Research to Inform Policy: Contributions of the Mathematica Center for Studying Disability Policy

Introduction

The Social Security Disability Insurance (DI) and Supplemental Security Income (SSI) programs provide vital income support to some of our country's most vulnerable populations. The programs help alleviate poverty and facilitate access to low-cost health insurance for working-age people with disabilities and their dependents, as well as for low-income families of children with significant disabilities. DI was established in 1956, and SSI followed in 1974. Much has changed since then, and the programs must evolve to continue to meet the needs of the public, particularly nonelderly Americans with disabilities. To make reasoned decisions, administrators and policymakers must have information about program participants and about the external and operational factors that affect the programs and the populations they support. The Social Security Administration (SSA) established the Disability Research Consortium (DRC) in 2012 to provide high-quality objective information to help guide policy decisions. In the DRC's first 6 years, its researchers developed substantial new information that has provided insights toward improving the circumstances of people with disabilities and the design and administration of the programs that serve them.

This article summarizes selected findings from DRC work conducted by researchers at the Mathematica Center for Studying Disability Policy (CSDP), often in collaboration with other institutions, most notably SSA, the University of Illinois at Chicago, the University of Massachusetts Medical School, and Virginia Commonwealth University. (Formerly known as Mathematica Policy Research, the organization shortened its name to “Mathematica” in 2019.) The research addresses five broad topic areas: DI applicants and their ability to remain in the labor force; factors affecting participation in the federal disability programs; the characteristics, well-being, and employment of disability program participants; special populations of people with disabilities; and access to health insurance for people with disabilities.

DI Applicants and Their Ability to Stay in the Labor Force

The number of DI beneficiaries has nearly tripled over the last 30 years (SSA 2017a). Changes in labor force demographics account for some of that increase, but the growth has exceeded what would be expected based on demographics alone and has resulted in fiscal pressure on the DI trust fund. In 2015, the Old-Age, Survivors, and Disability Insurance (OASDI) trustees projected DI trust fund depletion in 2016 (SSA 2015b). In response, Congress enacted a temporary payroll tax reallocation from the Old-Age and Survivors Insurance (OASI) Trust Fund to the DI trust fund under the Bipartisan Budget Act of 2015; as a result, the projected date of DI trust fund depletion was deferred to 2022 (SSA 2015a). Since then, however, DI awards have declined steeply and unexpectedly; the OASDI trustees now project DI trust fund depletion in 2052 (SSA 2019).

The immediacy of the DI program's financial troubles in 2015 focused policymaker and researcher attention on the factors contributing to DI enrollment growth and on possible approaches to limiting that growth by helping DI applicants and potential applicants stay in the labor force. A growing body of research suggests that a well-targeted intervention, if occurring in the first few weeks after a medical condition threatens an employee's ability to work, can substantially improve the odds of job retention (Ben-Shalom, Christian, and Stapleton 2018). Too often, however, workers with medical problems do not receive prompt support that would help them to avoid suboptimal medical outcomes and preventable labor force withdrawals.

Under the DRC, CSDP researchers and their collaborators have added to the available evidence in two ways. First, they have generated new information on the characteristics and outcomes of DI applicants. Second, they have examined options for identifying workers at risk of preventable disability as early as possible, when a well-designed intervention is most likely to succeed.

Characteristics of DI Applicants and Their Application Outcomes

Three CSDP studies sought to help policymakers and program administrators anticipate changes in application volume and implement policies for documenting the characteristics and outcomes of DI applicants, all with the goal of helping workers with medical problems stay in the labor force. A fourth study assessed the use of private organizations that provide nonattorney representation services to DI applicants in early-intervention efforts.

Thompkins and others (2014) examined the demographic, employment, and program-participation characteristics of DI applicants and individuals at risk of applying for DI. The authors found that individuals' employment and earnings declined before they applied for DI, with the biggest changes occurring in the 6 months before application. Four at-risk groups had relatively high rates of DI application (13 percent to 22 percent) although they represented small proportions of all DI applicants (2 percent to 5 percent): new private disability insurance beneficiaries, new workers' compensation (WC) recipients, new unemployment insurance beneficiaries with disabilities, and workers with disabilities at risk of unemployment benefit receipt.

Contreary and others (2017) determined that certain DI applicants are more likely than others to receive a DI allowance based on their preapplication work history. In their analysis of medical disability determinations, the authors found that DI allowance rates were higher for applicants who worked consistently either until application or until shortly before application. These applicants had relatively well-paying jobs, often with benefits such as private health insurance. Applicants who either had been out of the workforce for a relatively long period before application or had an intermittent work history were less likely to receive a DI allowance, and they tended to rely more on means-tested and social insurance programs—such as unemployment insurance benefits and WC—for support.

The two studies point to aspects of early intervention that are both promising and challenging. The four at-risk groups identified in Thompkins and others (2014) could benefit from early intervention to help them maintain employment, but each represents only a small portion of all DI applicants, thus limiting the scope of such an intervention. The findings of Contreary and others (2017) imply that different policy approaches might work for different target populations. Interventions that target workers still connected to an employer could serve some of the roughly half of DI applicants with recent attachments to the labor force and help that population remain employed. Applicants with less stable work histories and no recent connection to an employer might require a different, more comprehensive array of supports to regain employment.

