Social Welfare Policy and Public Assistance for Low-Income Substance Abusers: The Impact of 1996 Welfare Reform Legislation on the Economic Security of Former Supplemental Security Income Drug Addiction and Alcoholism Beneficiaries

Abstract

Prior to January 1, 1997, individuals with drug- or alcohol-related disabilities could qualify for federal public assistance through the Supplemental Security Income (SSI) program. During the welfare reforms of the Clinton administration, this policy was changed resulting in lost income and health care benefits for many low-income substance abusers. This paper examines the historical underpinnings to the elimination of drug addiction and alcoholism (DA&A) as qualifying impairments for SSI disability payments. Following this, empirical evidence is presented on the effect this policy change had on the subsequent economic security of former SSI DA&A beneficiaries. Findings indicate that study participants that lost SSI benefits suffered increased economic hardship following the policy change. These findings have important implications for future social welfare policymaking decisions.

The legitimacy of drug addiction and alcoholism as disabling impairments to occupational functioning has been a point of contention among politicians, physicians, and substance abuse treatment professionals and advocates for some time. The lack of a clear understanding of substance abuse coupled with many subjective interpretations of the phenomenon has made defining drug addiction and alcoholism problematic. Ambiguity related to the nature and consequences of drug addiction and alcoholism has made constructing social welfare policy and providing public assistance for drug addicts and alcoholics a difficult proposition. Philosophical and ideological divisions among political constituencies (e.g., politicians, recovering addicts and alcoholics, and substance abuse treatment professionals) have complicated efforts to address this issue. In 1974, with the initial implementation of Supplemental Security Income (SSI), the federal government attempted to assist low-income substance abusers by categorizing drug addiction and alcoholism (DA&A) as a disability. Under this public assistance program, low-income individuals dependent on drugs or alcohol and unable to work had the opportunity to claim their condition as a disability. Individuals with substantiated claims were mandated to substance abuse treatment, to having a representative payee, and to continued disability reviews in exchange for monthly cash assistance and health care benefits (i.e., Medicaid) (Social Security Amendments Act, 1972). Unfortunately, due to a variety of administrative and programmatic problems, as well as a shifting political environment, this attempt by the federal government to assist low-income substance abusers was relatively short-lived.

In the summer of 1996, Congress passed the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA; PL 104-193), more commonly referred to as “welfare reform,” as the culmination of a campaign by politicians to redefine the American welfare system. In March 1996, Congress preceded the PRWORA with passage of the Contract with America Advancement Act (PL 104-121). It was in this act, the Contract with America Advancement Act, that Congress mandated the elimination of drug addiction and alcoholism as eligibility categories for SSI disability benefits (Davies, Iams, & Rupp, 2000). When Congress eliminated the DA&A category, it estimated that 75% of former SSI DA&A beneficiaries (approximately 125,000 out of 166,666 recipients) would requalify for SSI benefits under a different disability category, such as chronic physical or mental illness (Watkins, Podus, Lombardi, & Burnam, 2001). Unfortunately, only 35% of beneficiaries were reclassified (Lewin & Westat, 1998). This left over 110,000 low-income substance abusers without monthly income or health care benefits.

In this paper a brief description of the historical, social, and political antecedents that influenced the elimination of substance abuse as a disability category is provided (see Hunt, 2000; and Hunt & Baumohl, 2003, for a more thorough description of this topic). Following this, a number of implementation issues that affected the saliency of the DA&A program are presented. Lastly, empirical evidence on the effects terminating the DA&A category had on the economic security of former SSI DA&A beneficiaries is examined.

Policy Change Antecedents

Social and Political Issues

The elimination of drug addiction and alcoholism as a category for SSI disability eligibility was influenced by a combination of social and political reactions. Moral interpretations of substance abuse called into question its legitimacy as a disability. Opponents of the benefit continued to argue that substance abuse was “self-inflicted,” more a characterological deficit than a disease or medical condition. Concerns related to drug addicts and alcoholics as worthy or deserving recipients of welfare benefits, and the legitimacy of drug addiction and alcoholism as a disability category, can be seen in the differential treatment DA&A beneficiaries were subjected to by policymakers and program administrators. First, DA&A beneficiaries were the only group receiving public assistance that were required to receive treatment or have their benefits suspended (Gresenz, Watkins, & Podus, 1998). This criterion did not apply to any other disabled group. Individuals suffering from chronic health or mental health problems were not required to seek treatment to avoid losing benefits. Second, DA&A beneficiaries were the only group of public assistance recipients mandated to have representative payees. While severely and persistently mentally ill beneficiaries may have representative payees, mentally ill beneficiaries are not mandated to use a third party (Rosen & Rosenheck, 1999). Finally, following passage of the Social Security Independence and Program Improvements Act of 1994 (PL 103-296), SSI benefits for DA&A recipients were limited to 36 months (Davies, et al., 2000). This implies, at the very least, that public officials did not consider drug addiction and alcoholism to be chronic disabilities. This sentiment was reinforced in May 1994 by the SSA Commissioner when she expressed to Congress that it was the expectation of DA&A recipients to take full responsibility for their recoveries and get off of the SSI disability payment rolls (Departments of Labor, 1994). This statement reflected the personal responsibility rhetoric being propagated by conservative politicians at that time.

