PERSPECTIVES: Long-Term Work Activity and Use of Employment Supports Among New Supplemental Security Income Recipients

Introduction

The Social Security Administration (SSA) oversees two programs that together provide income support to nearly 13 million working-age people with disabilities—Social Security Disability Insurance (DI) and Supplemental Security Income (SSI) (SSA 2014b). To qualify for either DI or SSI, an applicant must demonstrate inability to work at substantial levels because of a long-term, medically determinable impairment. The Ticket to Work and Work Incentives Improvement Act of 1999 (Ticket Act) sought to encourage and facilitate the work-seeking efforts of disability program participants and reduce their reliance on benefits. Ticket Act initiatives were implemented gradually and they changed the disability programs in several ways. First, they provided beneficiaries with information about how work affects their benefits. Second, they offered beneficiaries more options for obtaining SSA-financed employment services. Third, they allowed beneficiaries to return more easily to the disability rolls after unsuccessful work attempts. Fourth, they facilitated the processing of earnings information by SSA staff. Fifth, they established the Medicaid Buy-in program, which allows states to expand access to Medicaid for workers with disabilities who meet the SSI and DI medical eligibility criteria but do not receive cash benefits from either program because of their earnings.

Assessing the work efforts of SSI recipients and DI beneficiaries—and the effectiveness of work incentives such as the Ticket Act's signature initiative, the Ticket to Work (TTW)—poses a challenge because the most widely available and commonly used statistics on participant work efforts provide only cross-sectional monthly or annual perspectives. For instance, according to SSA (2008), 2.2 percent of working-age SSI recipients had their payments suspended for work under the Section 1619(b) work incentive (described later) in December 2007. Such cross-sectional statistics may be confused with cumulative statistics; for example, some observers might believe that only 2.2 percent of SSI entrants will ever forgo cash payments for work (at least temporarily), although the percentage of participants who actually do so is several times higher. Such misinterpretation could have a substantial bearing on policy or other decisions.

In this study, we use administrative data to examine, from a long-term cumulative perspective, the extent to which SSI recipients work and eventually stop receiving SSI payments because of work. We follow award-year cohorts of working-age SSI recipients (that is, new SSI recipients aged 18–64) for up to 11 years after they enter the rolls, and we produce cumulative statistics on their employment experience and use of work incentives. Because many recipients are on the rolls for many years, long-term cumulative data on their outcomes provide a more complete picture of their work-seeking efforts and how those efforts may be impacted by changes in policy and the economy. By following recipients for several years after award, we are able to (1) record the extensive longitudinal interaction between SSI and DI participation, (2) provide information on the length of time between SSI award and the achievement of important milestones, and (3) examine the extent to which awardees forgo cash payments because of work for at least part of the study period.

We follow annual SSI award cohorts from 1996 through 2006, and track all cohorts through 2007. Because the period we analyze spans years before and after passage of the Ticket Act, it reflects the experiences of recipients under both the pre-TTW rules and the initiative's original rules. More recent TTW regulations, implemented in July 2008, are not reflected in this analysis; however, our results will serve as a useful baseline for future analyses of their effects.¹

Much of the analysis focuses on the 2001 cohort because useful changes in the administrative data were first implemented in that year. The statistics for earlier cohorts, although more limited, still provide some insights about the effects of provisions of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA), which tightened SSI eligibility in several ways. For example, PRWORA instituted a more restrictive SSI medical eligibility standard for children, in part to offset the expansion of eligibility that resulted from the Supreme Court's 1990 Sullivan v. Zebley decision, and required all SSI child recipients to undergo eligibility redetermination under the adult standard upon reaching age 18 (Hemmeter and Gilby 2009). PRWORA also eliminated SSI and DI eligibility for persons whose drug abuse or alcoholism is material to eligibility and, with some exceptions, made citizenship a requirement for SSI eligibility.²

The methodology used in this article is similar to that used in Liu and Stapleton (2011), a study of DI awardees, except that we substitute SSI programmatic incentives for DI incentives. For example, instead of tracking the completion of a trial work period (applicable only to DI), we track qualification for Section 1619(a) and 1619(b) exemptions from restrictions on program eligibility because of work earnings (applicable only to SSI).

