Trends in Outcomes for Young People with Disabilities: Have Policies Aimed at Social Integration and Economic Independence Been Effective? Ann E. Horvath-Rose, Ph.D.* David C. Stapleton, Ph.D. Cornell Center for Policy Research 4301 Connecticut Ave. NW Suite 330 Washington, DC 20008 *Corresponding author. Phone (202) 719-7835; Fax (202) 719-7878; email aeh8@cornell.edu Abstract This paper is a first attempt at measuring long-term trends in outcomes for youth (aged 15 – 21) and young adults (aged 22 – 29) with disabilities, from 1981 through 2000, using the Current Population Survey, and focusing on change from the 1988 peak of the business cycle to the 1999 peak. We find that prevalence of work limitations declined for males, but increased for females. The growth for females is mostly accounted for by growth for black females. We also compare trends in an array of socioeconomic outcomes for those with and without work limitations. Despite a substantial reduction in the educational attainment gap between these two groups, the gaps in employment, earnings, and poverty widened substantially. Trends in marital status, parenting, and teen childbearing make it apparent that welfare reform contributed substantially to growth in the prevalence of work limitations for females. Health insurance coverage for those with work limitations has been fairly stable, but there has been a large shift from private to public coverage. These trends could be due to: i. factors that determine whether individuals report themselves to be work-limited, ii. factors that affect individual outcomes, regardless of self-reported work limitation status, iii. or factors that do both. Keywords: disability, work limitations, trends, demographics, employment, income I. Introduction A leading goal of current disability policy is to change the physical and social environment in ways that will help people with disabling physical and mental conditions participate more fully in major social activities, such as schooling, employment, marriage and parenting; enjoy a higher standard of living; achieve greater economic independence; and live more independently. During the last two decades there have been major changes in policies and programs affecting persons with disabilities. Many of these changes are particularly relevant for youth and young adults with severe disabilities, who often face lifetimes of social segregation, poverty, and dependence. Yet little research has been done analyzing long-term trends in outcomes among youth and young adults with disabilities.1 Without such research, we cannot know whether changes in the physical and social environment have been so substantial that fewer young people with disabling physical and mental conditions consider themselves to have major activity limitations, whether those changes are induced by public policy or other factors. We will also not know whether those who do have activity limitations are participating more fully in major social activities, enjoying a higher standard of living, achieving greater economic independence, or living more independently. This paper is a first attempt at measuring trends in these outcomes for youth (defined as those aged 15-21) and young adults (aged 22-29). For this purpose, we use data from the Current Population Survey (CPS), covering the period from 1981 to 2001. The analysis is necessarily limited by the nature of the disability and other questions asked in the CPS, but we consider the CPS to be the best starting point for such an analysis because it is large enough to obtain reasonably precise estimates for many interesting statistics; it has been fielded annually in a very consistent fashion over the last two decades; and it is the official federal source of population statistics for many of the outcomes of interest. II. Definition of Disability There is substantial ambiguity and disagreement in the literature about exactly how disability should be measured, but there is widespread acceptance of the concept that disability is inability, or reduced ability, to participate in a major life activity (e.g., education, work, marriage, parenting, etc.) as a consequence of a physical or mental condition and its interaction with the environment. We use “activity limitation” to refer to such limitations. Social science has moved away from the “medical model” of disability, which ascribed activity limitations to a person’s physical or mental condition alone, toward a model in which the environment plays a major role in determining whether that condition leads to an activity limitation. A medical condition might cause certain basic “functional limitations,” such as inability to walk, see, concentrate, or lift even a small weight, but a functional limitation alone does not prevent a person from participating in major activities. There are physical and social environments in which individuals with severe functional limitations can, and do fully participate in major activities. Definitions of disability that are used in the literature vary widely. These definitions include ones that focus on medical conditions and functional limitations; on participation in programs that target people with disabilities; and on activity limitations. Most programmatic definitions of disability have a medical orientation. While self-reported activity limitations seem conceptually aligned with the concept of disability that is adopted by social scientists, we do not know the extent to which individuals consider both medical and environmental factors when they assess their own situation. Such assessments are highly subjective. Further, as attitudes about medical conditions and disability change, people may change their self-assessments even if there is no change in either their medical conditions or their environment. Attitudes, themselves, might be considered to be part of the environment. Wittenburg and Maag [11] find that programmatic definitions are most commonly used in the literature on youth and young adults with disabilities. The programs most commonly referenced are special education, SSI, and vocational rehabilitation. Second most common are self/parent-reported activity limitations. The data sets that allow for analysis of long-term trends most commonly use self/parent-reported limitation data. Therefore, when thinking about trends in the proportion of a population that reports such limitations, we have to go beyond thinking about trends in personal characteristics, such as medical conditions, to trends in the physical and social environment that affect self reports. The latter trends can have real effects on activity limitations, and can also change self-perceptions of limitations. The population that reports activity limitations might differ substantially from a population that would be identified by more objective scientific definitions and the difference between the two might vary over time. We use data from the Annual Demographic Survey (March supplement) of the Current Population Survey (CPS), an annual nationally representative cross-sectional survey of approximately 150,000 non-institutionalized civilians, 50,000 households, conducted by the Bureau of Census for the Bureau of Labor Statistics. As the primary source of information on the labor force characteristics of the U.S. population, the CPS is rich in information about employment, unemployment, income, earnings, hours of work, program participation, health insurance, and educational attainment, as well as demographic characteristics including age, sex, race, marital status, and living arrangements. Since 1981 the CPS has also included a question about work limitations, asked of the reference person in the household: Does anyone in this household have a health limitation or disability which prevents them from working or which limits the kind of amount of work they can do? [If so,] who is that? (Anyone else?) There has been much discussion in the literature about the usefulness of the work limitation definition of disability (Hale 2001 [24]; Hardy and Pavalko 1986 [15]; Kaye 2002 [21]; Kirchner 1996 [5]; Kruse and Schur 2000 [9]; McNeil 2000 [14]). A recent paper by Burkhauser et al. [20] addresses the criticisms by comparing trends based on the CPS work limitation measure to trends from measures that focus on medical conditions (based on the National Health Interview Survey) and functional limitations (based on the Survey of Income and Program Participation) over substantial sub-periods of the last two decades. They find that, for persons aged 25-61, “although work limitation questions underestimate the size of the broader population with health impairments, they can be used to monitor trends in outcomes, such as employment, for those with disabilities.” More specifically, they find that over the last two decades trends in employment rates for people with disabilities do not depend in a substantial way on whether a medical condition, functional limitation, or activity limitation measure of disability is used, especially if peaks in the business cycle are compared. Work-limitation measures are clearly sensitive to the business cycle, however, with a larger share of adults reporting work limitations during recessions. These findings have encouraged us to look at trends in outcomes for youth and young adults with self-reported work limitations using the same CPS question. As we do so, however, it is important to keep in mind that the reporting of work limitations can be affected by changes in the economic or policy environment. This might be especially true of youth and young adults, whose adult roles and self-perceptions are just emerging. It is also important to keep in mind that this definition omits individuals who might consider themselves to be limited in other major life activities, but not work. Because the definition of disability used in this paper is based on work limitations, we will refer to the population we study as persons with work limitations. Furthermore, while we report trends for 15-29 year olds, we expect the results for 15-18 year olds, who have generally not yet entered the work force, to be less informative than those for older age groups. Because a very large share of those people without disabilities who work have entered the labor force by the age of 22, including those who obtain a college degree, we focus more heavily on persons aged 22-29. Between 15 and 29 there is a great deal of age-dependent variability in the outcomes we consider, regardless of activity limitations. That is, many of the outcomes, such as employment, educational attainment and marriage, materialize as an individual matures (“maturation effects”). Therefore, trends in the outcomes we measure are sensitive to changes in the age composition of the population within the 15-29 age group. For example, a relative increase in the number of persons in their late 20’s could lead to an increase in marital rates among 22-29 year olds that are solely due to the change in the age composition, rather than a change in behavior. For this reason we calculate age adjusted outcome series using fixed 1990 population weights for narrower age groups. Within the 15 to 21 age group, 15-18 year olds and 19-21 year olds are weighted based on 1990 population weights; within the 22 to 29 age group, 22-24 and 25-29 year olds are weighted analogously. Statistics for the full age range, 15-29, are also weighted to reflect the 1990 population weights within these narrow age groups. Due to small sample sizes, the trends we report are based on three year moving averages. Furthermore, all income data are presented in real 1998 dollars. Based on the CPS, there were 4,438,518 persons aged 15 to 29 with work limitations in the (non-institutional) population in 2000, representing 2.6 percent of all persons in this age group (Table 1). Of these, 58.5 percent were aged 22-29, down from 68.6 percent in 1990. The change in the age distribution reflects the fact that in 1990 the youngest members of the baby boom generation were in their late 20s, while by 2000 they had all exited the 15 – 29 year old age group, and their children were entering it.2 III. Changes in Programs and Policies Relevant to Youth and Young Adults The 1980s and 1990s have seen many changes in policies related to education and employment of persons with disabilities. The goals of these policies are to improve access to public education for children with disabilities, facilitate transitions from school to post-school activities, enhance their ability to work in competitive employment, and provide better employment opportunities. The Education for All Handicapped Children Act of 1975, which first integrated children with disabilities into public schools, was amended in 1986 to link school and adult services for youth with disabilities in order to facilitate their post-school transition. The Act was renamed the Individuals with Disabilities Education Act (IDEA) and amended in 1990 and 1997. These amendments put more emphasis on preparing individual transition plans (IEP) for students with disabilities. The goal of the transition planning is to help youth with disabilities develop vocational and life skills that will ultimately lead to employment as adults. As a result of a 1990 Supreme Court ruling in Sullivan v. Zebley, SSI eligibility for children was significantly expanded. Prior to 1990, in order to be eligible for SSI a child’s