Disapproving politicians and moral entrepreneurs expressed further concern for the DA&A program and its beneficiaries with claims of inappropriate use of cash benefits. Many felt that DA&A recipients used their benefits to buy drugs and alcohol (Satel & Glazer, 1993). According to Anderson and others (2002), many DA&A recipients could be characterized as “lying, swindling, drug addicts who were squandering taxpayers [sic] money on dope” (p. 266). Senator William Cohen (R-ME) was quoted in a number of prominent newspapers criticizing the DA&A program’s approach to helping low-income drug addicts and alcoholics. For example, in a 1994 New York Times article, he stated that “[h]undreds of millions of scarce Federal dollars” were being used by DA&A recipients to buy illegal drugs (Cohen, 1994a, p. 15). In the Washington Post, he stated that DA&A benefits were “often used to perpetuate addictions, rather than cure them, and the addicts and alcoholics ride along on a drug-laden train fueled by their continuing disabilities” (Cohen, 1994b, p. 17). Such comments were indicative of the political dissatisfaction with the DA&A program that existed among some elected officials during the early 1990s. While this position was not universally regarded by members of Congress, the bipartisan support it did receive would eventually compromise the political saliency of the DA&A program.

Disharmony among constituents of the DA&A program would escalate when it became evident that many substance abuse treatment and service providers were also dissatisfied with the DA&A program. Television programs broadcast stories with recovering alcoholics, shelter operators, and substance abuse treatment administrators and clinicians, all stating how SSI payments were harming drug addicts and alcoholics, and preventing effective treatment (Goldstein, Anderson, Schyb, & Swartz, 2000). A story on CBS’s 60 Minutes described how the SSA was providing drug addicts and alcoholics with public assistance for the explicit use of perpetuating “addictive lifestyles” (p. 217). NBC’s Dateline aired a similar segment not only criticizing DA&A beneficiaries for using public funds to purchase drugs and alcohol, but demonstrating opposition to the DA&A program from substance abuse treatment providers and individuals in recovery (Hunt, 2000). These mediated accounts of DA&A recipients spending their public assistance benefits on drugs and alcohol, and of members of the substance abuse treatment community criticizing a program specifically designed to help drug addicts and alcoholics, provided considerable ammunition for opponents of the DA&A program.

Implementation Issues

The greatest impediment to the successful implementation of the DA&A program was the administration of the substance abuse treatment and representative payee mandates. According to policy, DA&A beneficiaries were required to participate in a substance abuse treatment program (if appropriate treatment were available) and receive SSI payments through a representative payee (Davies, et al., 2000). To administer the treatment mandate of the program, the SSA developed a plan to create a system of independent state-contracted referral and monitoring agencies. These agencies were set up to refer DA&A beneficiaries to substance abuse treatment services, as well as monitor treatment compliance (Watkins, Wells, & McLellan, 1999). If beneficiaries failed to comply with the substance abuse treatment mandate, they would be subject to progressive sanctions up to, and including, loss of benefits (SSA, 1995). The trouble with the representative payee mandate was not whether a beneficiary had a representative payee; this requirement was handled administratively upon entry into the program. The problem that developed was more related to whom the benefit checks were being sent.

Treatment Mandate

One reason implementing the mandated substance abuse treatment requirement for DA&A recipients was so difficult was that appropriate resources were never provided. Congress never adequately compensated for the operational responsibilities that were assumed by the SSA in acquiring a new service population, nor were funds appropriated to treat DA&A beneficiaries (Hunt, 2000). By law, the SSA could not pay for treatment for beneficiaries; it had to rely on Medicaid and Medicare. Substance abuse treatment, however, was not a federally mandated Medicaid benefit. Furthermore, Medicaid coverage varied by state. While some states included some type of substance abuse treatment in their coverage, others did not. DA&A beneficiaries that resided in states that did not cover treatment were relegated to the public treatment system (e.g., county-funded substance abuse treatment services). This resulted in substantial barriers to treatment and great inconsistency in treatment provision across states. Beneficiaries in many rural areas found it difficult to even locate appropriate treatment services. Beneficiaries requiring methadone maintenance or dual diagnosis treatment services were further limited by a lack of such services in many areas (Hunt & Baumohl, 2003). The actual number of DA&A recipients that ever made it to treatment is unclear. Reports vary ranging from nine percent (Government Accounting Office, 1994) to a self-reported 47% (Office of the Inspector General [OIG], 1994b). The lack of available and appropriate substance abuse treatment resources compromised the intent of the original policy mandate.