We find that 9.8 percent of the 2001 SSI awardees had given up their SSI payments for work in at least 1 month by December 2007. That figure is about 4.5 times the 2.2 percent cross-sectional 1619(b) qualification rate observed for December 2007. The fact that we follow recipients over a long period accounts for part of that difference, but two other factors also apply. First, because we track only SSI recipients with initial awards during the study period, we do not include those who have stayed in current-pay status for many years, as the cross-sectional statistics do. Second, we find that a substantial minority of SSI recipients did not qualify for 1619(b) status when they gave up SSI payments for work, and their SSI eligibility was terminated. If we exclude from our counts persons whose SSI payments were terminated, the percentage of 2001 awardees who were in nonpayment status because of work in at least 1 month by December 2007 is only 8.4 percent, compared with 9.8 percent if they are included. Thus, statistics on 1619(b) qualification understate the extent to which recipients ever forgo SSI payments for work for two reasons: they do not track recipients over a long period, and they do not count persons who gave up their SSI eligibility entirely.

We also find, however, that nearly half of SSI participants who gave up SSI payments for work still received DI benefits. That is, the percentage of SSI awardees who eventually gave up benefits from both programs because of work was only about half the percentage of those who gave up only SSI payments. For instance, although 9.8 percent of the 2001 cohort gave up their SSI payments for work by December 2007, that figure drops to 5.5 percent when we exclude those who received DI benefits in months with no SSI payments.

Other statistics provide information on employment milestones achieved prior to suspension or termination of payments. For example, by December 2007, 19.4 percent of the 2001 SSI award cohort achieved positive countable earnings (PCE) in at least 1 month; and 10.4 percent attained at least 1619(a) status, which allows recipients to retain SSI eligibility while earning at or above the substantial gainful activity (SGA) level. Close to 11 percent of recipients in the 2001 cohort had enrolled in employment services from providers that were eligible for payment from SSA. However, a large majority of recipients who had their SSI payments suspended or terminated for work did not enroll in such services.

Young recipients were much more likely than older awardees to forgo their SSI payments for work. In the 2001 award cohort, 19.2 percent of awardees aged 18–19 and 14.9 percent of those aged 20–39 at the time of award achieved that status, compared with only 6.3 percent, 3.7 percent, and 1.8 percent of those aged 40–49, 50–61, and 62–64, respectively. Finally, analysis of the 1996 SSI award cohort, which we follow for 11 years, indicates that a large majority of those awardees who found work and earned enough to give up all of their SSI payments did so within 5 years of award.

Background

In this section, we briefly describe the SSI program, the work incentives available to SSI recipients, and the DI program and its interactions with SSI. We also highlight the findings of earlier studies.

SSI Eligibility and Benefits

To qualify for adult SSI disability benefits, an applicant aged 18 or older must demonstrate that he or she is unable to engage in SGA because of a medically determinable impairment that is expected to last at least 12 months or result in death. SSI's medical eligibility criteria are identical to those for DI. In 2014, SSA considered SGA to be the equivalent of the work required to have unsubsidized earnings above $1,070 per month for nonblind applicants.³

SSI is a means-tested program, with federally set limits on income and assets. To receive any federal SSI payments, countable income and assets must be below those limits. In 2014, the countable monthly income limits were $721 for an individual and $1,082 for a couple. Countable asset limits, which have not changed since 1989, are $2,000 for an individual and $3,000 for a couple. The federal SSI payment is the difference between countable income and an annually adjusted amount called the federal payment rate. (Some states supplement the federal SSI payment for certain categories of recipients.) Countable income includes all monthly income above $20 from sources other than work, plus half of earnings in excess of $65 or any other earnings disregards, such as allowed impairment-related work expenses.⁴ Most SSI recipients also qualify for Medicaid coverage.⁵ This is true even if their labor earnings are high enough to reduce their SSI cash payment to zero but are still beneath a higher variable limit set under the Section 1619(b) work incentive (described below).