Schimmel Hyde, Wu, and Gill (2018) focused on a specific group of applicants: older workers (aged 51 to 66) who were initially denied DI benefits based on residual ability to work in a current, former, or other occupation. The authors found that few older DI applicants who were denied benefits for this reason worked at a substantial level following denial. More commonly, they continued to pursue benefit award, often successfully; nearly two-thirds were ultimately allowed DI benefits after appealing the initial decision or reapplying, and many of the rest claimed OASI benefits before full retirement age. The findings suggest that returning to work after exiting the labor force and seeking benefits is difficult. This may be because of health limitations, but may also reflect difficulties in finding work at older ages. To forestall labor market exits and Social Security benefit claims among older workers who experience new disabilities, employers could provide workplace accommodations or retraining to allow them to remain working.

Intermediary organizations that provide nonattorney representation services to people applying for DI or SSI are a prominent but understudied part of the disability-service landscape, primarily because client-level data are not available. Luca and Ben-Shalom (forthcoming) used operational data from one such organization to describe how it screens potential clients and supports current clients throughout the DI application process. The authors found that the intermediary's screening process turns away a substantial proportion of prospective clients whom they assess to have a low likelihood of allowance. This results in a DI allowance rate for the intermediary's clients that is substantially higher than that of all DI applicants. However, the authors were not able to track outcomes for the many individuals who were screened out by the intermediary or who were screened in but ultimately did not use its services. It is also impossible to know what the outcomes for those who were screened in would have been if they applied for DI on their own. Information about these three groups could provide a more complete picture of the intermediary's role in DI applications, including the potential incorporation of similar screening tools in the DI and SSI disability determination process.

Workers at Risk of DI Application and Their Employers

A critical component of effective early intervention is the timely identification of the particular workers who are most likely to benefit from intervention services and supports. This means identifying workers soon after they have developed medical conditions that interfere with their ability to work and distinguishing between those who would and would not benefit from expanded services (Stapleton and others 2015). Early identification and engagement of workers at risk of job loss and DI entry is relatively straightforward—though not easy—in WC systems and private and public short-term disability insurance programs, in which workers are required to file claims to obtain benefits (Ben-Shalom 2016). Two CSDP studies examined options for using information in WC and short-term public and private disability insurance claims to identify workers who develop medical conditions that put them at risk of a prolonged work disability and DI entry.

Neuhauser, Ben-Shalom, and Stapleton (2018) analyzed California's statewide data on State Disability Insurance (SDI) and WC claimants. The authors found that 13 percent of SDI claims and 19 percent of WC claims that lasted for at least 8 days ultimately lasted for 12 months or longer. The long-term SDI and WC claimants were similar in demographic and diagnostic characteristics to DI awardees nationwide, except that they were somewhat younger. For early-intervention purposes, the SDI and WC data suggest that information available in initial short-term disability claims—such as the claimant's age, sex, medical diagnosis, and wage—can help target workers whose claims are likely to last for 12 months or longer and who are therefore at risk of DI entry. However, it is also important to collect more information at the time of filing to distinguish between those whose odds of return to work would benefit from earlier assistance and those whose would not.

Along similar lines, Contreary, Ben-Shalom, and Gifford (2018) used data on private short-term disability insurance claims from a large database of employer-based benefit programs to document that age, primary diagnosis, and industry are predictive of exhaustion of short-term disability benefits and transition to long-term disability insurance and risk of DI application. The authors also found that rapid attrition of short-duration claims from the sample indicates that waiting even a few weeks can substantially increase the efficiency of targeting efforts. The authors acknowledged the importance of considering the potential trade-offs involved in delaying the start of any intervention, as well as options for initial screening that could improve earlier targeting.

A less direct approach to early intervention involves encouraging employers to help workers remain employed after illness or injury. This can be accomplished by holding firms partially responsible for the DI benefits paid to their recent employees, whether through a mandatory short-term disability program (Autor and Duggan 2010) or by applying an experience rating to the DI portion of the Federal Insurance Contributions Act premium (Burkhauser and Daly 2011). Another approach would require employer-provided group long-term disability insurance policies to pay 100 percent of benefits over an initial period (Stapleton and others 2017). In analyzing the implications of such proposals, Stapleton and others found that mandatory short-term disability benefits and DI experience-rating would place a relatively large cost burden on firms employing low-wage workers, particularly for those with fewer than 500 workers. The authors concluded that firms with high potential liabilities might seek to accommodate and retain workers with challenging medical conditions, but they might also be less likely to hire or retain workers at high risk of medical problems, which would disproportionately include low-wage workers. Because employers providing long-term disability insurance group coverage tend to be large and pay relatively high wages, requiring that the policies pay 100 percent of benefits for an initial period would be less problematic for low-wage workers, but it would increase long-term disability insurance premiums and, in the absence of countervailing policy measures, would likely reduce coverage.

Factors Affecting DI and SSI Eligibility and Participation

Numerous medical and nonmedical factors affect eligibility for DI and SSI benefits—not only in an initial award decision but also in deciding whether to continue benefits once awarded. Understanding these factors and their effect on program participation is important for assessing current policies, considering new ones, and forecasting future participation and costs. Four CSDP studies under the DRC focused on the eligibility determination process and trends in the medical characteristics of applicants. Three others considered the role that regional health-condition and disability-status characteristics play in determining geographic variation in program participation, and another study took a closer look at paths to termination of DI or SSI benefits after program entry.