Because the SSA had to defer the provision of treatment service requirements to existing public substance abuse treatment programs, the increased bureaucracy made effective implementation of the treatment mandate much more difficult. In theory, the referral and monitoring agencies (RMAs) were to administer the treatment mandate for the SSA. In reality, for most of the life of the DA&A program, most states did not have a functional RMA. The first RMAs chosen by the SSA were state vocational rehabilitation programs. When these agencies proved ineffective, Congress, as part of President Reagan’s Omnibus Budget Reconciliation Act of 1981, authorized contracts with private rehabilitation programs with the caveat that no rehabilitation program would be reimbursed for services until a beneficiary achieved nine continuous months of employment (Berkowitz, 1987). This mandate, which was reflective of the Reagan administration’s agenda to make the receipt of public assistance benefits more difficult, thereby reducing federal spending on such programs, discouraged most agencies from participating in the RMA system. As a result, by 1984, there were only 10 states with RMA contracts (Hunt & Baumohl, 2003). This created more problems for the SSA due to the fact that, if no RMA was available, then the regional SSA office had to perform the task. Since resources were not provided for such an undertaking, many DA&A beneficiaries went unmonitored. In 1992, the SSA issued a Request for Proposals with the intention of placing an RMA in all 50 states and the District of Columbia. Ironically, it would not be until the last year of the program, after its fate had been determined, that the SSA would have an RMA in every state (Hunt, 2000).

Representative Payee Mandate

Problems of implementation related to the representative payee mandate were less related to compliance than to propriety. Representative payees of DA&A beneficiaries were responsible for managing the recipients’ cash benefits so that funds were spent appropriately (e.g., housing, food, and clothing). Representative payees could be “a nonprofit social service agency, a governmental social service agency or public guardian, another organization, a family member or other interested person” approved by the SSA to act in the best interest of the beneficiary (SSA, 1996, p. 1). However, many representative payees were neither related, nor organizationally connected, to the beneficiary (OIG, 1994a). Some, rather, were bartenders, liquor store owners, and fellow DA&A recipients (Gresenz, et al., 1998). Political opponents of the DA&A program grasped on to this information using it to full advantage.

Another problem related to the representative payee system had to do with large lump-sum retroactive payments. Retroactive benefit payments to DA&A beneficiaries were the result of back funds accumulating during an applicant’s claim or appeal process. As a result, low-income substance abusers with substantiated disability claims would receive “back pay,” a disbursement that could range from a few hundred to a few thousand dollars (Satel, Reuter, Hartley, Rosenheck, & Mintz, 1997). Drug addicts and alcoholics that were able to acquire all of their retroactive benefit payment from inappropriate or irresponsible representative payees were placed in a very vulnerable position. Surplus cash often translated into prolonged episodes of drug use and the occasional drug overdose. Instances of misuse of public assistance funds to purchase alcohol and drugs by DA&A recipients received much attention from the media and conservative political opponents (Satel, et al., 1997). In one 1992 case, a heroin addict arrested for possession in Bakersfield, California, was found to have thousands of dollars in cash from retroactive SSI payments. A more damaging case, occurring at the same time and in the same city, concerned a man found dead from a drug overdose in a motel room following the purchase of a large amount of drugs with retroactive funds (Hunt, 2000). Such highly publicized, albeit rare, horror stories proved very damaging to the entire DA&A program.

Program Growth

A major problem affecting the implementation of DA&A program provisions was the unexpected and unprecedented growth the program experienced in the early 1990s. In 1990, there were approximately 24,000 SSI DA&A beneficiaries on the rolls; by 1996, that number had increased to nearly 170,000 (Gresenz, et al., 1998). This sudden explosion in the number of program participants compounded already existing SSA staffing, communication, and resource issues. Not only were SSA employees inexperienced and inadequately trained to deal with low-income drug addicts and alcoholics, now they were overwhelmed by the growing magnitude of the problem.

The sudden expansion of the DA&A program can be attributed to a combination of factors. The creation of more RMAs, the broadened interpretation of substance abuse as a disabling disorder resulting from a number of federal court cases, the shifting of GA recipients to SSI by some states and counties, and a moratorium on continuing disability reviews by disability reviewers in the mid 1980s for political reasons all contributed to the sudden growth of the program (Hunt & Baumohl, 2003). The placement of RMAs in more states across the country during the early 1990s increased the number of SSI applicants determined to have a drug addiction and alcoholism disability. SSA regional staffs were less hesitant about designating DA&A claimants as drug- or alcohol-disabled if they did not have to deal with the responsibility of referring and monitoring applicants. Hunt and Baumohl (2003) speculate that in states without RMAs, “disability-determination teams were less inclined to approve drug addicts and alcoholics in the absence of a way to ensure treatment compliance” (p. 28). This possibility indicates the significance having an RMA played on the number of applicants determined to have a DA&A disability in a particular region.