impairment had to appear on a list of eligible impairments; unlike adults, there was no additional functional assessment performed. After the Supreme Court ruling, the eligibility rules were changed such that children whose impairments were not on the official list of impairments were given an individualized functional assessment (IFA) to determine eligibility. The new rules stated that, in determining the severity of a child's impairment, SSA must consider whether the limitations caused by the impairment substantially reduce the child's ability to engage in “age-appropriate” activities. These changes put additional financial resources into the hands of parents of children with disabilities, which would presumably encourage parental investment in their children and better adult employment outcomes. The changes might also have the opposite effect, however, by increasing awareness of reliance on SSI as an alternative to reliance on earnings in adulthood. Kubik [13] finds that the expansion of SSI eligibility for children resulted in greater detection and treatment of health problems in low-income children. This contributed to the large post-Zebley expansion of SSI -- a 260 percent increase in the number of child (under age 18) SSI recipients, from 265,000 in (December) 1989 to 955,000 in 1996 [22]. The welfare reform legislation of 1996 reversed a substantial component of the eligibility expansion for children. Specifically, the IFA was eliminated as a way of determining if a child’s impairment met eligibility requirements. However, by December 2001 the number on children under 18 on the rolls had fallen by only eight percent from its 1996 peak, to 881,000. In the same month, the number of SSI recipients aged 15 – 18 was about 225,0003 -- 26.9 percent of the CPS estimate of the number of persons with work limitations in this age group (836,255) in 2000. Congress passed the Americans with Disabilities Act (ADA) in 1990. This major piece of legislation provides civil rights protections to persons with disabilities. The Act states that: No covered entity shall discriminate against a qualified individual with a disability because of the disability of such individual in regard to job application procedures, the hiring, advancement, or discharge of employees, employee compensation, job training, and other terms, conditions, and privileges of employment. A primary objective of the ADA is to integrate persons with disabilities into the labor force by increasing their employment opportunities. In order to achieve this, the ADA mandates that employers provide “reasonable accommodations to the known physical or mental limitations of an otherwise qualified individual with a disability who is an applicant or employee” and makes it illegal to deny “employment opportunities to a job applicant or employee who is an otherwise qualified individual with a disability.” The ADA was an attempt to change work place environments, and employer personnel policy, in such a way that a person’s disabilities would not interfere with their ability to perform their job. Since the passage and implementation of the ADA there has been a decline in the employment rate of people with disabilities. Some researchers attribute the decline to the ADA itself, arguing that employers want to avoid the cost of accommodation and/or fear that if they hire a person with a disability they will be at risk of a lawsuit for future personnel actions (Acemoglu and Angrist 2001 [7]; DeLeire 2000 [23]). Others, however, find this evidence less than compelling (Kruse and Schur 2002 [9]; Blanck et al. 2003 [17]), and there are competing explanations for the decline, including expansions in disability income support programs (Burkhauser and Stapleton 2003 [19]). IV. Findings 1. Prevalence of Work Limitations A comparison between the two peak economic years 1988 and 1999 shows that the prevalence of work limitations (i.e., the percent of the population reporting work limitations) for 15 to 29 year olds declined by 6.4 percent (Figure 1). There was a substantial decline for males in this age group (15.9 percent) but the prevalence of work limitations actually increased for females (5.9 percent). The decline in prevalence for males is observed among whites (18.2 percent) and Hispanics (32 percent), but not in the black population (Figure 2).4 For females, although there was a slight decline among whites (2.5 percent) and Hispanics (3.5 percent), there was a sizeable increase in prevalence for black females (35.6 percent); this increase in prevalence is driving the overall increase in prevalence for females (Figure 3). i. Males The decline is most dramatic for males aged 25-29 (Figure 4). Between 1988 and 1999, the prevalence of work limitations for this group decreased by 28 percent, from 4.3 percent to 3.1 percent. The decline was not steady, however. Prevalence grew somewhat during the recession of the early 1990s, and then declined persistently through the economic expansion after reaching a peak of 4.5 percent in 1993 – a much larger decline than observed in the expansion of the late 1980s. An important issue we cannot address with the CPS data is the extent to which this decline reflects: a) a decline in the prevalence of disabling physical and mental conditions; b) advances in medical and rehabilitation technologies treatment that reduce the likelihood of such conditions leading to functional limitations; c) environmental changes that reduce the likelihood that functional limitations lead to work limitations; and d) changes in self-perceptions of work limitations that are not due to real changes in physical or mental conditions. Given the rapidity of the decline, it seems unlikely that it could be due solely to declines in prevalence of physical and mental conditions or to improvements in their treatment. As the CPS data are for the non-institutionalized population, one possibility is that changes in institutionalization of people with physical and mental conditions might have contributed to the prevalence trends. However, efforts to de-institutionalize those with severe mental and cognitive disorders would presumably increase prevalence in the non-institutionalized population. At the same time, increases in incarceration rates conceivably had a negative effect on prevalence for males. There has been a sharp increase in the share of young incarcerated males, with only a small increase for young females.5 Although prevalence of work limitations might be relatively high among those incarcerated, it would have to be extraordinarily high for observed increases in incarceration to substantially reduce prevalence in the non-institutionalized young male population. Also, the incarceration rate for males increased much more for blacks than for others, so if this were the explanation the prevalence rate for blacks should have fallen relative to the prevalence rate for other groups; instead, as shown above, the opposite occurred.6,7 The most obvious environmental factor is improvement in the job market, but we have to wonder why improvement in the second half of the 1980s was not accompanied by a decline similar to the decline in the late 1990s. Also, declines in the proportion with work limitations for older males during this same period are much less substantial. In fact, there is an increase in prevalence among men aged 45-54.8 The same applies to young females, although there is a convincing explanation for that difference, as we discuss below. There are several possible explanations that appear worthy of further exploration. One possible explanation is that the six-year, 1.3 percentage point decline for 25-29 year old males from 1994 to 1999 is partly the reflection of a four-year, 1.0 percentage point decline for 15-18 year old males from 1986 to 1989. The younger group in the earlier period represents essentially the same birth cohort as the older group in the latter period. If the earlier decline did contribute substantially to the later decline, it would imply that policies that successfully reduce self/parent reports of work limitations of youth with disabilities would also reduce self-reports of work limitations when they are young adults. It would also suggest that the decline for 25-29 year olds is likely to end, or even reverse somewhat, reflecting the reversal of the trend for 15-18 year olds in the early 1990s. This begs the question of why the earlier decline occurred and ended. Was the decline caused by innovations in educational policy? Was it caused by the economic expansion of the late 1980s? Did it end because of the 1990 recession? If so, should it have continued more vigorously in the 1990s expansion? Did it end because of the expansion of SSI for children? It seems likely that the later decline for the oldest group is not solely the result of the earlier decline for the youngest group because a) the former is larger in length and magnitude than the latter, and b) there is little decline for those aged 22-24 between 1992 and 1995, when the relevant birth cohorts were in that age group. Nonetheless, we should not dismiss the idea that the earlier decline contributed substantially to the latter decline. Another possible explanation of the decline is that economic growth of the 1990s was stronger or of a different nature than in the late 1980s, creating unprecedented opportunities for young males with disabilities. Perhaps rapid growth of information technologies, which helped fuel the more recent growth, also generated jobs that many young males with disabilities who had received computer training while in school were able to compete for. If so, however, why was the effect for the oldest age group greater than for younger age groups? A third possible explanation of the decline is that the ADA expanded opportunities for young adults with disabilities to the point where fewer considered themselves to be work limited. Perhaps the finding is strongest for the oldest age group because those who experienced the onset of a disabling condition while working were less likely to develop work limitations and lose their jobs as a result of the ADA. The pro-work message of the ADA might have also encouraged more young adults with disabilities to consider themselves not work limited, holding their real opportunities constant. A final possible explanation of the decline is that advances in the treatment of affective and other psychiatric disorders, especially drug therapy (e.g. Prozac, the first of the new anti-depressants, was introduced in January 1988), might have reduced functional limitations that are caused by these disorders, which, in turn, reduced work limitations. Psychiatric disorders account for a larger share of disorders among young adults with disabilities than among older adults, and onset frequently occurs during the teenage years and early 20s [16]. We also note there was an expansion in the percent of males aged 25-29 with work limitations from 1983 to 1987, when SSA expanded availability of SSDI and SSI benefits for persons with psychiatric disorders, following legislation enacted in 1984. ii. Females While the results indicate that work limitations are less prevalent among young females than among young men, over this same period prevalence for young adult females increased or remained constant, while prevalence for young adult males declined (Figure 5). Prevalence for young females aged 22-24 increased by 10.8 percent and the prevalence for females aged 25-29 remained constant. Most dramatically, for 19-21 year old females, prevalence increased by 14.8 percent. During the economic expansion preceding 1988, prevalence trends for males and females were much more similar. The factors that explain the decline in prevalence for young males have likely put downward pressure on the prevalence for young females. However, there are female-specific factors that counter-balance those downward forces. Perhaps most important, state welfare reforms in the early 1990’s that culminated in the overhaul of the federal welfare system in 1996 may have created an incentive for young adult welfare recipients, mostly females, to identify themselves as work limited, in part because of interest in qualifying for exemptions from welfare program requirements and in part to qualify for SSI [10]. The 1996 conversion of the Aid to Families with Dependent Children program to the block grant Temporary Assistance for Needy Families (TANF) program, as well as earlier state reforms, created an environment in which low-income mothers may be more likely to consider themselves to be work limited, because establishment of such limitations could result in exemption from new requirements and even eligibility for SSI benefits. Reforms under TANF as well as pre-TANF reforms in some states limit welfare receipt to two consecutive years and five years in a lifetime, whereas there was no time limit on AFDC receipt. These reforms also impose strict new work requirements on welfare recipients. Many states provide exemptions to both of these requirements for recipients with disabilities [25]. While the physical or mental condition of most parents in TANF families likely did not change post-welfare reform, the change in policy created an environment in which some welfare recipients could not comply with the new rules associated with welfare receipt due to a pre-existing health condition or disability that, under the previous