Obesity, Other Physiological Measures, and Vocational Factors

Obesity prevalence in the United States has increased rapidly in recent decades. Because obesity raises the risk of many significant medical conditions, the increased prevalence has important implications for disability prevalence and disability program participation. Schimmel Hyde and others (2016) examined trends in obesity among disability program applicants from 2007 through 2013 using SSA data collected electronically at the time of application. They found that applicants were much more likely than the full working-age population to be obese (40 percent versus 29 percent in 2013). They likewise found that obesity prevalence rose faster among disability program applicants (an increase of 2.8 percentage points, or 7.5 percent) than among the overall working-age population (1.8 percentage points, or 6.7 percent) over the study period. Although the findings do not establish a causal relationship between obesity prevalence and application filings, they suggest that such a connection is important.

SSA removed obesity from its medical Listing of Impairments for adults in 1999, which meant that the disability determination process could no longer find new DI and SSI applicants to be medically eligible based on obesity alone (although it could be considered a contributing factor). Stahl, Schimmel Hyde, and Singh (2016) found that the change led to a decline of about 60 percent in the number of applications in which SSA adjudicators cited obesity in the primary impairment field between 1999 and 2000. After 1999, the share of applications with obesity in the secondary impairment field steadily increased, so that by 2012, the share of applications citing obesity reached pre-1999 levels. Further, initial allowances declined for applicants with obesity cited as a primary or secondary impairment, as one might expect after delisting. The share of applicants who received an initial allowance based solely on meeting the criteria contained in SSA's medical listings fell precipitously after the policy change, from about 62 percent in 1990–1999 to about 10 percent in 2000–2012. Conversely, the share of applications citing obesity that were allowed at the initial level based on medical and vocational factors (that is, accounting for age, education, and work experience as well as the medical factors) increased after 1999, placing additional administrative burdens on the disability determination process and presumably delaying allowances for many applicants.

Obesity is just one example of numerous physiological markers that might predict disability. Blue and others (2017) assessed the extent to which 19 distinct and objective physiological measures can predict DI and SSI benefit receipt among adults aged 51 to 65. The authors found that, after controlling for age, sex, and other predictors of disability program participation, the physiological measures added substantially to their ability to predict SSI and DI participation. The findings indicate that SSA could use physiological measures in forecasting program enrollment and conducting disability determinations, but the authors note that much more work is required before implementing policies suggested by this line of research.

Another DRC study examined the evidence base for SSA's medical-vocational guidelines. Adjudicators use these guidelines to assess whether a DI or SSI applicant with a significant medical condition is able to return to a past occupation or, if not, may adapt to new work. Mann, Stapleton, and de Richemond (2014) conducted an exhaustive literature review to identify evidence that supports the way that SSA incorporates the vocational factors—age, education, and work experience—into its determinations of whether an applicant can perform work that he or she has not performed in the past. The authors concluded that no rigorous evidence directly supports the way the determination process currently uses vocational factors, nor is evidence available that would enable SSA to make better use of vocational factors in determining the applicant's ability to work at a substantial level. Filling this evidence gap could improve the disability determination process. For example, it would be useful to identify causal links between particular vocational factors and return to work following either a permanent layoff or the onset of a significant medical condition. Information gathered on injured worker experiences for the Retaining Employment and Talent After Injury/Illness (RETAIN) demonstration might suggest avenues for further research in this area.

Determinants of Geographic Variation in Program Participation

Geographically, DI and SSI participation among the working-age population and SSI participation among children varies widely. Three CSDP studies examined the nature and causes of that variability.

Gettens, Lei, and Henry (2018) sought to explain geographic variations in DI and SSI participation among working-age adults. The authors found that county-level variation in disability prevalence (based on self-reported activity limitations recorded in the American Community Survey) and socioeconomic characteristics explained most of the variation and that inconsistent disability program administration was not a significant contributor to geographic variation in program participation.

Two studies considered geographic variation in child SSI caseloads and outcomes. Schmidt and Sevak (2017) examined county-level variation in child SSI caseload growth. They found that, nationally, county-level variation in observed factors such as poverty and unemployment rates, disability and health conditions, and the share of children in special education explains only 30 percent to 40 percent of the growth in child SSI caseloads from 2003 to 2008, and about 25 percent of the growth observed from 2008 to 2011. Although the importance of these factors in explaining caseload growth varied substantially by region and across states, observable factors could not explain most of the variation. Hemmeter, Mann, and Wittenburg (2017) examined the age-18 redetermination outcomes of former child SSI recipients and found substantial variation; cessation rates by state ranged from 20 percent to 47 percent. Southern states had higher cessation rates relative to other regions even after accounting for differences in the characteristics of the state caseloads. Both studies point to the importance of a variety of geographic factors—both measurable and unobserved—that affect SSA program caseloads, including state and local service environments and the health, culture, and expectations of regional populations.

Beneficiary Experience After Program Entry

Better knowledge of the long-term experiences and characteristics of DI and SSI awardees could help policymakers develop programs tailored to the needs and circumstances of various subgroups. Anand and Ben-Shalom (2018) examined the various work- and program-related milestones achieved and the pathways followed by new DI and SSI awardees. Examples of work-related milestones include completion of a trial work period (TWP) and suspension of benefits because of work earnings. Examples of program-related milestones include attainment of the OASI full retirement age and, for those who were initially awarded either DI or SSI benefits, a subsequent award for the other program as well. The authors found that most DI and SSI awardees achieved few (if any) work- or program-related milestones in the 10 years after their initial award. Furthermore, many of the awardees who achieved initial work- or program-related milestones did not make additional progress toward exiting the program. For example, 4.9 percent of DI-first awardees completed a TWP, but only 2.9 percent subsequently had their DI benefits suspended because of earnings. Similarly, 6.2 percent of SSI-first awardees achieved positive countable earnings, but only 1.9 percent subsequently had their SSI payments suspended because of earnings. Interventions that identify awardees who achieve initial work milestones and support their continued efforts toward self-sufficiency might improve return-to-work outcomes and reduce reliance on benefits.