SSI Work Incentives

The incentives relevant to this analysis include the earned-income exclusion, Sections 1619(a) and 1619(b) of the 1987 Social Security Act, and TTW. The earned-income exclusion removes from countable income the first $65 of earned income, plus half of all additional earnings. Section 1619(a) allows SSI recipients to receive some cash payments even when their earned income is at or above the SGA level, and Section 1619(b) extends their SSI eligibility (including Medicaid coverage) indefinitely—even if their labor earnings are so high that their SSI cash payment is zero—as long as their earnings remain below a state-determined threshold.⁶ These provisions encourage work by reducing the risk of losing cash assistance and essential medical coverage when earnings increase.

SSI recipients may also receive employment services (for which SSA pays) if the recipient subsequently attains sufficient earnings levels over a specified period. TTW, a performance-based voucher program that was implemented during 2002–2004, is the most recent version of this work incentive. At the time of SSI award, the recipient receives a “ticket” that he or she may present to any employment network to obtain services. Employment networks include all state vocational rehabilitation (VR) agencies and other private and public entities that meet SSA criteria and agree to participate. Although the Rehabilitation Act of 1973 requires the state VR agencies to serve recipients, other employment networks may decline recipient requests for services. Payments to employment networks are based on clients' monthly earnings milestones and months with no SSI payments because of work. Before the rollout of TTW (including the 1996–2001 portion of our study period), SSA in essence paid only for services provided by state VR agencies under a system with a less stringent performance requirement.⁷ Since the TTW rollout, state VR agencies have retained the option to serve recipients under the earlier payment system on a case-by-case basis, rather than under either of TTW's new payment systems.

DI Eligibility, Benefits, and Beneficiary Interactions with SSI

In contrast with SSI, eligibility for DI is not means-tested; instead, eligibility for disabled-worker benefits requires the individual to have worked and contributed to the DI trust fund via payroll taxes for a sufficient period to attain “disability-insured” status.⁸ The benefit amount is based on past earnings—the higher the lifetime earnings of the beneficiary (or other relevant individual), the higher the benefit. DI benefits begin only after a 5-month waiting period that starts with the first month for which SSA determines the DI beneficiary was unable to engage in SGA (the “disability-onset” month). Twenty-four months after DI benefits begin, beneficiaries are automatically eligible for Medicare. DI beneficiaries with sufficiently low assets and income (including their DI benefits) also are eligible for SSI payments. SSI-only recipients may eventually become eligible for DI if they earn enough to become disability insured.

The interactions between SSI and DI eligibility are particularly relevant to the cumulative statistics presented in this study. The DI benefits of individuals initially determined to be eligible for both programs are included in SSI countable income. Because the 5-month DI waiting period does not apply to SSI, an individual who qualifies for both SSI and DI may initially receive only SSI payments, and then have his or her SSI payments reduced by the amount of the DI benefit (minus $20) when DI benefits start. Following Rupp, Davies, and Strand (2008), we distinguish between persons for whom the SSI payment is reduced to zero when they become eligible for DI (“serial beneficiaries”) and those who continue to receive cash benefits from both programs (“joint beneficiaries”).⁹ Our award cohorts exclude serial beneficiaries because we do not expect the existence of SSI to affect their behavior once they start to receive DI benefits. However, we include many others who receive DI benefits.

Previous Findings

The value of long-term cumulative statistics on SSI recipients has long been recognized. Past studies have addressed the length of time spent on the SSI disability rolls among children and working-age adults (Rupp and Scott 1995); the differences between long-term cumulative statistics and point-in-time statistics on SSI applications, caseloads, and awards (Pickett and Scott 1996); the effects of the age and diagnostic composition of cohorts of new DI and SSI awardees on their length of stay on the rolls (Rupp and Scott 1996); and the reinstatement rates for SSI recipients who had their cases closed and payments stopped (Kochhar and Scott 1998).

