programs also should be developed. The cutbacks in GA support for employable people have placed many individuals in a catch-22 situation. That is, when they were eligible for GA, they were often subject to work and training requirements and hence may have received employment training. However, after being eliminated from GA on the assumption that they could work, they were no longer eligible for employment and training services (Axinn & Levin, 1997). Because many people cut off from assistance have difficulty obtaining and holding jobs (Danziger & Lehrman, 1996; Halter, 1989), social workers should advocate for quality employment and training programs for people with severe employment deficits.

Social workers should play an integral role in developing realistic definitions of employability, both in broad policy discussions and in local program settings (Axinn & Levin, 1997). Of particular concern is ensuring that people without real possibilities for meaningful work in the modern economy are not screened out of GA programs. More effective screening instruments to assist in these determinations would be a significant contribution. Developing accurate profiles of people who are screened out of GA programs because of perceived employability also would be useful in advocacy efforts. For example, profiles demonstrating that people with severe educational deficits or physical disabilities were screened out could be used to advocate for less stringent employability definitions, better screening tools, and improved employment and training programs.

Given the rapid changes occurring in social services programs, social workers must keep abreast of opportunities to link GA applicants to other programs. For example, most GA recipients traditionally have been ineligible for Medicaid because they did not meet the Medicaid categorical eligibility requirements. However, if GA programs increasingly serve children and families who lose eligibility for TANF, eligibility for Medicaid or the Children's Health Insurance Program should continue. Because recipients often do not understand these program linkages and lose Medicaid eligibility (Ellwood, 1999), it is incumbent on caseworkers to inform GA applicants about all services for which they may be eligible.

Finally, because individuals eliminated from GA are an unorganized and invisible population, longitudinal research that follows GA leavers is required. Research on employment patterns and the use of other services such as shelters, medical care, and emergency services may help establish the survival strategies used after exiting GA. This research would be an important complement to the many studies being conducted on TANF leavers, and the results could be important in advocating for further changes to the safety net of last resort.

[FIGURE 1 OMITTED]

Table 1

Characteristics of State General Assistance Programs: 1989-1998

 Number of States

Characteristic 1989 1992 1996

Type of program (a)

 Statewide uniform 24 22 22
 Statewide mandated with 9 10 11
 local variation
 Local discretion (programs 10 10 10
 allowed but not required)
 No program 9 8 9

Target group

 Disabled/unemployable adults 41 42 41
 Employable adults with children 30 26 23
 Employable adults without children 27 20 16

Administrative authority

 State 16 16 16
 State-local sharing 11 13 13
 Local 14 13 12

 Number of
 States

Characteristic 1998

Type of program (a)

 Statewide uniform 23
 Statewide mandated with 10
 local variation
 Local discretion (programs 9
 allowed but not required)
 No program 10

Target group

 Disabled/unemployable adults 40
 Employable adults with children 17
 Employable adults without children 14

Administrative authority

 State 16
 State-local sharing 13
 Local 11

(a)For 1989, 1996, and 1998, numbers do not total 50 states. For these
years, some states (California and Maryland, 1989; Colorado and
Nebraska, 1996 and 1998) had more than one type of program that differed
in scope of coverage. Thus, totals are as follows: 1989 (52); 1992 (50);
1996 (52); 1998 (52).
Table 2

Work Requirements and Time Limits in State General Assistance Programs
for People Classified as Employable: 1989-1998

 Number of States

Program Characteristics 1989 1992 1996

Have GA program for employable
 adults (with or without children) 30 29 25
 Time limits on assistance (a) 7 11 3
 Adults with children 7 5 1
 Adults without children 4 10 3
 Work requirements 25 24 20
Percentage of all states that have
 programs for employable adults 60 58 50
Percentage of states with program
 for employable adults (b)
 that have time limits 26 38 13
Percentage of states with programs
 for employable adults that have
 work requirements 83 83 80

 Number of
 States

Program Characteristics 1998

Have GA program for employable
 adults (with or without children) 19
 Time limits on assistance (a) 4
 Adults with children 2
 Adults without children 4
 Work requirements 15
Percentage of all states that have
 programs for employable adults 38
Percentage of states with program
 for employable adults (b)
 that have time limits 22
Percentage of states with programs
 for employable adults that have
 work requirements 79

(a)Information on duration of assistance was not available for the
following states: Colorado, Maryland, and Virginia (1989); Colorado
(1996 and 1998).

(b)Because information was not available, percentages were based on the
number of states with programs for employable adults with or without
children as follows: 1989 (27); 1992 (29); 1996 (24); 1998 (18).

Original manuscript received August 27, 1999

Final revision received February 2, 2000

Accepted April 26, 2000

References

Axinn, J., & Levin, H. (1997). Social welfare: A history of the American response to need. New York: Longman.

Burke, V. (1981). Welfare background paper: Description of selected major programs. Washington, DC: Congressional Research Service.

Burt, M. (1992). Over the edge. Washington, DC: Urban Institute.