Beneficiary Characteristics, Financial Well-Being, and Work

DI and SSI are important parts of the social safety net for youths and working-age adults with disabilities, providing many with income support and access to health insurance. Understanding the characteristics and circumstances of SSI and DI program participants, the adequacy of the benefits they receive, and their work activity is fundamental to determining how well the SSI and DI programs fulfill their missions—and to considering how they might be improved. Beneficiary employment is particularly salient because earnings can supplement SSI and DI income and some beneficiaries may become self-sufficient and exit the disability programs. SSA devotes substantial effort to encouraging and supporting beneficiary employment through a variety of SSI and DI work-incentive provisions and through programs such as Ticket to Work and Work Incentives Planning and Assistance.

CSDP researchers and their collaborators have conducted numerous DRC-sponsored studies of beneficiary characteristics, well-being, and employment. We summarize selected findings from these studies below.

Beneficiary Characteristics

To make information about beneficiaries more readily available, CSDP developed a series of papers and data briefs based on the rich information collected through the National Beneficiary Survey (NBS), a nationally representative survey of working-age disability beneficiaries sponsored by SSA. By pooling NBS data from multiple years, researchers were able to study relatively small and nationally representative subgroups of SSI and DI participants—groups about which little information was previously available. For example, DRC studies profiled working-age SSI and DI beneficiaries with intellectual disabilities (Livermore, Bardos, and Katz 2017), those who were parents of minor children (Livermore and Bardos 2016), and those who were high earners (Livermore and Bardos 2015a). Others, described in later sections of this article, profiled young adults (Bardos and Livermore 2016) and beneficiaries with psychiatric disabilities (Livermore and Bardos 2017). The findings of these profiles demonstrate the many ways in which particular beneficiary subgroups differ from the “average” beneficiary and the importance of considering the heterogeneity of the DI and SSI beneficiary populations in the face of programmatic changes and reforms.

The NBS also supports analyses of how beneficiary characteristics and experiences have changed over time. Livermore, Sevak, and Shenk (2019) found a large increase in the share of DI-only beneficiaries who had work goals or work expectations, from 34 percent in 2005 to 43 percent in 2015. The estimated difference between the 2005 and 2015 outcomes remained large and statistically significant after accounting for changes in beneficiary characteristics over the period. The study also found that beneficiaries reporting work goals or expectations in 2015 were significantly older than the work-oriented group in 2005. The findings suggest that demand for return-to-work support among beneficiaries may have increased over time, and that such supports should address the needs of an older target group to be effective.

Financial Well-Being of Beneficiaries

Although many beneficiaries rely on the cash benefits DI and SSI provide, the benefit amounts are modest. The average monthly DI disabled-worker benefit marginally exceeds the poverty threshold for a single individual, and the average SSI payment is well below that threshold. CSDP studies have confirmed high poverty rates among beneficiaries, with 12 percent to 20 percent of working-age DI-only beneficiaries in poverty and 31 percent to 48 percent of working-age SSI recipients in poverty based on data from national surveys (Schimmel Hyde and others 2018). Moreover, the higher risk of poverty among DI beneficiaries persists after DI benefits convert to OASI benefits (Wu and Schimmel Hyde 2019). Four years after full retirement age, the number of former DI beneficiaries who began to receive DI benefits before age 62 and were in poverty was triple the share of those who began to receive OASI benefits after the full retirement age. Of course, beneficiaries may have had lower incomes and higher poverty rates before they began to receive benefits; neither study accounted for income or earnings before program participation.

Poverty is not distributed evenly across all beneficiaries, and certain characteristics are associated with increased financial vulnerability. Compared with other DI-only beneficiaries, those in poverty were significantly less likely to have completed high school or to be married, and were more likely to have children younger than 18 (Livermore and Bardos 2014). They also reported more activity limitations and poorer general health than those with higher incomes. Working beneficiaries were significantly less likely than nonworking beneficiaries to be in poverty, holding other observable characteristics constant. However, beneficiaries in poverty were as likely to be working as were those in higher-income households. Among working beneficiaries, earnings and hours worked did not differ significantly based on poverty status. However, poor beneficiaries were significantly more likely to be working in sheltered or supported work settings, which typically offer low wages.

The high rates of beneficiary poverty raise the question of whether and how beneficiaries are able to meet their financial obligations—especially given that their medical conditions or impairments may expose them to significant costs. To find out, Gettens and Henry (2019) interviewed a small sample of disability program beneficiaries in a single metropolitan area. The authors found that beneficiaries generally constrain their spending to align with their modest income. Many beneficiaries reported having very low housing expenses, restraining their food consumption, curtailing discretionary spending, or occasionally going without one or more basic needs. Comparing data from multiple surveys for 2009, Schimmel Hyde and others (2018) found that the shares of beneficiaries who also received housing, energy, or food assistance from public programs varied widely among the survey sources, but were not trivial in any instances. Those benefits may effectively increase the resources available to the family but are not included in standard measures of income. A minority of beneficiaries in 2009 supplemented their benefits with earnings. Depending on the survey source, earnings amounts varied, but generally represented less than 10 percent of individual income. As a percentage of beneficiaries' family income, however, earnings ranged from about 25 percent to 30 percent, reflecting the inclusion of spousal earnings for married beneficiaries. Yet, that share is substantially lower than that of nonbeneficiaries, for whom earnings represented about 85 percent of family income.