welfare system they did not consider to cause a work limitation. TANF also required welfare mothers to cooperate with paternity establishment. For females already receiving informal child support payments, formalization of the payments results in a net loss of income, because child support payments reduce monthly welfare benefits dollar for dollar. This, along with the other requirements, increased the financial incentive for welfare mothers with disabilities to apply for SSI, which does not have time limits, work requirements, or paternity establishment requirements. As noted, the increase in prevalence for females was driven by the increased prevalence for black females. Black families make up a disproportionately large share (38 percent in 1999) of the population receiving welfare benefits.9 Therefore an environment created by welfare reforms in which more females consider themselves work limited would have had a greater effect on the black population than on the white or Hispanic populations. In the absence of welfare reform we might have seen a much smaller increase or even a decrease in prevalence for young black females, and an even greater decrease for white and Hispanic females. While the disparity by race in prevalence may be explained to some extent by welfare reforms, the differences in prevalence by race and ethnicity among females merits more study. The second factor that might have put upward pressure on prevalence of work limitations for young females is the positive trend in their labor force participation. In the past, females who chose to stay out of the labor force but who had a health condition that would have limited the type or amount of work they could do might not have classified themselves as work limited because the health condition was immaterial to their work decision. Because more females are working, more experience the onset of disability while working and more are eligible for public or private disability benefits as a result. Others might experience onset as they are preparing to enter the labor force, so they consider themselves work limited, whereas their counterparts in earlier years, who were preparing to be homemakers, might not have. As noted earlier, trends for males and females were much more similar before the early 1990s than afterward. As growth in females’ labor force participation occurred both before the early 1990s and afterward [12] it seems likely that the welfare reforms are the more important explanation of the differences in trends in the latter period. In summary, while the factors that contributed to a decline in prevalence for young males in the 1990s probably put downward pressure on prevalence for young females, it appears that this pressure was more than offset by the effect of welfare reform. Growth in labor force participation of females might also have contributed to this trend. Statistics we present later regarding trends in the parental status of this population are consistent with this interpretation. For both males and females, it appears that environmental change played a key role in determining trends in the prevalence of work limitations. For males the change might reflect a mix of changes in education, improvements in job opportunities, and changes in perceptions all of which could have had a negative impact on prevalence. These same forces likely affected females, but welfare reform and long-term increases in labor force participation appear to offset the effects of other factors. These changes likely caused changes in the composition of the population with work limitations, both overall and for each sex. That is, the distribution of personal characteristics (as opposed to environmental characteristics) changed within each group, because those whose answers to the work limitation question are most sensitive to the environmental changes (i.e., those “at the margin” of identifying themselves as work limited) probably have different distributions of personal characteristics than the total work limited population. For example, one compositional change is an increase in the share of the work limited population that is female, apparently reflecting the effect of welfare reform and increases in labor force participation on female self-reports of work limitations. As we examine how mean outcomes for the population with work limitations have changed, it is important to consider whether compositional change caused by environmental factors might have played a significant role. 2. Educational Attainment Increased educational attainment is a cornerstone of recent efforts to improve the transition into adulthood for young adults with disabilities and to achieve the goals of independent living and participation in the workforce. Compared with youth and young adults with no work limitations, the data show that youth and young adults with work limitations are more likely to have no education, less likely to have a high school degree, and less likely to have attended college or to have a college degree. The trends, however, show that there has been substantial progress in reducing the gap between educational attainment for those with and without work limitations. Although almost every person aged 15-29 without a work limitation has at least some education, a substantial share of those with work limitations do not. That share declined considerably, however, between the late 1980s and mid 1990s from 6.4 percent in 1988 to 2.8 percent in 1997, then remained steady through 2000 (Figure 6).10 The share of work limited young adults with less than a high school degree decreased in the mid-1990s while the share with a college degree increased (Figure 7). Unfortunately, the positive trends in high school completion seem to level off at the end of the 1990’s. It seems likely that compositional changes in the population with work limitations caused by environmental factors depressed the positive trends in educational attainment. Improvements in job opportunities in the 1990s might well have been greatest for those with relatively high levels of education, and perhaps they were more likely than others to exit the work limited population (for reasons stated above). We conjecture, too, that the upward pressure from welfare reform on prevalence for females likely had a depressing effect on mean education for females with work limitations. Put differently, it seems likely that in the absence of such compositional change, increases in educational attainment might have been stronger. The disparities in educational attainment by sex among youth and young adults with work limitations are similar to those observed among youth and young adults with no work limitations. Among 22-29 year olds, young males are more likely to have less than a high school degree and young females are more likely to have at least a high school degree and to have attended college. During the welfare-reform period of the mid-1990s, educational attainment for young females with work limitations declined relative to attainment for young males, which could reflect the influence of welfare reform on reporting of work limitations, but by the end of the period the male-female differential is about the same as in the late 1980s. 3. Employment Rate There is a consensus among researchers that employment rates for working age persons with disabilities decreased in the 1990s, despite the ADA and the healthy economy in the late 1990s [19]. Although our analysis focuses on a younger age group than other studies, it also shows that employment rates for work limited 15-21 and 22-29 year olds decreased between the two peak economic years 1988 and 1999 (Figures 8 and 9). The decreases were greatest for males with work limitations, with declines of 28.4 percent among 15-21 year old males and 23.5 percent among 22-29 year old males. The decreases for females with work limitations were lower, at 13.7 percent and 11.3 percent for 15-21 and 22-29 year olds respectively.11 Furthermore, while the 1990s saw an increase in educational attainment among persons with work limitations, Figure 10 shows that the employment rate of work limited 22-29 year olds with high school and college degrees decreased by over 16 percent between 1993 and 200012, 13. According to some researchers, the movement, in the late 1990s, toward an integrated educational system for all students has increased the emphasis on academics to the exclusion of vocational education, career education and other experiences related to the development of social skills and independent living [4]. Thus, the findings are consistent with the argument that, although students with disabilities are increasingly being taught the academic information needed to graduate from high school, upon graduation, they are falling behind on basic employment and life skills. It is possible that a portion of the decrease in employment rates is due to compositional changes in the work-limited population that were caused by changes in the environment. If improvements in education and better job opportunities are major causes of the decline in prevalence for males in the 1990s, then those who report no work limitations as a result of these trends are likely employed at much higher rates than those who remain in the work-limited population. Differences between the trends for males and females might reflect welfare reform, but the effect of welfare reform on employment rates for females with work limitations is complex. Mothers who enter this population as a result of the reforms might be less likely than others in this population to work. At the same time, however, the reforms themselves might have encouraged many females who would have been in this population anyway to work. Growth in the labor force participation of females could also contribute to the differences in trends for males and females. The factors that caused the employment rate decline for males aged 15-21 are unlikely to have been factors that affect prevalence of work limitations because prevalence for this age group did not change. It is mathematically possible, however, that the factors causing the substantial decline in prevalence for males aged 22-29 also caused the entire decline in their employment rate (see Appendix I). Even if that were true, however, at best we would have to conclude that the employment rate is unchanged among those young males who would have considered themselves work limited under the environments of both 1988 and 1999. A similar, but more complex analysis for females shows that exits for work and entrances due to welfare reform cannot fully explain the decline in the employment rate for females aged 15-21, but could easily explain the entire decline for those aged 22-29. In an analysis of older groups in the working-aged population of people with work limitations, Houtenville and Daly [2] found pervasive evidence of employment rate declines during this period for all age, sex, education, and race groups, yet most groups did not experience substantial declines in prevalence. Burkhauser and Stapleton [19] review the evidence on the causes of the decline for older groups and conclude that policy changes are the like cause of the decline. As we discuss further in the concluding section, it seems likely that these same policy changes are a cause of at least part of the decline in the employment rate for younger adults with work limitations of both sexes. 4. Income A main goal of disability policy is to boost economic independence of persons with disabilities. A crucial step toward this goal is increased personal income for this population. Despite efforts aimed at increasing opportunities for persons with disabilities, work limited young adults have not been successful in closing the income gap between themselves and non-work limited young adults; in fact a comparison between the two peak economic years 1988 and 1999 shows an increase in the gap. In 1988, mean income of 22-29 year old work limited males was 56.9 percent of the income of non-work limited males in the same age group, falling to 44.6 percent in 1999, due mainly to a decrease in real income for work limited males. From 1988 to 1999 work limited 22-29 year old females’ income as a percent of non-work limited 22-29 year old females’ income fell from 65.1 percent to 60.0 percent; this decrease was primarily due to an increase in the real income of non-work limited females. The income gap between males and females aged 22-29 with work limitations has narrowed considerably; by 1999 female income was 92.8 percent of male income. However, the narrowing gap is attributable to a decrease in male income rather than to an increase in female income. Figures 11 and 12 depict the components of income for young adults. These components include: earnings; welfare payments; Social Security and SSI payments (included together due to frequent confusion by respondents between the two); and other income including worker’s compensation, unemployment compensation, veteran’s payments, private disability insurance payments, and miscellaneous other sources. While earnings comprise the vast majority of income for non-work limited young adults, this is not the case for work limited young adults. Work limited young males saw their real earnings decline dramatically (28.3 percent) from 1988 to 1999, causing earnings as a percent of income to drop from 64.0 percent to 56.8 percent. While there was essentially no change in real earnings for young work limited females from 1988 to 1999, earnings as a percent of income decreased from 62.4 percent in 1988 to 59.1 percent in 1999. Between 1981 and 1999 Social Security (DI) and SSI payments as a percent of total income for young adults with work limitations varied greatly making up anywhere from a low of 12.3 percent in 1981 to a high of 25.9 percent in 1997 for young males and a low of 13.3 percent in 1984 to a high of 24.5 percent in 1998 for young females. A comparison of 1988 and 1999 reveals an increase in Social Security and SSI payments as a percent of income by 60.5 percent for young males and 36.6 percent for young females. These increases can in part be attributed to compositional changes in the work-limited population for both males and females. The decreased prevalence for males left a pool with more severe work limitations who were more likely to be eligible for DI and/or SSI payments. Furthermore, females who identified themselves as work limited as a result of the welfare reforms were more likely to receive SSI payments, resulting in an estimated 54.7 percent increase in DI/SSI payments to young females from 1988 to 1999. While welfare payments make up a very small share of income for males for the entire period, they make up a larger, but decreasing share for females. For young females the share of income attributable to welfare payments decreased from 7.9 percent in 1988 to 4.9 percent in 1999, a decline of 37.8 percent. Although welfare recipients may be entering the work limited population, they seem to also be opting to leave the welfare rolls in favor of DI/SSI. Increased SSI payments more than offset the decrease in welfare payments, over the period the sum of DI/SSI and welfare payments increased by 14.1 percent for females. 