Center on Social Welfare Policy and Law. (1992, September). No relief for the poor: 1992 state cutbacks in AFDC, GA and EAF (No. 167). Washington, DC: Author.

Commonwealth of Pennsylvania, Department of Public Welfare. (1982). Status report. Harrisburg, PA: Governor's Office.

Coulton, C., Crowell, L., & Verma, N. (1992). General assistance program reductions in Cuyahoga County: Summary report. Cleveland: Case Western Reserve University, Mandel School of Applied Social Sciences, Center for Urban Policy and Social Change.

Danziger, S., & Lehrman, J. (1996, March-April). How will welfare recipients fare in the labor market. Challenge, 30-35.

Ellwood, M. (1999). The Medicaid eligibility maze: Coverage expands, but enrollment problems persist (Occasional Paper No. 30). Washington, DC: Urban Institute.

Gallagher, L. J., Uccello, C. E., Pierce, A. B., & Reidy, E. B. (1999). State general assistance programs, 1998. Washington, DC: Urban Institute.

Gilbert, N., & Terrell, P. (1998). Dimensions of social welfare policy. Boston: Allyn & Bacon.

Ginsberg, L. (1998). Conservative social welfare policy: A description and analysis. Chicago: Nelson Hall.

Hagen, J. (1999). Public welfare and human services: New directions under TANF? Families in Society, 80, 78-90.

Halter, A. (1989). Welfare reform: One state's alternative. Journal of Sociology and Social Welfare, 16, 15 1-162.

Halter, A. (1992). Homeless in Philadelphia: A qualitative study of the impact of state welfare reform on individuals. Journal of Sociology and Social Welfare, 19, 7-20.

Halter, A. (1996). State welfare reform for the employable poor: The facts behind the assumptions. Social Work, 41, 106-110.

Henly, J., & Danziger, S. K. (1996). Confronting welfare stereotypes: Characteristics of general assistance recipients and postassistance employment. Social Work Research, 20, 217-227.

Holzer, H. J. (1996). What employers want: Job prospects for less-educated workers. New York: Russell Sage Foundation.

Karger, H., & Stoesz, D. (1998). American social welfare policy: A pluralist approach. New York: Longman.

Katz, M. (1986). In the shadow of the poorhouse: A social history of welfare in America. New York: Basic Books.

Lav, I., Lazere, E., Greenstein, R., & Gold, S. (1993). The states and the poor: Budget decisions hurt low income people in 1992. Washington, DC: Center on Budget and Policy Priorities.

Lewin/ICF and James Bell Associates. (1990). Characteristics of GA programs, 1989 (Prepared for Office of the Assistant Secretary for Planning and Evaluation). Washington, DC: U.S. Department of Health and Human Services.

National Association of Social Workers. (2000). Code of ethics of the National Association of Social Workers. Washington, DC: NASW Press.

Nichols, M., Dunlap, J., & Gallagher, L. J. (1992). National general assistance survey. Washington, DC: Urban Institute.

Personal Responsibility and Work Opportunity Reconciliation Act of 1996, P.L. 104-193, 110 Stat. 2105.

Peterson, P. (1995). State response to welfare reform: A race to the bottom? In I. V. Sawhill (Ed.), Welfare reform: An analysis of the issues (pp. 7-10) Washington, DC: Urban Institute.

Piven, F., & Cloward, R. (1993). Regulating the poor: The functions of public welfare. New York: Vintage Books.

Schneider, R., & Netting, F. (1999). Influencing social policy in a time of devolution: Upholding social work's great tradition. Social Work, 44, 349-357.

Social Security Administration, Office of Research, Evaluation, and Statistics. (1998). Annual statistical supplement to the Social Security Bulletin. Washington, DC: Author.

Sylvester, K. (1992, February). The war against the able-bodied poor. Governing, 24-26.

Trattner, W. (1999). From poor law to welfare state: A history of social welfare in America. New York: Free Press.

Uccello, C., & Gallagher, L. (1997). General assistance programs: The state-based part of the safety net. Washington, DC: Urban Institute.

Uccello, C. E., McCallum, H. R., & Gallagher, L. J. (1996). State general assistance programs 1996. Washington, DC: Urban Institute.

U.S. Department of Health and Human Services, Administration for Children and Families. (1999a). Changing welfare caseloads since enactment of new welfare law [Online]. Available: www.acf.dhhs.gov/news/stats/aug-sept.htm.

U.S. Department of Health and Human Services, Administration for Children and Families, Office of Planning, Research, and Evaluation. (1999b). Unpublished data files on state general assistance caseloads.

Steven G. Anderson, PhD, is assistant professor, School of Social Work, University of Illinois, Urbana-Champaign, 1207 West Oregon, Urbana, IL 61801; e-mail: sandersn@uiuc.edu. Anthony P. Halter, DSW, is associate professor, School of Social Work, University of Illinois, Urbana-Champaign, and Brian M. Gryzlak, MSW, is a graduate student, Department of Sociology and Anthropology, Western Illinois University, Macomb.