During the Reagan and Bush administrations, there were a number of federal court cases that challenged the government’s competence in determining drug addiction and alcoholism disability claims. A 1990 class action suit (Wilkerson v. Sullivan) brought against the Secretary of the Department of Health and Human Services asserted that disability claims based exclusively on severe alcoholism had been mishandled by the SSA resulting in a high rate of error in determining DA&A disability cases. The suit claimed “the Secretary had willfully violated the regulations by never allowing severe alcoholism alone to be a disabling impairment” (Hunt, 2000, pp. 57-58). The federal district court found for the plaintiffs but the decision was overturned by the 3^rd Circuit Court of Appeals with the advise that the Secretary impart on SSA and DDS staffs the importance of accurately identifying DA&A claimants (Hunt & Baumohl, 2003). The full impact this court decision had and the resulting pressures it put on Administration officials is unclear; however, over the following two years the number of DA&A beneficiaries doubled.

Motivated by the placement of more RMAs and the liberalizing of DA&A disability eligibility interpretations, many states and counties along with advocates for low-income substance abusers saw an opportunity to transfer their most severely affected clients from the state and county welfare rolls to the federal government’s (Schmidt, 1990). Moving clients from General Assistance (GA) to SSI was seen as a win-win proposition by state and county social service agencies. The state or county would no longer be financially responsible for the client, and the client would receive increased cash benefits, access to substance abuse treatment, health care coverage, and a representative payee to help manage the resources of the drug- or alcohol-disabled beneficiary. A newly burgeoning “troubled persons” industry also benefited as recovery homes and therapeutic communities gained access to a steady stream of federal dollars.

Due to a political controversy created by the Reagan administration’s attempt to reduce the number of individuals collecting disability benefits in the 1980s, for a period of approximately two years the SSA ceased all continuing disability reviews (CDRs) and denials of mental impairment claims (Berkowitz, 1987). A change in CDR criteria (i.e., the removal of a “medical improvement” provision requiring beneficiaries to have improved medically before they could be denied benefits) made termination of disability benefits easier. The Reagan administration utilized this change in review criteria to questionably terminate many disabled individuals, a large number of whom had severe mental impairments. Outraged disability reviewers refused to conduct CDRs. The resulting moratorium created a backlog of CDRs. Once the moratorium was lifted, the SSA was faced with an exorbitant number of overdue reviews. This created a negative exit to entry ratio that the Administration had a very difficult time resolving. DA&A beneficiaries that successfully completed treatment were often not re-evaluated, and continued to collect disability benefits. In an 18-month period between 1994 and 1995, RMAs referred 2182 cases to the SSA for a CDR following successful completion of a substance abuse treatment program. Only 32 cases were terminated (OIG, 1997).

Summary

Many factors contributed to the elimination of the DA&A program. Socially and politically, the program was an easy target for conservatives. Throughout the lifetime of the SSI DA&A program (1972–1996), the federal government was predominantly lead by conservative politicians intent on reducing the scope and magnitude of social welfare in this country. During the Clinton administration, conservative ideology attributing social ailments such as poverty, unemployment, and substance abuse to individual factors, and a campaign of welfare reform predicated on moral themes of personal responsibility, self-sufficiency, and hard work, defined a sociopolitical environment that was not conducive to providing public assistance to low-income drug addicts and alcoholics. Negative perceptions of the program provided by media horror stories and a lack of support from the substance abuse treatment community reinforced the conservative perspective that cash assistance to drug addicts and alcoholics only promoted and encouraged substance abuse. The division among political constituencies became more damaging when it became evident that the SSA had little enthusiasm to administer the program. The few proponents the program did have were overwhelmed by the opposition. Beneficiaries had no power, and many progressives did not want to appear “soft” on drugs in an election year.

Besides the social and political factors conspiring against the DA&A program, problems related to poor implementation were equally overwhelming. Much of this can be attributed to a lack of resources in the face of unprecedented program growth. However, even before the program started to grow so rapidly in the early 1990s, the SSA failed to effectively manage the treatment and representative payee mandates. The referral and monitoring system was severely inadequate. Few beneficiaries ever went to treatment, treatment effectiveness was not monitored, and continued disability reviews were not always conducted. Negative attitudes toward the DA&A program and its beneficiaries among service providers and SSA employees made implementation that much more difficult. With these social, political, and programmatic problems contributing to the instability of the DA&A program, its demise should not be surprising. The effects eliminating this program had on low-income substance abusers are important for understanding the implications of this policy change.