The Great Recession and Beneficiaries' Well-Being and Work

Early evidence suggested that the 2007–2009 recession affected workers with disabilities more than other workers (Kaye 2010), and that many left the labor force to seek disability benefits (Maestas, Mullen, and Strand 2015). Evidence gathered more recently suggests that the employment rate of workers with disabilities took many years to return to prerecession levels (Kessler Foundation 2018). However, the recession's effects on the employment and income of SSI and DI beneficiaries were not immediately known. Two CSDP studies examined how beneficiary employment, reliance on SSA program benefits, and poverty changed before, during, and after the Great Recession.

Levere and others (2018) found that beneficiaries were less likely to have their cash benefits suspended or terminated for work (STW) during and after the recession. The number of DI and SSI beneficiaries with at least 1 month of STW status in 2011 was about 25 percent lower than it had been in 2008. However, those achieving the STW milestone in postrecession years retained that status for as many months as their counterparts before and during the recession. The effects of the recession on achieving STW status were more pronounced for SSI recipients than for DI beneficiaries. Following cohorts of beneficiaries based on their initial eligibility year, the study found steeper declines in the likelihood of STW status for SSI recipients (including those concurrently receiving DI) during the recession than for DI-only beneficiaries. This likely reflects a combination of differences between beneficiaries of the two programs: first, in the strength of previous connections to the labor force; and second, in how earnings affect benefits.

Livermore and Bardos (2015b) found that, in absolute terms, the labor force participation rates of disability program beneficiaries and nonbeneficiaries had similar declines from 2006 through 2010, but the proportional change was significantly larger for beneficiaries. Declines in income resulting from the recession translated into absolute increases in poverty rates that were similar for beneficiaries and nonbeneficiaries. Proportionally, the increase in the poverty rate was larger for nonbeneficiaries than for beneficiaries because a smaller share of the former was in poverty before the recession. The findings suggest that steady income from DI, SSI, and other safety-net programs contributed to the relative stability of beneficiary income, weakening the negative effects of the recession on their financial status.

Other Factors Affecting Beneficiary Work Activity

The desire of many beneficiaries to work, coupled with the possibilities of reduced dependence on benefits, increased beneficiary income, and reduced federal outlays, has focused policymakers' attention on options for increasing beneficiary employment. In recent years, SSA has modified administrative review of beneficiary work efforts to adjust benefits more accurately and timely and to reimburse vocational rehabilitation (VR) agencies for the role they play in helping beneficiaries work.

Schimmel Hyde and O'Leary (2018) examined SSA reimbursements to VR agencies that served beneficiaries who wished to work. For beneficiaries who received VR services and subsequently maintained substantial employment, the estimated value of benefits forgone for work was about 10 times greater than that of SSA reimbursements to VR agencies. Notably, only 4 percent of beneficiaries served by VR agencies generated an SSA reimbursement. The authors concluded that because benefits forgone for work are so much greater than payments, VR agencies may have been eligible for SSA reimbursement for other beneficiaries, had they sought payment and completed the requisite paperwork. The funds provided by unclaimed SSA reimbursements could free up scarce VR agency resources to serve more of the approximately 40 percent of beneficiaries who applied for VR but were not served when they initially applied.

Delays in reporting and processing beneficiary earnings information can lead to benefit payments that should not be made, sometimes for several months or years. Avoidance of work-related overpayments is often cited as a reason why some beneficiaries do not work, or choose to keep earnings below the threshold that results in benefit suspension. Hoffman and others (2019) documented that 71 percent of DI beneficiaries with earnings sufficient to put them at risk of a work-related overpayment during 2010–2012 were overpaid in at least 1 month. The median overpayment amount was $9,282, approximately equivalent to 9 months of the average monthly DI benefit amount at the time. Reflecting the fact that a very small share of beneficiaries had sufficient earnings to be at risk of overpayment during the period, those receiving work-related overpayments represented 1.9 percent of all DI beneficiaries.

Kregel (2018) interviewed 84 DI beneficiaries who had received a notice of overpayment to learn about their experiences. Of those interviewed, 51 percent immediately stopped working upon receiving the overpayment notice. Those who quit working said they did so because they feared they would receive another overpayment, or were frustrated by the overpayment. Although some of these beneficiaries may have used overpayment fears and frustrations as justification for not working, Hoffman and others (forthcoming) found evidence suggesting that overpayment notices may trigger a reduction in the share of beneficiaries with substantial earnings.

Special Populations

Several DRC studies conducted by CSDP have focused on youths with disabilities or people with mental health conditions. Both groups are among the disability program beneficiaries who receive SSI or DI benefits for the longest periods (Riley and Rupp 2015). Hence, they both represent important target populations for services designed to avoid a lifetime of poverty and, ultimately, to reduce long-term federal expenditures for their support.

Youths with Disabilities

Youths with significant disabilities encounter various barriers in their transition to adulthood, including health-related challenges, a lack of appropriate work supports, less developed social networks, and limited awareness of their rights to disability-related workplace accommodations (Rangarajan and others 2009; Shandra and Hogan 2008). Youths and young adults receiving SSI face additional barriers and incentives because of their households' limited resources and poverty. Such factors contribute to youths with disabilities experiencing poorer educational and employment outcomes as adults than their nondisabled peers (Mann and Honeycutt 2014; Newman and others 2011).