5. Family Poverty An important indicator of the standard of living of youth and young adults is the trend in their family income as a percent of the federal poverty threshold. Figures 13 through 16 reveal that work limited youth and young adults are much more likely to live in poverty than non-work limited youth and young adults. Between the peak economic years 1988 and 1999, the percent of male work limited 15-21 year olds with family income below the poverty threshold decreased 20.6 percent from 25.7 percent in 1988 to 20.4 percent in 1999. Despite this significant decrease, the 1999 family poverty rate for 15-21 year old work limited males was still substantially higher than the family poverty rate for non-work limited males in this age group, which remained constant between 1988 and 1999 at just under 14 percent. The percent of female work limited youth with family income below the poverty threshold was just over 24 percent in both 1988 and 1999, but had spiked to 36.2 percent in 1991. One partial explanation of why family poverty rates fell for young males with work limitations relative to rates for young females with work limitations is that the SSI expansion for children resulted in many more awards to young males than to young females. . In 1999 7.1 percent of males aged 15-21 received SSI compared to 5.1 percent of females in the same age group [22]. Non-work limited youth with family income below poverty remained close to constant in 1988 and 1999 at just above 17 percent, with a smaller spike in 1993 and 1994 to 21.4 percent. It is reasonable to assume that a significant proportion of 15-21 year olds are dependents within their family unit and that family income is primarily determined by the income of their parents. Therefore, the fact that the family income of work limited youth is more likely to be below the poverty line than the family income of non-work limited 15-21 year olds leads to important questions of causality. * Does the family’s poverty status contribute to the likelihood that a dependent in the family will have a health condition or disability that results in a work limitation? Do other factors specific to low income families that may contribute to the increased prevalence of work limited dependents? Such factors might include: lower quality of pre and post-natal care available to low income mothers; lower quality of child care and schools available to children in low income families; and the increased likelihood of risky behaviors (such as drug use) by the parents and the dependent. * Does the dependent’s physical or mental condition contribute to the family’s poverty status? Is family income lower because parental time that could be spent in the labor market is instead spent caring for the work-limited dependent? The low level of earnings by work limited young adults discussed in the previous section has led to an extraordinarily high family poverty rate in this population. Despite recent implementation of policies designed to improve the standard of living for persons with disabilities, the percent of work limited young adults with family income below the poverty threshold has increased from 1988 to 1999. The percent of young work limited males with family income below the poverty line increased from 22.0 percent in 1988 to 26.2 percent in 1999, an increase of 19 percent. While the increase was not as large for young females, they had a much higher family poverty rate at the beginning of the period, and it was still higher at the end of the period. The percent of young females living in families below the poverty threshold rose 7 percent, from 32.3 percent in 1988 to 34.6 percent in 1999, with the rate peaking at 37.5 percent in 1993. In comparison, 9.6 percent of non-work limited males and 15.5 percent of non-work limited females lived in families with income below the poverty threshold in 1999. 6. Health Insurance Coverage Long-term trends show a slight increase in the share of work limited youth and young adults with health insurance coverage. From 1988 to 1999 coverage increased by as little as 2.7 percent for 22-29 year old females and as much as 14 percent for 15-21 year old males (Figures 17 and 18). However, the public/private mix of coverage changed greatly over this period. i. Public health insurance coverage Public health insurance coverage includes Medicare and Medicaid coverage. Medicaid is a jointly funded, Federal-State health insurance program that primarily serves low-income elderly, families with children, and individuals with disabilities. Access to Medicaid increased in the 1990s through the SSI, TANF and CHIP programs. Medicare is a health insurance program for persons or spouses of persons who worked for at least 10 years in Medicare-covered employment and are over 65 years old or have received Social Security Disability Insurance benefits for at least 24 months. Relatively few young people with work limitations have Medicare coverage. We have combined these two programs for reporting purposes due to likely confusion among respondents about which of these two programs they or their family member is enrolled in. Public health insurance coverage among youth with work limitations increased dramatically after the 1990 Supreme Court ruling in Sullivan v. Zebley expanded SSI eligibility criteria for children. From 1988 to 1999 the share of work limited 15-21 year old males and females with only public health insurance coverage increased by 51.2 percent and 33.5 percent respectively (from 25.7 percent to 38.8 percent for males and from 30.5 percent to 40.7 percent for females). The larger increase for males likely reflects the fact that 64 percent of SSI child recipients are male [22]. Coinciding with expanded access to Medicaid in the 1990s, from 1988 to 1999, the shares of 22-29 year old males and females with only public health insurance coverage increased by 32 percent and 18.4 percent respectively (from 29.3 percent to 38.7 percent for males and from 38.7 percent to 45.8 percent for females). Probably due to the relationship between Medicaid and welfare benefits, work limited females are more likely than work limited males to have only public health insurance coverage. Furthermore females in the older age group are more likely than younger females to be covered by Medicaid or Medicare. ii. Private health insurance coverage Due to limitations in the CPS and the lack of consistent questions about private health insurance, we had to construct a proxy measure for private health insurance coverage. We included as covered by private health insurance all persons with health insurance coverage through their own employer, their spouse if living in the same household, and their children or stepchildren under the age of 25 and living in the same household. While this proxy measure might misrepresent the level of coverage in a given year, we suspect that trends in this measure would be quite similar to those in a more accurate measure. Based on the proxy measure, the share of work limited youth and young adults with only private health insurance coverage declined substantially for youth and young adults with work limitations. The smallest decline, 16.6 percent, was for males aged 15-21 followed by a decline of 22.9 percent for females in the same age group (from 38.1 percent to 31.8 percent for males and from 38.4 percent to 29.6 percent for females.) Among young adults the share covered by private health insurance declined by 36.2 percent and 25.6 percent for males and females respectively (from 30.1 percent to 19.2 percent for males and from 31.7 percent to 23.5 percent for females.) Young adults are less likely to be covered by only private health insurance than youth. The reason a smaller share of the older group has private health insurance coverage is that a large majority of the 15-21 year olds with coverage are, according to our definition, covered because a parent is covered. We do not extend this assumption to 25-29 year olds. Therefore, even though many more 25-29 year olds have own-employer insurance than 15-21 year olds, the increase in this coverage is not enough to offset the decrease in coverage under parental policies. iii. Relationship between employment and health insurance coverage Employed work limited young adults are less likely to have health insurance provided by their employer than employed non-work limited young adults. However, Figure 19 shows that between 1982 and 2000 the trend for both work limited and non-work limited 22-29 year olds was toward a decrease in employer provided health insurance. Although there is much variability between 1988 and 1999, a comparison between these two peak economic years reveals that while employed work limited males experienced a 26.7 percent decrease, the percent of employed work limited females with employer provided health insurance rose 14.1 percent. Over this period the percent of employed non-work limited young adults whose employer paid for health insurance decreased by about 10 percent for both males and females. Since the mid-1980’s not employed, work limited young adults were more likely to have health insurance coverage than employed, work limited young adults, with the likelihood increasing from 1988 to 1999 (Figures 20 and 21). However, the coverage mixes for employed and not employed young adults differ greatly. While a large (but decreasing) majority of employed persons have private health insurance coverage, a large (and increasing) majority of not employed persons have public coverage. Public health insurance coverage increased for both employed and not employed work limited young adults between 1988 and 1999. The most dramatic increase (58.9 percent) occurred among employed work limited young males (from 12.7 percent to 20.2 percent). This might in part be attributed to the decreased prevalence of work limitations among young males. The decreased prevalence likely left a pool of males with more severe work limitations, who, although employed, were still eligible for Medicaid. Furthermore, the decreased prevalence also meant that males who did not qualify for Medicare (that is, males who did not qualify for DI benefits) were less likely to be in the work limited population in the late 1990s, leaving a greater percent of employed work limited males who qualified for Medicare. Employed and not employed females had smaller increases in public health insurance coverage of 13.8 percent and 16.4 percent respectively. While the welfare reforms may have contributed to an increased prevalence of work limitations among welfare recipients (mostly female) who qualify for Medicaid, the total effect could have been dampened because a large portion of these females would not be eligible for Medicare due to the lack of sufficiently long work history, thus decreasing the share of Medicare coverage for employed and not employed females. 