The Impact of Welfare Reform Legislation

The impact of welfare reform legislation on marginalized and disadvantaged populations requires special attention by policy researchers and analysts. Previous research on 1996 welfare reform outcomes has indicated that, while the number of welfare recipients has significantly declined, the quality of life of former public assistance (i.e., Temporary Assistance for Needy Families, Supplemental Security Income) beneficiaries has not necessarily improved. Many former public assistance beneficiaries that have been forced to leave the social welfare system as a result of newly-mandated time limits or reformed eligibility criteria have reported either sustained or increased economic and material hardship (Acs & Loprest, 2007; Norris, Scott, Speiglman, & Green, 2003). Welfare reforms that result in abrupt changes to established mechanisms of income maintenance and health care security among low-income individuals, such as the elimination of the SSI DA&A program, are expected to have an even greater effect on what must be considered a vulnerable population.

In the following study, income levels for a panel of 412 former SSI DA&A beneficiaries from Northern California were examined following elimination of the SSI DA&A category. Self-reported income data (i.e., amount and source of income) were gathered at baseline, as well as 6-, 12-, 18-, 24-, and 42-month follow-up interviews. These interviews began just prior to the policy change in December, 1996, and were concluded in November, 2000. Longitudinal income data were analyzed using mixed model techniques to determine changes in levels of economic support over time. A primary interest of the researchers was to compare longitudinal income outcomes of former SSI DA&A beneficiaries that were able to retain SSI income benefits under another disability category with those that lost benefits. The ultimate goal of this research was to determine the effects eliminating formal income maintenance benefits (i.e., SSI) had on the subsequent economic security of low-income substance abusers.

Methods

Study Sample

The study sample was selected from a population of SSI DA&A recipients residing in and around the San Francisco Bay Area as of December 1996. A sampling frame consisting of SSI DA&A beneficiaries from four Northern California counties (Alameda, San Francisco, San Joaquin, and Santa Clara) was developed by Maximus, the referral and monitoring agency responsible for tracking treatment placement and compliance among SSI DA&A beneficiaries in those areas. From this sampling frame, 775 individuals were randomly selected. Ultimately, 519 study participants would meet all inclusion criteria for the study and complete baseline surveys. At the 42-month follow-up, a final panel of 412 study participants was established. Of the potential 2472 interviews available over the lifetime of the study from this panel (412 study participants × 6 waves of data), only 30 follow-up interviews were missed. Dropout analysis of the 107 study participants excluded from the final panel did not differ significantly, with the exception of ethnicity. Forty-one percent of the study dropouts were Caucasian compared to 29% of the study’s final panel.

Variables

Income

This study was concerned with two types of income variables: 1) total income for each study participant at each interview, and 2) primary income source following termination of the DA&A category. Total income was calculated by adding all reported income sources in the 30 days prior to an interview. Income sources ranged from legitimate employment to formal mechanisms of income support (i.e., public assistance and social insurance) to informal sources, such as monies from friends and family, panhandling, gambling, prostitution, and other illegal activities. Primary income source reflected a study participant’s primary source of income over the 3.5 years following the policy change. In order to assess differences in the economic security of study participants that requalified for SSI benefits under another disability and study participants that lost SSI benefits, three primary income groups were constructed. The first primary income group was SSI retainers. These study participants reported receiving SSI benefits in more than half of the months leading up to a follow-up interview; they also needed to be “on SSI” for a majority of their follow-up interviews. This group accounted for 47% (n = 193) of the study sample. The second primary income group consisted of study participants that reported some other type of public assistance (i.e., Temporary Assistance for Needy Families or General Assistance) as their primary source of income following termination of the DA&A category. This comparison group was characterized as having lost SSI benefits, but still relying primarily on public assistance. This primary income group accounted for 21% (n = 87; 31 TANF, 56 GA) of the study sample. The final primary income group, a residual group, was labeled the “no public assistance” group. These study participants were characterized as having lost SSI benefits and relying on informal resources and/or employment as their primary income source. These individuals accounted for 32% (n = 132) of the study sample.

Clinical characteristics

Issues related to mental health and substance abuse were a concern for this population. In order to assess the impact mental illness, and alcohol and drug use had on study participants’ ability to recoup or obtain income, variables related to these clinical characteristics were included in the study. The mental health variable was quite liberal; a dichotomous variable indicating the self-reporting of any serious mental health symptomatology (e.g., serious depression, serious anxiety, hallucinations, and suicidal ideation) at each follow-up interview. The indicators for substance abuse were any illicit drug use, or any heavy alcohol use (i.e., five or more drinks on five or more occasions in the past 30 days for men; four or more drinks on five or more occasions in the past 30 days for women) at any point following the policy change.

Demographics

Demographic variables included age (in years), gender, ethnicity, education, county of residence, employment, and time (in days) since first interview. Ethnicity was categorized as Caucasian (reference), African American, Hispanic, and Other; education was dichotomized to indicate high school graduate or equivalent. Dummy variables were constructed for the four Northern California counties (Alameda, San Francisco, San Joaquin, and Santa Clara); and employment was categorized as 1) no employment, 2) worked 20 hours or less per week, and 3) worked more than 20 hours per week. Variables for the interaction between income determinants (i.e., primary income source and employment) and clinical characteristics (i.e., any mental health symptom, any illicit drug use, and heavy alcohol use), and time since first interview were also constructed.