Bardos and Livermore (2016) profiled young adults (aged 18 to 29) participating in the SSI and DI programs. The authors found that a large majority of young adult beneficiaries were interested in employment, and their employment rates were more than twice those of older beneficiaries. Yet most young adults who worked did so part-time in low-paying jobs. Further, relative to older beneficiaries, young adult beneficiaries received less public and other income support and experienced higher rates of poverty. The study also found that, despite their strong interest in work, young beneficiaries faced numerous significant barriers to employment and economic independence. A large share had no experience in the labor market, and many had less than a high school education or its equivalent. Additional training, and the availability of other supports and opportunities that address these barriers, might improve outcomes for this group and help some to achieve greater economic well-being. The Promoting Readiness of Minors on SSI (PROMISE) demonstration, now in the evaluation phase, tests whether providing such supports will help SSI youths achieve greater self-sufficiency as young adults.

State VR agencies are an important resource for youths with disabilities who want to work and become independent as they transition to adulthood. Before the passage of the 2014 Workforce Innovation and Opportunity Act (WIOA), VR agencies served few applicants younger than 16. Honeycutt and others (2015) found that 8 percent of youths aged 16 to 24 with disabilities applied for state VR services from 2004 through 2006. Of those who applied, 56 percent eventually received services, and of those who received services, 56 percent were employed for at least 3 months at case closure. The findings suggest that more youths with disabilities might benefit from using state VR services. WIOA provisions require VR agencies to use 15 percent of their federal funding on preemployment transition services and to provide those supports to students before they apply for VR services. Those provisions, and others, have likely increased the number and percentage of youths with disabilities receiving VR services relative to the pre-WIOA period analyzed in this study. In another study of the 16–24 age group, Honeycutt and others (2017) examined a specific type of service provided by state VR agencies that is believed to be particularly effective in improving adult outcomes: postsecondary education supports. The authors found a positive association between the delivery of those services and employment at case closure. Although the findings of both studies suggest positive effects of state VR services on employment, the supporting statistics are descriptive and therefore do not provide evidence of causation.

Honeycutt and others (2017) also found that many youths who participated in the federal disability programs and received state VR services had positive employment outcomes and subsequently reduced their dependence on SSA program benefits. Nearly 30,000 of those youths and young adults applied for state VR services in each year from 2004 to 2006, accounting for 4 percent of SSA program participants aged 16 to 24. Those VR customers were less likely to be employed at case closure than were their same-age nonbeneficiary counterparts. Nonetheless, 48 months after application for VR services, 14 percent of youths receiving SSI or DI benefits had at least 1 month of benefit suspension because of earnings. These statistics point to both the connections between young federal disability program participants and state VR agencies and the program savings that can result from those connections.

Hoffman, Hemmeter, and Stegman Bailey (2018) examined the long-term outcomes of child SSI recipients whose eligibility was terminated in their age-18 redeterminations and how those outcomes were associated with their receipt of state VR services as youths. The authors tracked the outcomes for these former child SSI recipients at ages 27 to 30, and differentiated them by whether they reported using VR services or vocational training before age 18. The study found that more than half (58 percent) had no earnings during the year studied, and few of those with earnings had accrued amounts that met the level defined by SSA as substantial gainful activity. Holding observable characteristics constant, those who had received VR services or other vocational training before age 18 were more likely to earn above the substantial gainful activity level and less likely to be receiving federal disability program benefits. Although the findings do not provide causal evidence of the effects of the VR services, they suggest that services provided by state VR agencies could help SSI youths to avoid returning to the federal disability programs as adults.

People with Mental Health Conditions

About one-quarter of working-age DI and SSI beneficiaries have a mental health condition recorded as their primary impairment (SSA 2017a, 2017b). Mental health conditions can affect educational attainment, obtaining a job, productivity, and labor supply. CSDP studies have examined the characteristics of beneficiaries with mental health conditions, their barriers to work, and the long-term effects of interventions designed to improve their employment outcomes and reduce their reliance on public programs.

Livermore and Bardos (2017) estimated that 46 percent of all SSI and DI adult beneficiaries had at least one mental health condition. Although the SSA estimate of about one-quarter considered only the primary impairment that qualified an applicant for benefits, Livermore and Bardos also considered secondary impairments that SSA noted as contributing to medical eligibility and survey respondents' self-reported reasons for their activity limitations. The authors also found that about one-third of the beneficiaries that SSA found eligible for disability benefits on the basis of a mental impairment did not attribute their activity limitations to a mental health condition in their survey responses. Relative to other beneficiaries, those with mental health conditions were more likely to report a desire to work but were no more likely to be employed, perhaps because they were also more likely than other beneficiaries to report a variety of employment barriers. Of those barriers, the most frequently cited—after poor health—was discouragement over previous work attempts. This finding suggests that employment interventions aimed at preventing failed work attempts, if implemented timely and effectively, might improve employment outcomes.