Few studies have focused specifically on the long-term cumulative work-related experiences of SSI recipients, however. The studies most similar to this one are two by Scott (1989, 1992). The 1989 study produced statistics on recipients who were newly awarded SSI disability payments in the last quarter of 1981, using a 1 percent sample file.¹⁰ Scott estimated that 7.5 percent of those awardees became ineligible for SSI payments because of excess income (other than Social Security benefits) within 4 years. However, excess income was not necessarily the recipient's earnings from work; for example, it might have represented a spouse's income.

Scott's 1992 study examined the postapplication work experience of all recipients on the SSI disability rolls in December 1988, again using a 1 percent sample file. He estimated that 22 percent of SSI disability recipients had had some postapplication work experience as of that month, including 4.3 percent with postapplication work experience of 5 years or more. Years on the rolls varied by recipient, however, and the percentage that achieved earnings high enough to lower cash payments to zero was not reported.

Two recent studies are particularly relevant to this one, even though they do not specifically focus on employment outcomes or the use of work incentives. Rupp and Riley (2011) analyzed longitudinal patterns of disability program participation and the interactions between SSI and DI program rules. The authors followed working-age disability awardees for 60 months after award and found substantial program interactions, with about a quarter of first-time awardees participating in both SSI and DI over the study period. Rupp and Riley (2012) examined how longitudinal patterns of SSI and DI program participation were associated with Medicare and Medicaid public health insurance coverage. Following new working-age disability awardees from 12 months prior to award through 72 months after award, the authors documented complex interactions between SSI and DI participation and the timing of Medicare and Medicaid coverage. Throughout the rest of this article, we discuss how interactions between the SSI and DI programs affect some of the central long-term statistics our analysis produced.

Data Sources

The SSI award cohorts used in this study, as well as most of the statistics presented for those cohorts, were developed from analytic administrative data files constructed for the TTW evaluation. The 2008 version of the files used here, collectively called the 2008 Disability Analysis File (DAF), contains extensive information on the more than 20 million individuals who received DI benefits, SSI payments, or both in at least 1 month from January 1996 through December 2008 (Hildebrand and others 2010).¹¹

To obtain information on enrollment for VR employment services, we also merge DAF records with state data on closed VR cases from Rehabilitation Services Administration Case Service Report (RSA-911) files for fiscal years 1998 through 2008 made available under an interagency agreement between SSA and the Department of Education. For the purpose of this analysis, we include only cases that were closed after eligibility for services was determined.¹²

Earnings are only recorded in SSI administrative data when the cohort member is actually on the SSI rolls, and some earnings might not be reported. Therefore, some of the statistics we report here also required access to SSA's Master Earnings File (MEF), which includes annual earnings data derived from tax reports under rules established by the Internal Revenue Service (IRS). SSA maintains an extract of earnings records for SSI and DI beneficiaries represented in the DAF. To comply with security requirements for the earnings data, qualified SSA staff produced the statistics based on those records and verified that they do not disclose personal information.

Although the 2008 DAF includes data through 2008, we have analyzed data only through 2007 because we expect SSA to revise many of the 2008 values for key variables at a later date. These revisions occur because of delays in the reporting of earnings and the processing time required for determining work-incentive status. In addition, although we report service enrollment statistics through 2007, the more recent years are subject to substantial revisions because of the nature of the RSA-911 data: Enrollment for a case is not captured in the file until the case is closed. For example, enrollment by an SSI recipient in 2007 is recognized only if the recipient's VR case closed before September 2007 or the recipient assigned his or her ticket to the state VR agency. Hence, we describe the service enrollment estimates for 2005 through 2007 as preliminary.¹³

The data have additional limitations that stem, like those described above, from having been collected for administrative rather than research purposes. The statistics we report are all based on data that have an important administrative purpose and are generally reliable but are subject to errors reflecting the processing of postentitlement work, as well as alterations because of changes in the postentitlement processes. If such errors occurred consistently over time, they would not affect trends in statistics across award cohorts. However, SSA's focus on reducing the backlog of postentitlement work, especially between 1999 and 2002, might mean that some observed trends reflect changes in administrative processing rather than changes in policy or the economic environment, and the size of any effect from such changes is potentially substantial.