7. Marriage Work limited young adults (22-29) are less likely to be currently married than non-work limited young adults (Figure 22). Declines for both groups reflect increases in the share never married (Figure 23). They also reflect trends in divorces/separations relative to the number ever married (“dissolution rate”), but these vary by sex and work limitation status (Figure 24). The dissolution rate for males without work limitations is very stable throughout the period. The rate is higher, but also stable, for males with work limitations until 1993, hovering around 22 – 23 percent. It then declines sharply to 17 percent by 1995, when it reverses and increases to 27 percent by 1999. The decline roughly coincides with a period of growth in prevalence of work limitations for males in this age group, and might reflect a compositional change; during this period, growth in prevalence probably reflects the effects of the economy on employment, and those who enter this population were perhaps more likely to be currently married than those already in this population. The reverse might have happened as the economy expanded – those men left behind in the work-limited population might have had less success in both the labor market and the marriage market. It is interesting, however, that we did not see a similar phenomenon during the expansion of the 1980s. The dissolution rate for females without work limitations declined very gradually over the entire period. The behavior of the dissolution rate for females with work limitations is quite different, however, and is almost the mirror image of that for males. The latter difference likely reflects the role of welfare reform in prevalence trends for females. There was substantial growth in the rate for those with limitations from 1988 (28 percent) to 1995 (34 percent), followed by a reversal and larger decline to 1999 (24 percent). We would expect the business cycle, by itself, to have a compositional effect that is similar to that for men, but welfare reform would have the opposite effect, at least in the early period. That is, movement of welfare recipients into the work-limited population might have caused the increase in the dissolution rate for females in this age group. That would not explain the reversal later on, however. One possible explanation of the later decline is that the reforms themselves eventually encouraged more mothers in this group to marry. One of the objectives of the reforms was to increase the incentives for parents to marry. Time limits, work requirements, and paternity reporting requirements all made marriage a more attractive option for unmarried parents with an opportunity to marry. To assess this possibility further, we briefly consider marriage rates for work limited females aged 22-39. We use the larger age group to increase the precision of estimates by welfare status. Among welfare recipients there has been an increase in marriage rates since 1994 (Figure 25), and in a decline in dissolution rates (Figure 26). In contrast, trends for work-limited females who are not welfare recipients are in the opposite direction. For persons at low income levels, decisions about marriage and divorce may indeed be closely linked to the financial implications of each. Policies designed to alleviate poverty such as AFDC/TANF, SSI and the EITC may have both intended and unintended impacts on marital status. It is possible that these changes contributed to changes in marriage rates for both males and females with work limitations, even though those changes are in opposite directions. Consider a young couple with young children, prior to welfare reform and the expansion of the EITC. Neither parent is work-limited, the father has a low-income job and the mother is not working. Now consider what happens if the father becomes work-limited due to an accident or illness. He is eligible for some disability benefits, but less than he was earning at his job. The mother could supplement family earnings through work, but she and the children would be eligible for AFDC benefits to the extent that the father’s disability income is not counted against those benefits.14 After welfare reform and the expansion of the EITC, however, the scenario would be different in several ways. First, TANF time limits and work requirements increase the incentive for the mother to work. Second, by doing so, she will become eligible for EITC payments on the basis of her earnings, but the amount would be reduced to the extent that the father’s disability income contributes the family’s adjusted gross income. Third, if the father’s disability income includes some SSI payments, her earnings might reduce the amount he receives.15 Therefore, welfare reform combined with the expanded EITC may create an incentive for separation or divorce for this couple after his disability onset because doing so would improve their financial situation. That is, after separation or divorce, assuming the children remain with the mother, the father’s SSI would not reduce the mother’s EITC, and the mother’s EITC would not reduce the father’s SSI. Now consider the same scenario, but this time the mother experiences disability onset, not the father. She would likely become eligible for SSI, and the amount of her payment might be higher if she were divorced than if she remained married. That would be true both before and after welfare reform, creating an incentive to divorce. However, due to the expanded ETIC, the family would also receive EITC based on the father’s earnings as long as the mother and father remain married (assuming again that the children would stay with the mother in the event of a divorce or separation). Thus, compared to the pre-reform period, there is less incentive for the husband to leave the family, and it is more likely that the couple will remain married. The welfare reform might therefore increase the chance that a married mother who experience’s disability onset remains married while reducing the chance that a married father who experience’s disability onset remains married. The asymmetry of the results for males and females reflects differences in traditional sex roles for males and females. The outcomes highlighted by these scenarios are supported in the data. Throughout the 1990s the probability of divorce or separation increased for not employed work limited males and decreased for not employed work limited females (Figure 27). In fact, in 1997, not employed work limited men surpassed not employed work limited females to become the most likely to divorce or separate. Work limited SSI recipients have a higher probability of divorce or separation than non-recipients (Figure 28). Furthermore, the probability of divorce or separation increased for male SSI recipients after 1997 but continued to decline for female recipients. 8. Parenting The CPS does not collect data about childbearing, but it does ask about relationships between adults and children within a household. Relationship data for household members allow us to determine whether an adult has own children living in the household. Except in the case of the head of household, however, this information does not allow us to distinguish between own children, step children, and, in some instances, nieces and nephews.16 Nor can we identify parents whose children live in a different household. Because children are much more likely to live with their mothers than their fathers when parents live separately, our “males with children (in household)” variable is undoubtedly less accurate as a measure of the proportion of males who are biological or adoptive fathers than our “females with children” variable is as a measure of females who are biological or adoptive mothers. The series we present begin in 1989, because the data were not consistent prior to 1988. As with other series, we report three-year moving averages. i. Females with Children From 1989 to 1999 there was a sharp increase in the percent of young (aged 22 – 29), unmarried work limited females with children, from 18.5 percent to 23.4 percent (Figure 29).17 Over this same period the percent of young, unmarried non-work limited females with children increased slightly, from 12.3 percent to 14.9 percent. The figure makes clear that young work limited mothers are much more likely to be single than non-work limited mothers. In fact, by 1993 just under half (49.6 percent) of young work limited females living with their children were single; by 2000 this figure had increased to 58.4 percent. Over the same period, the corresponding figure for non-work limited young females was under 35 percent. The finding that single young females with work limitations are much more likely to have children than single young females without work limitations is supported by data from the National Longitudinal Transition Study, a longitudinal, nationally representative survey of youth with disabilities. The results of this survey show that in the early 1990s 20 percent of young single females with disabilities had children [26]. The prevalence of single mothers among females with work limitations may be due in part to the fact that people with disabilities have less education and lower paying jobs on average; this in turn has a negative effect on their marriage market opportunities. Further, in the event of a non-marital birth, the possible need to qualify for TANF or SSI creates a disincentive to marry. The increase in single work limited females with children may not necessarily reflect a change in behavior of females with disabilities; that is, one cannot conclude from these statistics that more females with disabilities are having non-marital children or divorcing after having children. Rather, the increase in single mothers with work limitations may in part reflect an increase in single mothers reporting work limitations possibly due to the welfare reforms in the 1990s. Single females with children are the most affected by the welfare reforms; therefore they are most likely to seek to avoid the stricter welfare rules (such as work requirements, time limits and cooperation with paternity establishment) and/or to apply for SSI. For single mothers on welfare who, prior to welfare reform, had been receiving informal child support from their child’s father, there is an additional reason for an increased incentive to self-identify as work limited and apply for SSI as a result of the reform. Paternity establishment and child support reporting requirements included in the reforms only apply to TANF recipients, so these mothers can escape those requirements if they can move from TANF to SSI. More research is needed to determine the extent to which work limited females are more likely than others to become single mothers, and the extent to which becoming a single mother increases the chance that the mother will identify herself as work limited. ii. Males with Children Figure 30 shows the percent of males aged 22-29 by marital status who live with their children. Because a majority of children live with their mothers if the parents do not live together, the percent of single males (both work limited and non-work limited) who live with their children is very low. It is interesting, however, that the percent of work limited young males who are both single and living with their children is generally higher than the corresponding figure for non-work limited young males. A reason for this might be that disability income, including possible dependent benefits, reduces their opportunity cost of staying out of the labor force to care for their children. iii. Teenage Childbearing By using the relationship data and the ages of the females and their children, we were able to approximate the percent of 15-29 year old females who gave birth to at least one child as a teenager. It is important to note that the females who reported work limitations at the time of the CPS interview did not necessarily have the disability or health condition as a teenager, when conception and birth occurred. Starting in the mid 1990s there was a very large increase in the percent of work limited females aged 22-29 who had given birth as a teenager, this increase coincided with welfare reform (Figure 31). Between 1989 and 1999 the percent of work limited 15-29 year old females who had a teen birth increased by 47.1 percent, from 13.9 percent to 20.5 percent, compared to a 2.0 percent increase in the non-work limited population. Almost all of the increase was after 1994. This is the strongest evidence we have seen that welfare reform induced many young mothers to identify themselves as work limited. All of the females who were 22 to 29 in 1996 had aged out of the teenage population by 1993, and most had aged out earlier. Thus, it is very unlikely that welfare reform had an impact on their childbearing as teenagers. Instead, it seems much more likely that welfare reform encouraged many young females who had given birth as teenagers to identify themselves as work limited. One other possibility is that the post Zebley expansion in SSI eligibility for children, including teenagers under 18, contributed to this growth. As a result of the expansion, more teenage girls had SSI benefits, and it is possible that the benefits induced some to give birth18 and others to identify themselves as work limited. If that were true, however, we would expect to see an increase in teen births for females 15 – 21 in the early 1990s, but we do not (Figure 32). There is an increase for that age group in the late 1990s (after an unexplained decrease in the mid 1990s), but this is more consistent with the hypothesis that welfare reform induced more females who had given birth as teens to identify themselves as work limited. To further assess the cause of this increase, we examine the teenage birth series for females aged 25 – 29 (Figure 33). There is a 44.5 percent increase in this series from 1989 to 1999, with all of the increase occurring after 1994. Everyone in this population in 1995 turned 18 in 1988 or earlier (i.e., before Zebley), and turned 20 in 1990. Hence, it is very unlikely that their childbearing as teens was affected by welfare reform in the mid 1990s or by the post-Zebley SSI expansion for children, or that their self-identification as work limited is due to the post-Zebley expansion. There doesn’t seem to be a credible alternative to the hypothesis that welfare reform induced many females who had given birth as teens to identify themselves as work limited. 