Cook, Burke-Miller, and Bohman (2017) studied one such intervention: the Texas site of the Demonstration to Maintain Independence and Employment (DMIE), funded by the Centers for Medicare and Medicaid Services. The Texas DMIE provided case management, enhanced access to health care, and referrals for employment supports to employed adults with mental health conditions, with the goal of postponing or preventing application for federal disability benefits. In the original DMIE evaluation, Gimm, Hoffman, and Ireys (2014) found that DI and SSI receipt were reduced by 27 percent for intervention participants 1 year after enrollment, relative to the control group's rate. Cook, Burke-Miller, and Bohman followed the Texas DMIE participants for 5 years to assess the longer-term effects of the early intervention on employment. For participants overall, the study found no effects on employment or reliance on Medicaid as a source of health insurance. Yet among the participants identified as having a serious mental illness, the regression-adjusted estimates suggest that treatment-group participants were almost five times more likely than control-group participants to be employed during the follow-up period. The study provides some evidence of the long-term effect of earlier intervention for workers with mental health conditions, particularly when targeted to those with the more severe conditions.

Two other CSDP studies examined the long-term outcomes of SSI and DI beneficiaries with mental health conditions who received supported employment interventions. Cook, Burke-Miller, and Roessel (2016) conducted a 13-year follow-up study of beneficiary participants in the Employment Intervention Demonstration Project (EIDP), sponsored by the Substance Abuse and Mental Health Services Administration. The EIDP was an experimental study of supported employment programs targeting people with mental health conditions. Participants assigned to the treatment group received supported employment services that included personalized job-search assistance and ongoing vocational supports provided by multidisciplinary teams that coordinated employment and clinical services. The authors found that, relative to the control group, treatment-group members were almost three times more likely to be employed during the 13-year follow-up period. The treatment group also had higher average earnings, though the difference was small (about $24 per month), and were about 13 times more likely to attain STW status during the study period. The effects of the interventions on all three outcomes declined with time and eventually disappeared; most had ended by about 8 years into the follow-up period.

Baller and others (2017) studied the outcomes of participants in SSA's Mental Health Treatment Study (MHTS) in the 5 years after the demonstration ended. The MHTS targeted DI beneficiaries with schizophrenia and affective disorders. Treatment-group members were offered individual placement and support services, clinical case management, supplemental health insurance, and other medical supports (Frey and others 2011). Baller and others found that treatment-group members were more likely to be employed than were control-group members (45 percent versus 37 percent) and had higher earnings (roughly $14,000 versus $9,000), but were no less likely to have their DI benefits suspended or terminated because of earnings. The findings of both Cook, Burke-Miller, and Roessel (2016) and Baller and others (2017) suggest that supported employment might help beneficiaries work and improve their economic well-being, but they do not provide evidence that the interventions lead to long-term reductions in disability benefits under current program rules. This may in part be due to the limited (2-year) duration of the interventions and the chronic nature of mental health conditions; beneficiaries with these conditions might require ongoing medical and vocational supports to maintain long-term employment.

Health Insurance and Access to Health Care

Before the Patient Protection and Affordable Care Act (ACA) was enacted in 2010, many adults with disabilities who were not receiving either DI or SSI had limited options for obtaining health insurance. The availability of employer-sponsored plans was declining (Long and others 2016), particularly for those working in low-wage or part-time jobs (Claxton and others 2018), and nongroup coverage was often unavailable or prohibitively costly (Sommers 2006; Pizer, Frakt, and Iezzoni 2009). Many of the ACA provisions were significant for individuals with disabling conditions—in particular, the expansion of health insurance coverage, which is implemented at the state level. The ACA's Medicaid expansion option allowed coverage for adults with household incomes below 138 percent of the federal poverty level, thus decoupling Medicaid coverage from SSI receipt. Not all states exercised the option to expand Medicaid eligibility. Immediately after the policy change took effect in 2014, about half of the states opted to expand Medicaid. About a dozen others have expanded their programs since then (Kaiser Family Foundation 2019).

Because SSI and DI both confer public health insurance coverage, the programs provide more value to people with disabilities than cash payments alone. Kennedy and Blodgett (2012) speculated that the health insurance benefits that accompany federal disability benefits may motivate individuals to apply for SSI and DI even when they are indifferent to the income support the programs provide. DI beneficiaries are automatically enrolled in Medicare 24 months after entitlement for DI benefits. Almost all SSI recipients are eligible for Medicaid at the time of award, although in many states enrollment is not automatic, and a few states have Medicaid means tests that are more stringent than the SSI means test (Rupp and Riley 2016).

The ACA provisions were implemented in the early years of the DRC, providing opportunities for CSDP researchers to assess how the availability of health insurance coverage for people with disabilities affects disability program participation and service receipt. The next subsection summarizes some of those studies.

Health Insurance Coverage Expansions and the Disability Programs

The effect of the ACA provisions on DI and SSI application and award volume is particularly relevant to SSA. Application might have decreased if obtaining health insurance was an important motivation for enrolling in DI or SSI before the ACA, or it might have increased if the expanded coverage options enabled workers to free themselves of the “job lock” effect (maintaining employment to keep health insurance coverage). Application might also have increased if marketplace efforts to enroll adults in Medicaid or federally subsidized private plans increased awareness of potential SSI or DI receipt among those with disabilities—a phenomenon called the “welcome mat” effect.

In studying this issue, Anand and others (2019) found that ACA Medicaid expansions led to slower decreases in SSI application from 2014 through 2016—SSI application declined nationally during that period, but the rate of decline was slower in states that expanded Medicaid. The authors noted that this trend might be short-term—the study period immediately followed the Medicaid expansions—and does not necessarily mean that awards will follow a similar trend, especially if disproportionate shares of new applicants have less severe health conditions. The authors also found that the estimated effect of the ACA on SSI application varied substantially from state to state, highlighting the importance of local factors in influencing the outcomes of federal policies for people with disabilities and in considering the effects of national policies at the state level.