Statistical Analysis

Multivariate mixed random and fixed effects growth curve models were used to estimate changes in the total income of former SSI DA&A beneficiaries following the policy change. Several models of increasing complexity were estimated to understand the relationship between primary income source, employment, time, and post-policy-change income while controlling for demographic and clinical characteristics of the study sample. Mixed effects models can be conceptualized as two-level models. The first level is the individual or person level. This level models the expected change in income over time for each study participant. The second level is the population or group level. It is used to model the expected change in income over time between study participants. Mixed effects growth curve models have been shown to effectively estimate longitudinal outcomes (i.e., labor force participation and criminal behavior) for this population (Orwin, Campbell, Campbell, & Krupski, 2004). Since this study was interested in modeling income trajectories following termination of the DA&A category, baseline measures of income taken just prior to the policy change were adjusted for in most models.

Model 1, the unrestricted mean model, is the simplest model and estimated the grand mean total income along with level-1 and level-2 random effects. Model 2 included the time variable, days since the initial interview, both as a fixed effect for the mean rate of change in total income over time, and a random effect expressing between-person-level differences in the fixed effect time slope. The next two models added fixed effects for baseline income, primary income groups, demographic variables, and clinical characteristics in a stepwise fashion. Model 3 added fixed effect parameters for membership in one of three primary income groups: retained SSI (reference), other public assistance, and no public assistance, as well as a control for baseline income. Model 4 added demographic characteristics, such as age, gender, ethnicity, county of residence, and education, and fixed effects for clinical characteristics, such as the presence of any serious mental health symptoms, the use of an illicit drug, and the heavy use of alcohol. Model 5 added indicators of level of employment: no work (reference), worked 20 hours or less per week, and worked more than 20 hours per week. Covariates for cross-level interactions between the effect of work and time on income were also included. Stata 9.2 was used for all analyses (StataCorp, 2005).

Like most income data, the distribution of income for this study sample was positively skewed. Using a log transformation failed to address the distributional problems associated with skewed data. To address the positively skewed data, we used a bootstrap approach to empirically estimate the standard errors for the fixed and random effects. Simulation studies have demonstrated that bootstrapping effectively reduces the bias in standard errors associated with skewed income data and allows the coefficients to be displayed in their original dollar units (Afifi, Kotlerman, Ettner, & Cowan, 2007; Efron & Tibshirani, 1993).

Another issue with the self-reported income data was the presence of outliers. We assumed there were two potential sources of error leading to outliers. Outliers may have resulted from misreported or misrecorded data, or they may have occurred when study participants received a large one-time cash allotment, such as a retroactive SSI benefit or an inheritance. If income reported was more than four standard deviations above the mean, then the observation was considered eligible for winsorizing. This applied to 13 income data. Winsorizing the outliers was done by replacing the outlier with the next adjacent order statistic (Huber, 2002). For example, if a study participant reported $10,000 of monthly income from SSI and the next highest reported monthly SSI income value for that study participant was $850, $850 was substituted for $10,000. Study participants that indicated receiving a specific source of income but either refused to provide an amount or did not know how much they had received were assigned mean values.

Results

Differentiating study participants by primary income group revealed some demographic differences within the study sample (see Table 1). Study participants that requalified for SSI benefits under another disability category were significantly older than the rest of the study sample. Study participants that lost SSI benefits and relied on some other type of public assistance tended to be female and African American. Former SSI DA&A beneficiaries residing in Alameda County and San Francisco County were more likely have received public assistance (i.e., SSI, TANF, or GA) following the policy change, while study participants from San Joaquin County were more likely to have fallen into the no public assistance group. In comparison to the other three counties, San Joaquin County would be considered less urban and more agricultural. There was no difference in educational level between the three groups.

Table 1.

Baseline Demographic Information by Primary Income Group.

	Retained SSI n = 193	Other public assistance n = 87	No public assistance n = 132
Male (%)*	58.5	43.7	62.1
Age (years) (SD)***	45.1 (8.42)	41.3 (7.67)	41.5 (8.08)
Ethnicity (%)
African American**	43.8	52.9	32.6
Caucasian	32.8	19.5	28.8
Hispanic	17.2	17.2	20.2
Other	6.3	10.3	12.1
County of residence (%)
Alameda***	32.1	39.1	17.4
San Francisco***	37.8	24.1	15.2
San Joaquin***	18.1	27.6	53.8
Santa Clara	11.9	9.2	13.6
High school graduate (%)	53.4	54.0	60.6

Comparisons made using one-way ANOVA and Pearson’s chi-square tests.

^*p < 0.05;

^**p < 0.01:

^***p < 0.001.