9. Living Arrangements A primary goal of recent disability policy has been to facilitate independent living for persons with disabilities. Despite this goal there has been little change between 1989 and 1999 in the living arrangements of work limited young adults as shown in Figures 34 and 35. While young males with work limitations are much more likely to live with their parents than young females, due in part to their lower marriage rates, they are no less likely to do so in 1999 than they were in 1989, with 43 percent of young males with disabilities living with their parents in both years. Between 1989 and 1999, there was a slight decrease in the percent of young females with work limitations who lived with their parents, from 26 percent to 23 percent. It is possible, however, that the lack of change for young males with work limitations reflects a move toward greater independent living. If the prevalence of work limitations among young males decreased in the mid and late 1990s because of increased education and employment opportunities, then it might be that those males left in the work-limited population have more severe medical conditions or functional limitations, on average, than those who were in this population in 1988. That, alone, would lead us to expect an increase in the percent who live with parents, so no change would be an indication of progress. While the decline for in the percent of females with work limitations living with their parents might be sign of progress for similar reasons, we suspect that the main reason is for the decline is increased prevalence of work limitations among welfare mothers who live independently with their children. The percent of young males who live with their spouse and/or children decreased between 1989 and 1999, from 27 percent to 22 percent. At the same time, however, this statistic increased for young females, from 47 percent to 50 percent. The differences between the male and female statistics reflect the differences in marriage rates and parenting rates, discussed earlier. While it has been a policy priority to facilitate independent living by increasing employment opportunities for persons with disabilities, these statistics suggest that many young females with work limitations take a path to independence from their parents that is less likely to be available to young males with work limitations: they marry and/or have children. V. Summary and Conclusions This paper is a first attempt at measuring long-term trends in outcomes for youth (aged 15-21) and young adults (aged 22-29) with disabilities, from 1981 through 2000. Our specific objective was to assess the extent to which those in the non-institutionalized population with work limitations are starting to participate more fully in major social activities, enjoy a higher standard of living, achieve greater economic independence, and live more independently. We used data from the Current Population Survey (CPS) for 1981 through 2001 and focused on changes between the peak business cycle years of 1988 and 1999. Over that period, the prevalence of work limitations (i.e., the percent of the population reporting work limitations) decreased by 16 percent for males aged 15-29 and increased by 6 percent for females. Males aged 25-29 experienced the greatest decline in prevalence, decreasing by 28 percent. The increased prevalence for females is mostly due to a large increase for black females (36 percent). Changes in prevalence could be caused by changes in personal characteristics of the population, such as age and the prevalence of conditions that contribute to work limitations. Changes could also reflect changes in environmental factors that affect who considers themselves work limited, conditional on personal characteristics. For males, environmental factors that may have contributed to the decline include: changes in education policy; the economic expansion; the implementation of the ADA; and advances in both treatment of conditions and assistive devices. While these forces likely affect females too, it appears they are more than offset by the effects of welfare reform and long-term increases in labor force participation, both of which contribute to an increase in prevalence among females. Changes in outcomes for the work limited population could be due to: i.) changes in outcomes for people with given characteristics, or ii.) compositional changes in the work limited population (i.e., changes in personal characteristics of the work limited population, some induced by environmental factors, others affected by underlying population trends). We can at best only partially assess which of these two sources of change the outcome trends reflect. Furthermore, changes in characteristics of the work limited population that are due to environmental change cannot be readily distinguished from those due to underlying changes in population characteristics. However, we find substantial evidence that environmental changes have had considerable effects on a number of outcomes. The outcome trends show substantial progress in reducing the gap in educational attainment between those with and without work limitations. At the same time, however, the employment rates for work limited young males and females declined, by 25 percent and 12 percent, respectively. While compositional change might explain a substantial share of this decline, the data indicate that at least some of the decline is due to declines conditioned on personal characteristics. The income gap between work limited and non-work limited young adults widened over the period. Work limited young adults experienced a decrease in real earnings and AFDC/TANF payments as a share of total income, accompanied by partially offsetting increases in SSDI and SSI. The share of work limited young adults living in poverty also increased. There has been a slight increase in the share of work limited youth and young adults with health insurance coverage, but the more substantial trend is the large increase in public coverage and large decrease in private coverage. Our results show that young adults with work limitations are less likely to be married and more likely to be divorced than non-work limited young adults. Over the period there was substantial growth in the share of young work limited females who are unmarried mothers, as well as a sizeable (158 percent) increase in the share who had their first child as a teenager. There was no change in the share of work limited young males living independently from their parents and only a slight decrease for work limited young females. Taken at face value, the decline in the employment rate, decreased earnings, increased poverty, increased reliance on income support, and increased public health insurance coverage for work limited youth and young adults call into question the success of policies aimed at greater economic independence and a higher standard of living for persons with disabilities. Although efforts to close the educational gap between those with and without disabilities appear to have had some success, this success has not translated into better outcomes for young adults. In addition, young females with disabilities are increasingly likely to have a marriage end in separation or divorce, to be a single mother, and to have given birth as a teenager. This dire interpretation of the findings assumes that these trends are not due to changes in the personal characteristics of young persons with work limitations. It is possible however, that the decline in prevalence for males reflects environmental changes that caused young males with less severe medical conditions to stop perceiving themselves as work limited. The result would be that, on average, those in the work-limited population in recent years have more severe conditions than in the past. If this is correct, the reduction in the education gap is even more impressive, and adverse trends in other outcome measures for males are less troubling. Perhaps if we could adjust for changes in the severity of medical conditions of the work limited population, we might even see improvement in some of the outcome measures. Our analysis suggests that the decline in the employment rate for young males with work limitations cannot be fully attributed to increases in the severity of their medical conditions or to other changes in their personal characteristics. We cannot, however, rule out the possibility that trends in personal characteristics did play an important role. Certainly there were changes in the environment of the sort that might reduce the prevalence of work limitation, especially among those with less severe conditions. The size of the decline in prevalence, however, is not large enough to explain the entirety of the adverse outcome trends. Furthermore, a review of the extensive evidence on the causes of the employment rate decline for older males with work limitations suggests that the main causes are factors other than increased severity of medical conditions.19 The most credible evidence for older males with work limitations points to eligibility expansions for the DI and SSI programs, along with growth in the income replacement rate for those DI beneficiaries who had the lowest earnings prior to entering the program. Other evidence indicates that the ADA itself is the cause; allegedly, the costs of potential litigation in anti-discrimination lawsuits by employees with disabilities, along with the costs of accommodations, have discouraged employers from hiring people with disabilities. Many dispute this interpretation of the evidence, however. If either, or both, of these factors had an adverse impact on employment and other outcomes for older adults with disabilities, we would expect them to have had an adverse impact for younger adults with disabilities, too. In fact, it is possible that their impact on outcomes for younger adults with disabilities would be especially large, because of the nature of the eligibility expansions, the relatively low wages of younger workers, and the likelihood that any adverse impact of the ADA would be on hiring, rather than terminations. Growth in prevalence of work limitations for young females stands in stark contrast to declines in prevalence for young males and is hardly suggestive of increases in the severity of impairment. As we have seen, however, there is another explanation of both prevalence growth and adverse outcome trends for young females: welfare reform apparently induced many young, unmarried mothers to perceive themselves as work limited. Long-term growth in female labor force participation likely also increased the share of young females who perceive themselves as work limited. Females entering the work limited population because of these forces might typically have less severe medical conditions than others in this population, but nonetheless be less likely to work and more likely to depend on public programs. More research is needed on the causes of change in prevalence for young males and females, and the contributions of those causes to outcome trends. More specifically, research is needed to determine whether the changes in prevalence are the result of: i) changes in the prevalence of medical conditions that contribute to work limitations, or ii) changes in the environment that affect self-perceptions about work limitation. For environmental factors, it is important to distinguish between the roles of the childhood environment (e.g., schooling and income support for children with disabilities), and the young adult environment (e.g., work incentives and access to medical care). For females, we are convinced that welfare reform is a major cause of increased prevalence, but is it the only cause? In the absence of welfare reform, would there have been a decline in prevalence, similar to that for males? How do changes in the medical conditions of females with work limitations compare to changes for males?20 Perhaps the most disturbing finding from this analysis is that the observed reduction in the education gap for work limited young males and females is not accompanied by a closing of gaps in employment, economic dependence, and other outcomes. In the 1990s, young people with work limitations had attained substantially more education relative to non-work limited people in the same birth cohort than had earlier cohorts of young people with work limitations. Could factors affecting the trends in prevalence be both closing the education gap and increasing gaps for other outcomes? While this is possible, it is counter-intuitive. If this is not the explanation it would appear that educational reforms for children and youth with disabilities succeeded in increasing the most obvious barometer of educational outcomes, educational attainment, without measurably improving social outcomes for young adults. Is that because, despite the increase in attainment, schools have not succeeded in providing youth with disabilities the skills they need to support themselves as adults? Could the movement toward an