Levere and others (2019) also considered interactions between health insurance and SSI, focusing on historical instances of Medicaid and Children's Health Insurance Program (CHIP) expansions. They found that those health insurance expansions led to a decrease in both applications and awards for child SSI payments in the states where SSI awardees are not automatically eligible for Medicaid, but not in other states where SSI and Medicaid may require separate applications, even though Medicaid is often granted automatically to SSI recipients. Those findings suggest that making Medicaid available to low-income children with disabilities reduces SSI participation, but only if the transaction costs of Medicaid enrollment are substantial. In other words, Medicaid expansions in states that require separate applications for Medicaid and SSI do not affect SSI participation as much as they do in the states that formally link eligibilities. As with Anand and others (2019), this study found varied effects based on state-level policies, even under national policy reforms.

Levere, Hock, and Early (2019) analyzed the effect on SSI application of another type of ACA coverage expansion—namely, the provision that extended to age 26 the coverage of dependents on their parents' or guardians' health insurance. Before the ACA, dependents typically lost that coverage at age 19 or upon college graduation (Goldman 2013). Levere, Hock, and Early found a spike in SSI application around age 26 after ACA implementation. They estimated that the prospective loss of coverage resulted in about 3.8 percent more applications at age 26 in the 5 years following policy implementation than would have been filed at that age in the absence of the policy change. The spike might simply reflect delayed applications that would have occurred at earlier ages without the ACA provision. If so, public savings resulting from that provision are limited to reduced SSI and Medicaid expenditures for those years. The authors also speculate that the extension of private coverage to age 26 might have facilitated improved education and employment outcomes for some young adults with disabilities, ultimately enabling their self-sufficiency and thereby reducing public-health and income-support expenditures for many years.

Employment-Related Health Care Needs

Health insurance market reforms in the ACA (and in earlier state initiatives) generated interest in understanding whether newly available coverage would provide the necessary services and supports for individuals with disabilities who were interested in working. Historically, private insurance plans usually did not cover these types of supports—including personal assistance services, home- and community-based care, and durable medical equipment—but Medicaid did, to some degree. Broader availability of Medicaid coverage might have been particularly salient to workers with disabilities who wanted to remain attached to the labor force but required these types of supports to do so.

Two CSDP studies addressed this issue by drawing on the experience of Massachusetts, which implemented reforms in 2006 that later informed some of the ACA deliberations. One feature of the Massachusetts plan was a Medicaid buy-in, under which workers with disabilities could pay an income-based premium to maintain Medicaid coverage when their earnings would otherwise suspend their eligibility. Although the Massachusetts reforms made relatively low-cost coverage more widely available, people with disabilities who had health insurance reported that they continued to have unmet health care needs and high out-of-pocket costs (Gettens and Henry 2014). About 70 percent of workers and 67 percent of potential workers with disabilities used employment-related health care services, but 7 percent and 33 percent of those groups, respectively, had unmet needs for those services (Gettens, Henry, and Lei 2016).

Gettens, Hoffman, and Henry (2016) found that Medicaid—both as a sole source of coverage and as a “wraparound” to supplement other coverage—was particularly important in providing long-term supports to people with disabilities. Working-age people with disabilities in Massachusetts who were working, were enrolled in the Medicaid buy-in program, and also had Medicare or private coverage accrued an average of $427 per member per month in wraparound costs (in 2012 dollars). Community-based services and supports represented nearly two-thirds of that total, and personal-assistant services represented 60 percent of the community-based service costs. Thus, although the ACA's Medicaid coverage expansion may be especially helpful in covering these supports for workers with disabilities, not all states expanded Medicaid through the ACA, nor do all states have Medicaid buy-in programs.

Conclusion

In October 2018, SSA merged the DRC and the Retirement Research Consortium to form the Retirement and Disability Research Consortium (RDRC). Projects completed by CSDP researchers under the DRC and summarized above have already contributed to the knowledge base and helped lay the groundwork for potential RDRC research. Combining retirement and disability research under a single umbrella will be especially helpful for examining important topics that encompass the intersection of DI and OASI.

One of the goals of the DRC has been to make better use of rich data from administrative sources for which access is often difficult to obtain. Using these data helps to illustrate their value for researchers and paves the way for easier access and greater use in the future. Many CSDP studies used such data. Most were made possible by the collaboration of SSA researchers, who have comparatively easy access to the data, with CSDP researchers and their partners at other institutions. For example, several projects used SSA's Disability Analysis File, a restricted-access data file that combines data on 100 percent of working-age SSI and DI beneficiaries to enable researchers to conduct projects that would not have been feasible in the past. Others took advantage of administrative data linked to results of nationally representative surveys and other administrative sources. Several studies used administrative data to estimate long-term effects of earlier randomized controlled trials on employment and program outcomes. Two studies made innovative use of SSA's Structured Data Repository, which contains electronic information on every disability program application and disability determination from 2007 forward. Demand for access to these rich sources has only increased over time, and in response, SSA is vigorously addressing privacy and security concerns to make these and other data available to academic researchers and staff at other federal agencies.

Over its 6-year history, the DRC supported a broad range of rigorous disability research relevant to the federal disability programs, as intended. The studies have made important contributions to the body of research and have led to new investigations on existing and emerging issues of policy significance.