Table 2 displays the results of five multilevel growth curve models for estimating total income following termination of the DA&A category. The first model indicates that the mean income for all study participants up to 3.5 years following the policy change was $652.62, over $100 less than the average income prior to losing SSI DA&A benefits ($766.54; data not shown). The second model adds a fixed effect for time and a random effect for individual differences in the change of total income over time. From this model, the estimated increase of a study participant’s income was approximately $0.11/day over the course of the study, when not controlling for any other variables. After 3.5 years, this would equate into an average total income of approximately $724 for this study sample. When a study participant’s baseline income and primary source of income following the policy change are added to the model (Model 3), we find a significant decrease in the total income of former DA&A beneficiaries that were unable to retain or requalify for SSI benefits. When compared to SSI retainers, study participants that lost SSI benefits averaged over $236 less in total income following the policy change, adjusting for baseline income and the number of days since the initial interview.

Table 2.

Income Growth Models.

	Model 1: Random Intercept	Model 2: Unrestricted Growth	Model 3: Income Groups	Model 4: Demographic and Clinical Variables	Model 5: Employment Variables
Fixed Effects (SE)
Intercept	652.62 (8.04)***	583.85 (12.60)***	443.57 (32.40)***	852.38 (78.57)***	756.49 (72.85)***
Time since initial interview (days)		0.11 (0.02)***	0.11 (0.02)***	0.12 (0.02)***	0.03 (0.02)
Baseline income			0.35 (0.05)***	0.31 (0.04)***	0.26 (0.04)***
Other public assistance group			−236.55 (18.63)***	−246.74 (15.89)***	−278.32 (16.77)***
No public assistance group			−236.08 (28.12)***	−240.25 (24.83)***	−355.86 (22.05)***
Age (years)				−6.27 (1.24)***	−3.75 (0.98)***
Male				−60.79 (17.59)***	−47.40 (14.73)**
High school graduate				64.76 (14.96)***	40.59 (14.59)**
African American				−8.96 (21.72)	0.53 (19.82)
Hispanic				−24.29 (27.81)	−24.35 (23.56)
Other ethnicity				−76.31 (30.17)*	−36.18 (28.61)
Alameda County				−66.90 (21.63)**	−52.54 (17.43)**
San Joaquin County				−53.02 (20.84)*	−43.18 (24.22)
Santa Clara County				−4.02 (32.46)	−24.04 (29.03)
Any mental health symptom				−85.34 (29.96)**	−39.12 (23.62)
Any illicit drug use				35.55 (26.92)	35.51 (24.76)
Heavy alcohol use				−12.88 (29.03)	0.99 (20.45)
Worked 20 hours or less					193.33 (48.01)***
Worked more than 20 hours					370.28 (107.47)***
Worked 20 hours or less by time interaction					0.02 (0.10)
Worked more than 20 hours by time interaction					0.41 (0.12)***
Random Effects (SE)
Between-person	314.92 (14.98)***	317.10 (29.01)***	265.15 (27.75)***	260.90 (24.17)***	225.86 (23.28)***
Random time slope		0.31 (0.04)***	0.31 (0.04)***	0.31 (0.03)***	0.22 (0.02)***
Correlation		−0.29 (0.09)***	−0.30 (0.10)***	−0.37 (0.09)***	−0.56 (0.11)***
Within-person	340.30 (13.17)***	306.03 (28.34)***	306.16 (18.77)***	305.56 (26.04)***	291.80 (22.87)***

^*p< 0.05;

^**p <0.01;

^***p< 0.001.

Reference groups: Retained SSI (primary income group), San Francisco (county), and Caucasian (ethnicity).

Model 4 adds demographic, mental health, and substance abuse covariates to the model. Not surprisingly having a high school degree or its equivalence was positively related to income; however, increases in age and being male were associated with lower income. Residents of Alameda County and San Joaquin County experienced significantly less income than residents of San Francisco. This model also indicates that study participants from the Other ethnicity category earned significantly less income, as well as study participants reporting any serious mental health symptomatology, when adjusting for other covariates.

The final model adds employment and a cross-level interaction examining the effect of employment on individual differences in the time slope. This full model includes interaction coefficients for working 20 hours or less, working more than 20 hours per week, and time since initial interview. Other models (not shown here) tested the other interactions between employment, primary income group, any mental health symptom, any illicit drug use, heavy alcohol use, and time since initial interview; only the interaction representing working more than 20 hours per week and time since initial interview had a significant effect on the time slope. These results indicate that study participants who worked more than 20 hours per week had a significantly higher rate of income change over time, while study participants that were unable to achieve substantial rates of hourly employment experienced no change in their total monthly income over time.