integrated educational system for all students actually be decreasing post-education employment opportunities for young adults with disabilities by placing less emphasis on vocational and life skills? Or have skills increased, but young adult outcomes deteriorated because of other factors? Has increased income support for children with disabilities, through SSI, fostered increased dependence as adults? Have policies that appear to have adversely affected employment outcomes for older adults with disabilities offset gains in education? Our findings concerning welfare reform add to a growing body of evidence that the policy environment has a substantial influence on self-perceptions of disability or, more specifically, work limitations. This is not surprising when disability is viewed as the outcome of an interaction between a person’s characteristics and environment. The finding reinforces the importance of better understanding the separate contributions of medical conditions and the environment to disability, and the central role that public policy appears to play in the latter. These same findings also add to growing evidence of substantial interactions between welfare policy and disability policy. That evidence leads us to wonder whether current policies are fostering integration of unmarried, low-income parents with disabilities into mainstream social activities, reflecting the goals articulated in the ADA, or are instead segregating them from other unmarried, low-income parents and fostering long-term dependence on programs (SSI and DI) that were not purposefully designed to support their needs as unmarried parents (e.g., through provision of transitional child care or enforcement of child support). The finding that young work limited females are much more likely than their non-work limited peers to be unmarried mothers, and to be parenting a child borne when they were a teenager, leads to an important question of causality: are female youth with medical conditions that might lead to work limitations more likely than others to become single mothers, or, conversely, does becoming a single mother increase the chance that a young female will identify herself as work limited? If the former, are policies to reduce teenage childbearing effectively targeted at female youth with such conditions? If the latter, does single motherhood cause or exacerbate medical conditions (e.g. major depression) that could lead to work limitation? Or is it simply that welfare policy encourages them to identify themselves as work limited? Finally, the lack of decrease in prevalence of work limitations for young black males and the large increase for young black females deserves more study. The increase among black females may be partially due to the effects of welfare reform. However, the lack of decrease among black males over a period when prevalence decreased for white and Hispanic males is an indication that something else might be working to elevate the prevalence of work limitations in the black population. Acknowledgements We greatly appreciate the significant assistance provided by Andrew J. Houtenville in the production of the statistics we present in this paper. We are also grateful for the research assistance of Jennifer Duffy. This research was funded by the U.S. Department of Education’s National Institute on Disability and Rehabilitation Research through its support for the Rehabilitation Research and Training Center for Research on Employment Policy for People with Disabilities at Cornell University (cooperative agreement No. H133B980038). The findings and conclusions are our own, and do not necessarily reflect the views of the National Institute on Disability and Rehabilitation Research. References [1] A. Beck and P. Harrison, “Prisoners in 2000” Bureau of Justice Statistics Bulletin, August (2001), NCJ 188207. [2] A. Houtenville, and M. Daly, “Employment Declines Among People with Disabilities: Population Movements, Isolated Experience, or Broad Policy Concern?” in: The Decline in Employment of People with Disabilities: A Policy Puzzle, D. Stapleton and R. Burkhauser, eds., WE Upjohn Institute, Kalamazoo, Michigan, 2003, Chapter 3. [3] B. Garrett and S. Glied, “Does State AFDC Generosity Affect Child SSI Participation?” Journal of Policy Analysis and Management, Spring (2000), 275-295 [4] B. Guy, H. Shin, S.Y. Lee and M. Thurlow, State Graduation Requirements for Students With and Without Disabilities, Technical Report No. 24, Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes, 1999. http://education.umn.edu/nceo/OnlinePubs/Technical24.html (accessed April 2003). [5] C. Kirchner, “Looking Under the Street Lamp: Inappropriate Uses of Measures Just Because They are There,” Journal of Disability Policy Studies 7(1), 1996, pp. 77-90. [6] Committee on Ways and Means, U.S. House of Representatives, 1996 Green Book: Background Material and Data On Programs Within The Jurisdiction Of The Committee On Ways and Means, 1996. [7] D. Acemoglu and A. Joshua, “Consequences of Employment Protection? The Case of the Americans with Disabilities Act,” Journal of Political Economy, 109 (5), 2001, pp.915-57. [8] D. Gilliard and A. Beck, “Prisoners in 1997,” Bureau of Justice Statistics Bulletin, August (1998), NCJ 170014. [9] D. Kruse and L. Schur, “Employment of People with Disabilities Following the ADA,” Industrial Relations 42(1), 2000, pp. 31-66. [10] D. Stapleton, D. Wittenburg, M. Fishman, and G. Livermore, Transitions from AFDC to SSI Before Welfare Reform, Social Security Bulletin 64(1), 2001/2001, pp. 84-114. [11] D. Wittenburg, and E. Maag, "School to Where? A Literature Review on Economic Outcomes of Youth with Disabilities," Journal of Vocational Rehabilitation, (forthcoming). [12] H. Hayghe, “Development in Women's Labor Force Participation,” Monthly Labor Review 120(9), 1997, pp. 41-47. [13] J. Kubik, “Incentives for the Identification and Treatment of Children with Disabilities: the Supplemental Security Income Program,” Journal of Public Economics 73 (1999), 187-215. [14] J. McNeil, “Employment, Earnings and Disability,” Presented at the 75th Annual Conference of the Western Economic Association, 2000. http://www.census.gov/hhes/www/disable/emperndis.pdf. (Accessed March 2003). [15] M. Hardy and E. Pavalko, “The Internal Structure of Self-Reported Health Measures Among Older Male Workers and Retirees.” Journal of Health and Social Behavior 27(4), 1986, pp.346-357. [16] National Institutes of Mental Health, The Numbers Count: Mental Disorders in America, NIH Publication No. 01-4584, 2001. http://www.nimh.gov/publicat/numbers.cfm. (Accessed February 2003). [17] P. Blanck, S. Schwochau, and C. Song, “Is It Time to Declare the ADA a Failed Law?” in: The Decline in Employment of People with Disabilities: A Policy Puzzle, D. Stapleton and R. Burkhauser, eds., WE Upjohn Institute, Kalamazoo, Michigan, 2003, Chapter 9. [18] R. Burkhauser, A. Houtenville and D. Wittenburg, “A User Guide to Current Statistics on the Employment of People with Disabilities,” in: The Decline in Employment of People with Disabilities: A Policy Puzzle, D. Stapleton and R. Burkhauser, eds., WE Upjohn Institute, Kalamazoo, Michigan, 2003, Chapter 2. [19] R. Burkhauser and D. Stapleton, “A Review of the Evidence and Its Implications for Policy Change,” in: The Decline in Employment of People with Disabilities: A Policy Puzzle, D. Stapleton and R. Burkhauser, eds., WE Upjohn Institute, Kalamazoo, Michigan, 2003, Chapter 11. [20] R. Burkhauser, M. Daly, A. Houtenville, N. Nargis, “Self-Reported Work Limitation Data: What They Can and Cannot Tell Us,” Demography, 39(3), 2002, pp. 541-555. [21] S. Kaye, “Employment and the Changing Disability Environment” Mimeo, Disability Statistics Center, University of California, San Francisco, 2002. [22] Social Security Administration, Office of Research, Evaluation and Statistics, Children Receiving SSI, December 2000, 2001. [23] T. DeLeire, “The Wage and Employment Effects of the Americans with Disabilities Act,” Journal of Human Resources. 35(4), 2000, pp. 693-715. [24] T. Hale 2001. “The Lack of a Disability Measure in Today’s Current Population Survey,” Monthly Labor Review June (2001), 38-40. [25] T. Thompson, P. Holcomb, P. Loprest and K. Brennan, State Welfare-to-Work Policies for People with Disabilities: Changes Since Welfare Reform, Urban Institute, Research Report, 2000. http://www.urban.org/url.cfm?ID=1000043. [26] The National Longitudinal Transitions Study: A Summary of Findings. 1997. http://www.sri.com/policy/cehs/publications/dispub/nlts/nltssum.html. (Accessed March 2003). Appendix I: Compositional Change Model An important question is whether changes in the prevalence of work limitations is due to actual changes in the prevalence of the medical conditions and functional limitations that can contribute to work limitations, or to changes in the environment that affect self-reporting of work limitations. We would also like to know the extent to which observed changes in outcomes for the work-limited population are explained by compositional changes caused by environmental factors that have reduced prevalence. In this appendix we use simple mathematical models to explore the extent to which such compositional changes can possibly account for the decline in one important outcome, the employment rate. We begin with a model for males, and then consider a necessarily more complex model for females. Males We define the employment rate for work limited males in year t as: (1) , where Et denotes the number of work limited persons who are employed in year t, wt denotes the share of the population that is work limited in year t, and Nt denotes the population in year t. For purposes of this exercise, we calculate the largest decrease in employment rates that could be explained by the same factors that caused the decrease in prevalence of work limitations among young men, regardless of what they were. The size of the effect could be no larger than the decline that would occur under the following extreme assumptions. From year t-1 to year t, 1) there is a decline in prevalence solely because some males who would have considered themselves work limited under the environment of t-1 do not consider themselves work limited under the environment of t, and 2) all such individuals would be employed under either environment. In the absence of any other change, the employment rate for year t would be: (2) where d = wt-wt-1 and d < 0 if, as assumed, prevalence declines. Hence, et* - et-1 is the largest decline that could be explained by the factors that caused the decline in prevalence. If et < et* (that is, the actual decline is greater than the decline predicted under these assumptions) then there must have been some factor depressing the employment rate other than the same factors that reduced prevalence. If et > et* (that is, the actual decline is not as great as the decline predicted under these assumptions), then it is possible that the decline in the employment rate is entirely caused by the same factors that reduced prevalence, and there might even have been factors causing a partially offsetting increase. 15-21 year males For this group, there is essentially no change in prevalence from 1988 to 1999 (the rate is 2.5 percent in both years), so it appears that the cause of the decline in the employment rate is factors that did not affect prevalence for males in this age group. 22-29 year males For this group, there is a significant decline in prevalence. Based on age-adjusted data for males 22 to 29 years old: w1988 = .040, e1988 = .589, w1999 = .030, e1999 = .451, and d = -.01. Filling in equation (2) we get: (4) , or 45.2 percent. This is only slightly larger than the 1999 CPS estimate, which is 45.1 percent. Thus, it is mathematically possible that the decline in the employment rate for males in this age group is caused entirely by compositional changes resulting from the factors that reduced prevalence. Females As discussed in the text, the issue is more complicated for females because, while factors that reduced prevalence for males likely had a similar effect for females, welfare reform and long-term growth in labor force participation of females likely had a positive effect on prevalence for females, but not males. If exits from the population due to work occurred at the same rate for females as for males in the same age group, then d for females is the same as for males. We define the entrants into the work-limited population due to welfare reform and growth in labor force participation (LFP) as: (5) , where d is assumed equal to the value for males and wt and wt-1 are prevalence rates for females. Suppose that all females exiting the work-limited population were employed (as we assumed for males) and none of the females entering the work-limited population due to welfare reform or growth in LFP were working. The largest decline in the employment rate from year t-1 to year t that could be due to factors that reduced the prevalence of work limitations would be et* - et-1 where: (6) If these assumptions are correct and there are no other factors causing changes to the employment rate, then et = et*. If et < et*, then there must have been some factor depressing the employment rate other than compositional change caused by the same factors that reduced prevalence. If et > et*, then it is possible that the decline in the employment rate is entirely caused by compositional change due to factors that reduced prevalence, and there might even have been factors causing a partially offsetting increase. 