Our findings also suggest another way in which a changing backlog might affect cross-cohort comparisons. For example, the downward trend in the percentages of former SSI child recipients who received their first SSI payment as adults at ages 18 and 19 instead of at age 20 or older (see Ben-Shalom and others 2012, Exhibit III.1) could be consistent with an increasing backlog in age-18 redeterminations. Such a backlog would mean that former SSI child recipients make up a decreasing share of our SSI award cohorts—a change not accounted for in our cross-cohort comparisons.

Finally, studies conducted by SSA in 1999, 2002, and 2004 identified about 466,000 cases of SSI-only recipients who had earnings histories and were potentially disability insured (SSA 2006). Many of those recipients, known as special disability workload cases, were awarded DI benefits retroactively. Our statistics do not capture retroactive benefits; instead, they only pick up the DI participation of those cases at the time of their first special disability workload payment.

Study Population

All of the statistics presented in this article represent 100 percent of the relevant SSI population (which includes individuals concurrently receiving both SSI and DI benefits); that is, they are population statistics rather than estimates. We develop annual cohort files from 1996 through 2006 based on the month in which the recipient was first paid an SSI payment as an adult according to information in the DAF. Although it is possible for an individual to have multiple periods of payment receipt, he or she is assigned to just one cohort based on the year that corresponds to the individual's first payment as an adult.

We count those who initially become eligible for SSI before age 18 as SSI adults starting in the month in which SSA determines that they are eligible as an adult. In the majority of such cases, adult eligibility results from an age-18 redetermination, but the former SSI child recipient is often older than 18 when the decision is made. In the case of denial at the age-18 redetermination or a lapse in SSI eligibility before an age-18 redetermination occurs, the former SSI child recipient is counted as an SSI adult from the month in which payments are awarded as the result of a new application as an adult.¹⁴

As mentioned earlier, we exclude serial beneficiaries (those who entered SSI during the DI waiting period, but became ineligible for SSI payments when DI benefits started) from our cohorts. Some serial beneficiaries later returned to the SSI rolls, however. We include those who did, and assign them to a cohort based on the first month in which an SSI payment was received after they had been in nonpayment status for at least 13 months. Apart from serial beneficiaries, many other SSI recipients who also received DI benefits are included in our cohorts. Some of them were awarded DI first and later became eligible for SSI, some were nonserial beneficiaries who were awarded SSI and DI at the same time, and others entered the DI rolls only after a longer period following SSI award.

Even if work-seeking behavior does not change across cohorts, employment outcomes are likely to change simply because of changes in distribution by age and sex, thus reflecting the differing employment outcomes across the age-and-sex groups. To control for those demographic differences, we adjust all of the cross-cohort statistics presented in this study for age and sex using the 2001 cohort (the last year before TTW) as the index cohort.¹⁵ We develop outcome measures for each age-sex group, and the cross-cohort statistics presented are weighted averages of these group-specific measures. We adjust state series using the same weights so that cross-state comparisons are not influenced by differences in age-sex composition. Table 1 presents the size and age-sex composition of the cohorts.

Outcome Measures

For each cohort, we develop two types of outcome measures: monthly statistics on employment and the use of work incentives, based on program data; and annual statistics on employment and earnings, based on IRS data (Table 2). One key statistic can be constructed only for 2001 and later award cohorts: SSI nonpayment status following suspension or termination for work (NSTW).¹⁶ We designate NSTW if the individual's SSI payments are known to have been suspended for work in a given month or in a previous month and the individual had not yet returned to SSI current-pay status, attained age 65, or died. We cannot confirm that the individual was engaged in SGA, or even working, in each such month. For all cohorts, we can uniformly identify the subset of NSTW months in which the beneficiary was in 1619(b) status. Differences between the 1619(b) and NSTW statistics thus reflect months in which the beneficiary was ineligible for SSI, for reasons that are generally unknown.¹⁷