Discussion

The primary function of any welfare state is to ensure the economic security and basic material sufficiency of its citizens (Gilbert & Terrell, 2006). Findings from this study indicate that reforms to the Supplemental Security Income program that eliminated income and health care benefits to low-income substance abusers actually compromised the economic security of some of America’s most vulnerable and disadvantaged citizens. While some former SSI DA&A beneficiaries were able to attain substantial gainful employment (36.5% of our study sample that failed to requalify for SSI benefits reported working 20 hours or more per week at some point following the policy change; the average income for these study participants was approximately $1145 per month at the 42-month follow-up), many study participants (49.3%) reported never experiencing any employment following their loss of federal income maintenance benefits. The average total monthly income of study participants that lost their formal cash assistance and did not experience any employment up to 3.5 years following the policy change was only $379 (42-month follow-up; data not shown). Expectations by policymakers that former SSI DA&A beneficiaries that did not requalify for SSI benefits under another disability category would find employment and achieve a subsistent level of self-sufficiency were only partially realized. For a significant proportion of this population, the policy change was associated with reduced income and increased economic hardship.

Study participants that did requalify for SSI benefits under another disability category were able to recoup over 93% of their baseline income 3.5 years following the policy change, adjusting for inflation. For study participants that lost benefits and could not requalify for SSI benefits, income recuperation following elimination of the DA&A category was significantly more difficult. The group of study participants that relied on some other type of public assistance was the most severely affected, losing nearly 22% of their immediate economic security. The rationale that most former SSI DA&A beneficiaries would requalify for SSI benefits under another disability and carry on with their lives uninterrupted was neither prudent nor realistic. In the current study, nearly 40% of study participants that did not requalify for SSI benefits ended up relying on some other form of public assistance as their primary source of income following the policy change. Policymakers should be concerned that eliminating an eligibility category in a federal public assistance program simply resulted in a significant proportion of former beneficiaries shifting to less economically substantial forms of state and local public assistance (i.e., TANF and GA). This also led to increased involvement among informal mechanisms of care, such as family, friends, and communities, requiring these groups to absorb much of the responsibility and burden of providing economic support to former public assistance recipients. From this perspective, the policy change fell short of meeting many of its original goals.

Speculation that substance abuse among former SSI DA&A beneficiaries would have a negative effect on their ability to achieve self-sufficiency following the policy change was not supported by statistical evidence. For this study sample, alcohol and drug use did not have a significant effect on level of post-policy-change income. Surprisingly, this was also true for individuals reporting serious mental health symptoms. From previous research, we know a substantial proportion of former SSI DA&A beneficiaries that did not requalify for SSI benefits continued to suffer serious mental health issues following the policy change (Hogan, Speiglman, & Norris, 2007; Watkins, et al., 2001). It is interesting that in the current study our mental health indicator was not a significant predictor of income following the policy change. The fact that individual characteristics, such as alcohol and drug use, and mental illness did not have a significant effect on post-policy-change levels of income may alert researchers to examine more systemic or structural influences on the income maintenance of former welfare recipients, such as access to transportation and available child care services. This is an area for future research.

There were limitations to this study. The initial quasi-experimental research design limits our confidence to state that the policy change was the cause of the reduction in income for study participants that lost SSI benefits. Another important limitation of this study relates to measurement reliability and validity. Measures of income and employment were based on self-reports. The inaccurate reporting of these variables by study participants could have critical implications for this study. There were also limitations to the study’s sampling method and external validity. The study sample was limited by the sampling frame provided by Maximus, the referral and monitoring agency responsible for tracking SSI DA&A beneficiaries in Northern California. SSI DA&A recipients not known to Maximus were unavailable for selection. This undoubtedly resulted in selection bias. Finally, results from this study cannot be generalized to all former SSI DA&A beneficiaries. Northern California has regional variations that may have affected outcomes for this population. For example, during the time of this study, the economy in Northern California was very good, possibly allowing for better income and employment outcomes among study participants.

Conclusion

Social welfare policy reforms of the mid 1990’s were intended to reduce the number of Americans receiving public assistance by promoting personal responsibility and self-sufficiency. During this movement, politicians attempted to cleanse the welfare rolls of “undeserving” recipients and replace assumed welfare dependence with work. By eliminating the SSI DA&A category, politicians theorized that low-income individuals with a legitimate physical or mental health disability would retain their public assistance benefits through the SSI requalification process, and low-income substance abusers would be expelled from the public dole and forced to find alternative sources of income, ideally employment. Whether you philosophically agree with this method of welfare reform is not the current issue; what is important is to understand the effect such policymaking decisions have on social welfare beneficiaries.

Proponents of the policy change estimated that 75% of former SSI DA&A beneficiaries would requalify for SSI benefits under another disability category. When only 35% of this population retained their SSI benefits, policymakers should have realized that the social welfare of some of our society’s most vulnerable members had been compromised—a result that is antithetical to the goals and objectives of a progressive welfare state and a healthy society. Some former SSI DA&A beneficiaries were able to find employment and achieve self-sufficiency; however, a substantial proportion was left to suffer increased hardship. When making policy changes that affect marginalized and disadvantaged populations, policymakers should expect some negative consequences. If policymakers are purposefully going to cut a hole in the “safety net” of this country’s social welfare system, they must have contingencies for the poor and disabled that fall through it.