15-21 year old females The age adjusted estimates of key parameters for females 15 to 21 years old for 1988 (t-1) and 1999 (t) are: w1988 = .020, e1988 = .466, w1999 = .022, and e1999 = .402. For this age group, there is essentially no change in prevalence for males (as discussed above), so d is essentially zero for females. Filling in equation (6) we get: (7) , or 42.3 percent This is larger than the actual 1999 value of 40.2 percent. That is, the actual decline in the employment rate was 2.1 percentage points greater than the greatest decline we would predict as a result of compositional change due to environmental factors that caused changes in prevalence. Thus, it appears that some other important factor must have contributed to the decline in the employment rate for females in this age group. 22-29 year old females The age adjusted estimates of key parameters for females 22 to 29 years old for 1988 (t-1) and 1999 (t) are: w1988 = .030, e1988 = .541, w1999 = .031, and e1999 = .479. Based on the analysis of males in this age group, d = -.010 Filling in equation (6) we get: (8) , or 20.1 percent. This is much smaller than the CPS estimate of 47.9 percent in 1999. That is, the assumptions built into the model imply a decline in the employment rate that is much larger than the actual decline. Thus, it is plausible that compositional change caused by environmental factors could explain the entire employment rate decline. In summary, the analysis in this appendix leads us to conclusions about the role that compositional change caused by environmental factors could plausibly have played in the employment rate decline that vary by age and sex. For both males and females aged 15 – 21, we conclude that some factor other than these explains the decline. For 22 – 29 year old males and females, such factors might have played a much more significant role. For the older females compositional change due to two identifiable factors – welfare reform and growth in female labor force participation – plus unidentified environmental factors that reduce prevalence for both males and females could easily explain the entire decline in the employment rate. For the older males, it is mathematically possible that environmental factors that reduced prevalence account for essentially the entire employment rate decline. As we discuss in the text, however, it seems more likely that other factors also contributed to the decline. Tables Table 1. Estimated Non-institutionalized Persons with Work Limitations Figure Captions Figure 1. Prevalence of work limitations among youth and young adults, 1982-2000 Figure 2. Prevalence of work limitations among male youth and young adults by race and Hispanic ethnicity, 1982-2000 Figure 3. Prevalence of work limitations among female youth and young adults by race and Hispanic ethnicity, 1982-2000 Figure 4. Prevalence of work limitations among male youth and young adults, 1982-2000 Figure 5. Prevalence of work limitations among female youth and young adults, 1982-2000 Figure 6. Percent of youth and young adults with no education and less than high school education, 1982-2000 Figure 7. Educational Attainment of work limited young adults, 1982-2000 Figure 8. Employment rates of youth, 1981-1999 Figure 9. Employment rates of young adults, 1981-1999 Figure 10. Employment rates of young adults by educational attainment, 1982-2000 Figure 11. Income of male young adults, 1981-1999 Figure 12. Income of female young adults, 1981-1999 Figure 13. Family poverty levels of male youth, 1981-1999 Figure 14. Family poverty levels of female youth, 1981-1999 Figure 15. Family poverty levels of male young adults, 1981-1999 Figure 16. Family poverty levels of female young adults, 1981-1999 Figure 17. Health insurance coverage by youth with work limitations, 1982-2000 Figure 18. Health insurance coverage young adults with work limitations, 1982-2000 Figure 19. Percent of employed young adults with employer provided health insurance, 1982-2000 Figure 20. Health insurance coverage by employment status for work limited male young adults, 1982-2000 Figure 21. Health insurance coverage by employment status for work limited female young adults, 1982-2000 Figure 22. Marriage rates among young adults, 1982-2000 Figure 23. Never married rates among young adults, 1982-2000 Figure 24. Dissolution rates among young adults, 1982-2000 Figure 25. Marriage rates among 22-39 year old work limited females by welfare status, 1982-2000 Figure 26. Dissolution rates among 22-39 year old work limited females by welfare status, 1982-2000 Figure 27. Dissolution rates among work limited 22-39 year olds by employment status, 1982-2000 Figure 28. Dissolution rates among work limited 22-39 year olds by SSI recipiency, 1982-2000 Figure 29. Female young adults with children by marital status, 1989-2000 Figure 30. Male young adults with children by marital status, 1989-2000 Figure 31. Percent of female young adults who had a teen birth, 1989-2000 Figure 32. Percent of female youth who had a teen birth, 1989-2000 Figure 33. Percent of 25-29 year old females who had a teen birth, 1989-2000 Figure 34. Living arrangements of male work limited young adults, 1989 Figure 35. Living arrangements of female work limited young adults Figures Figure 1. Prevalence of work limitations among youth and young adults, 1982-2000 Figure 2. Prevalence of work limitations among male youth and young adults by race and Hispanic ethnicity, 1982-2000 Figure 3. Prevalence of work limitations among female youth and young adults by race and Hispanic ethnicity, 1982-2000 Figure 4. Prevalence of work limitations among male youth and young adults, 1982-2000 Figure 5. Prevalence of work limitations among female youth and young adults, 1982-2000 Figure 6. Percent of youth and young adults with no education and less than high school education, 1982-2000 Figure 7. Educational Attainment of work limited young adults, 1982-2000 Figure 8. Employment rates of youth, 1981-1999 Figure 9. Employment rates of young adults, 1981-1999 Figure 10. Employment rates of young adults by educational attainment, 1982-2000 Figure 11. Income of male young adults, 1981-1999 Figure 12. Income of female young adults, 1981-1999 Figure 13. Family poverty levels of male youth, 1981-1999 Figure 14. Family poverty levels of female youth, 1981-1999 Figure 15. Family poverty levels of male young adults, 1981-1999 Figure 16. Family poverty levels of female young adults, 1981-1999 Figure 17. Health insurance coverage by youth with work limitations, 1982-2000 Figure 18. Health insurance coverage young adults with work limitations, 1982-2000 Figure 19. Percent of employed young adults with employer provided health insurance, 1982-2000 Figure 20. Health insurance coverage by employment status for work limited male young adults, 1982-2000 Figure 21. Health insurance coverage by employment status for work limited female young adults, 1982-2000 Figure 22. Marriage rates among young adults, 1982-2000 Figure 23. Never married rates among young adults, 1982-2000 Figure 24. Dissolution rates among young adults, 1982-2000 Figure 25. Marriage rates among 22-39 year old work limited females by welfare status, 1982-2000 Figure 26. Dissolution rates among 22-39 year old work limited females by welfare status, 1982-2000 Figure 27. Dissolution rates among work limited 22-39 year olds by employment status, 1982-2000 Figure 28. Dissolution rates among work limited 22-39 year olds by SSI recipiency, 1982-2000 Figure 29. Female young adults with children by marital status, 1989-2000 Figure 30. Male young adults with children by marital status, 1989-2000 Figure 31. Percent of female young adults who had a teen birth, 1989-2000 Figure 32. Percent of female youth who had a teen birth, 1989-2000 Figure 33. Percent of 25-29 year old females who had a teen birth, 1989-2000 Figure 34. Living arrangements of male work limited young adults, 1989 Figure 35. Living arrangements of female work limited young adults 1 For a review of literature on post-secondary school outcomes of youth with disabilities see Wittenburg and Maag forthcoming [9]. 2 The baby boom generation consists of persons born between 1946 and 1964, so the youngest members were 26 in 1990 and 36 in 2000. 3 Based on SSA [20], there were 174,000 recipients age 15 – 17. We did not find a published figure for age 18, but assume it must be on the order of the 54,000 figure for age 17. 4 White does not include white Hispanics and black does not include black Hispanics. 5 The incarceration rate for young males aged 25-29 increased by one percentage point between 1990 and 2000, from 1.5 percent to 2.5 percent. The incarceration rate for young females aged 25-29 is much lower and increased by a smaller absolute amount, from 0.10 percent in 1990 to 0.16 percent in 2000 ([1] and [8].) 6 The incarceration rate for black males of all ages increased from 2.4 percent in 1990 to 3.5 percent in 2000 while the rate for all white males increased from 0.34 percent to 0.45 percent ([1] and [8].) 7 Another possible factor contributing to the decrease in prevalence of work limitations among young males is the growth in the homeless population, many of whom presumably have disabling physical or mental conditions. Although not technically excluded, we assume that coverage of the homeless population is poor in the CPS. The effect of growth in the homeless population on prevalence measures for the non-institutionalized population depends on the extent to which movement of people with disabilities into this population is from the non-institutionalized population versus the institutionalized population. 8 Calculated from the CPS. 9 The black population made up 13 percent of the total US population in 1999. 10The series before and after 1992 are not entirely comparable, this is because there was a change in the way that the educational attainment question was asked. Because three years of data are pooled, the change in 1992 leads to a gap in our education series for 1991 and 1992. The data for “no education” is assumed to be compatible before and after 1992. Even if the entire decline in the percent of work limited youth with no education between 1990 and 1993 is due to the change in data collection, this is only 49 percent of the total decline between 1988 and 1997, that is, 51 percent of the decline occurred in years that the data collection was consistent. 11 Employment rates for non-work limited 15-21 year olds also decreased between 1988 and 1999 by 9.9 percent for males and 8 percent for females. Employment rates for non-work limited 22-29 year olds did not fall over this period. 12 Questions about employment are asked in year t about year t-1 and questions about educational attainment are asked in year t about year t. In this variable data about employment in year t-1 is getting matched with data about education in year t. Therefore, it is possible that in year t the respondent had a different level of education than in year t-1. However, the impact of this potential error is minimized in the comparison between persons with and without work limitations because both groups are affected by this lag issue. 13 The change in questions about educational attainment does not allow for comparisons between years before and after 1993. 14 Under AFDC, Social Security benefits were treated as unearned income reducing AFDC benefits by $1 for each $1 of Social Security benefits. However, the SSI recipient (whether a child or an adult) is not regarded as a part of the AFDC unit therefore, all income and resources of the SSI recipient are ignored in determining the AFDC benefit [6]. 15 The rules for SSI eligibility for married couples state that if only one spouse qualifies for SSI, part of the ineligible spouse's income is considered to be that of the eligible spouse. In order for each person in the couple to be treated as an individual, the couple must be separated or living apart for over 6 months [6]. 16 We constructed our parent identification variable from three relationship variables: i. head of household matched with child of head; ii. female/male spouse of head matched with child of head (this could be the spouse’s stepchild); iii. female/male child of head matched with oldest grandchild of head (this could be the niece or nephew of the child of head). 17 Not married includes: single, divorced, separated and widowed. 18 Conceptually, the effect of expanded SSI eligibility on births for qualifying teenagers is ambiguous. There might be a positive income effect – those who qualify have more income and choose to use that income to help support a child. However, AFDC support would have been available once the child was born anyway. In fact SSI offers a source of income without childbearing, so for newly eligible teenage females the SSI expansion might have offset the childbearing incentive effect of AFDC. See Garrett, A.B. and S. Glied 2000 [3], for an interesting analysis of how SSI expansions for children affected AFDC receipt. 19 This point, and the discussion in the following paragraphs, is based on Burkhauser and Stapleton (2003) [19]. 20 The authors are currently conducting a follow-up study that will address these and related issues through analysis of the National Health Interview Survey. 55