We produce statistics on months in NSTW and 1619(b) status for all SSI recipients as well as for the subset of SSI recipients who did not also receive DI benefits. Differences between the DI and SSI work incentives enable some SSI recipients in NSTW or 1619(b) status to remain eligible for and receive a DI benefit. The most obvious example is when the beneficiary is in the trial work period. Another such situation is when countable income from all sources, including DI benefits and labor earnings, is sufficient to reduce the SSI payment to zero when labor earnings (after disregards) are lower than the SGA amount.¹⁸

Monthly earnings data would be ideal for our analysis because SSI payment amounts are based on monthly earnings levels. However, SSA has no administrative reason to collect earnings data for all working recipients in every month. Instead, our employment and earnings statistics are based on the most reliable earnings records available: annual IRS data from the MEF. These data have two substantial limitations in addition to their comparative infrequency. First, for new awardees, we cannot distinguish between earnings for preaward and postaward work; hence, we do not report employment and earnings statistics for the award year and the first postaward year.¹⁹ Second, not all earnings are reported to the IRS. An important exception for SSI recipients is sheltered workshop earnings, which are not subject to payroll taxes; thus, our annual employment and earnings statistics reflect competitive employment only, and miss any other earnings not reported to the IRS.

Many of the statistics we report are cumulative from the award year through the end of a given calendar year (usually 2007). For example, the reported percentage of recipients in the 2001 cohort with an NSTW month by the end of 2007 is an unduplicated count of individuals with at least 1 NSTW month as of the end of 2007. Cumulative statistics for the employment rate are an exception; the cumulative employment rate is for the period from the beginning of the second calendar year after award through the given calendar year.

Major Findings

We begin with findings for the 2001 cohort, the first cohort for which we have complete information on NSTW. Because of the importance of interactions with the DI program to the interpretation of the results, we first discuss the extent to which the 2001 SSI award cohort also received DI benefits, providing detail by age group. We then review the main work-effort milestones achieved by the 2001 cohort as of 2007, after which we compare key statistics across cohorts. That comparison is followed by findings from the MEF data, which contain information on earnings not recorded in SSI administrative data—either because the cohort member is off the SSI rolls or because of differences between MEF data and self-reported earnings.

2001 SSI Award Cohort

Like all SSI award cohorts, the 2001 cohort includes many awardees who also received DI benefits at some point. Such SSI awardees include those who (1) were awarded SSI and DI at the same time and had DI benefits low enough that their SSI payments were not terminated after the DI waiting period; (2) entered DI first but became eligible for SSI after spending down their resources or losing other sources of income; (3) entered SSI first but entered DI after accumulating the work experience necessary to meet the latter program's earnings-history criteria; and (4) were awarded SSI and, at some point, were also awarded either disabled adult child benefits or disabled widow(er) benefits.

More than 30 percent of the 2001 SSI award cohort received DI payments in at least 1 month during their SSI award year (Chart 1). The cumulative statistics show distinctive patterns by age at award. Although DI participation increased rapidly among individuals aged 18–19 at SSI award, it increased relatively little for those in the older age groups.

Chart 2 shows how many members of the 2001 SSI award cohort moved toward NSTW by December 2007. We find that 19.4 percent of the 2001 awardees had at least 1 month with PCE, 10.4 percent had at least 1 month of Section 1619(a) or 1619(b) status, and 8.4 percent had at least 1 month of 1619(b) status specifically. Notably, more recipients had at least 1 NSTW month by 2007 (9.8 percent) than had achieved 1619(b) status during that time.²⁰ Nearly half of those who reached 1619(b) status or NSTW were receiving DI benefits during those months (46.9 percent and 43.6 percent, respectively).

Chart 3 presents, for the 2001 SSI award cohort, the annual progression of cumulative percentages of recipients with PCE, 1619(b) status, and NSTW, as well as enrollment in employment services. It also shows the cumulative percentages that achieved 1619(b) status and NSTW without also receiving DI benefits. By the end of 2007, 5.5 percent of the 2001 awardees had been in NSTW and 4.5 percent had been in 1619(b) status while not receiving DI benefits in at least 1 month. Over the same period, 10.7 percent of recipients in the 2001 cohort had enrolled in services in at least 1 month. Similar analysis of the 1996 SSI award cohort over a longer period (11 years) indicates that a large majority of those awardees who found work and earned enough to give up all of their payments did so in the first 5 years after their award (Ben-Shalom and others 2012).

In Chart 4, both panels show the cumulative percentages of 2001 SSI awardees achieving NSTW by age group. The left panel tracks all recipients who attained NSTW in any month, while the right panel excludes recipients who collected DI benefits while in NSTW. Whether or not they received DI benefits, younger SSI awardees were much more likely to have at least 1 NSTW month than were older awardees.

As of December 2007, the total number of months in NSTW accumulated by the 2001 cohort was equivalent to 164 years per thousand recipients (left panel of Chart 5). That amounts to less than 2 months per beneficiary across a span of over 6 years, or 2.5 percent of all possible months. The two youngest age groups accounted for more than 70 percent of cumulative NSTW months, even though they represented less than 45 percent of the 2001 cohort. (The oldest group, ages 62–64, represented 4.5 percent of the 2001 award cohort but accounted for only 0.3 percent of the NSTW months—a proportion too small to be visible in the chart.) The cumulative number of months in NSTW that were not in DI current-pay status was equivalent to 75 years—less than 1 month per beneficiary over that period (right panel of Chart 5). Overall, 54.4 percent of the months in which recipients were in NSTW were months in which they received DI benefits.

Enrollment in Employment Services

Recall that Chart 3 includes cumulative statistics on employment service enrollment for the 2001 cohort. In an earlier study, we similarly analyzed the 1998 cohort—the first cohort with complete data on service enrollment—and found that 10.5 percent of that cohort had enrolled in services by 2007. Close to 60 percent of the 1998 cohort had achieved PCE in at least 1 month; 38.9 percent had attained Section 1619(a) or 1619(b) status, or both; and 31.7 percent had specifically achieved 1619(b) status. Most of the recipients who achieved PCE, 1619(a), or 1619(b) status did so after employment service enrollment, but many enrolled for services only after achieving one or more of those milestones. Moreover, almost 80 percent of the recipients who achieved NSTW by 2007 had not enrolled in employment services, or at least had not done so with providers that would be eligible for payment from SSA.²¹

Cross-Cohort Comparisons

We now assess how more recent cohorts have fared relative to earlier ones and whether changes in SSA policies, policies external to SSA, or the economic environment might have contributed to any observed cross-cohort differences. Charts 6–10 track and compare the experiences of seven cohorts at each of three intervals: the year of (or after) award, the third year after award, and the fifth year after award. Additionally, the charts track four later cohorts through the intervals that had occurred as of December 2007. (Charts showing employment-service enrollment and NSTW cover fewer cohorts because data on those outcomes are not available for every cohort.) Statistics for all cohorts are weighted to the 2001 cohort's age-sex composition.

In the absence of any change in policies from within or outside of SSA or in the economic environment, we would expect the charted columns and their component segments to be almost identical across cohorts. Besides changes in policy and the economy, changes across cohorts in the distribution of characteristics such as impairment type (but not age and sex, for which we already have adjusted the series) could cause cross-cohort variations, but such changes seem to occur gradually relative to changes in policy or the economic environment. Adjusting for age also accounts for some of the change seen in impairment types.

Because of the importance of program interaction to the interpretation of the results, Charts 6 and 7 show the extent to which each SSI award cohort also received DI benefits. For all cohorts, the share of SSI recipients who received a DI benefit in at least 1 month during their award year is at least 30 percent (Chart 6). The spike for the 1997 cohort might be due to the previously discussed tightening of SSI eligibility rules that followed the 1996 passage of the PRWORA. Those changes likely had a disproportionately negative effect on SSI-only awards, resulting in a higher percentage of joint SSI-DI awards. The decrease in DI participation during the SSI award year after 1997 and the increase after 2001 might be due to the improving economy in the late 1990s and the worsening economy after 2001, respectively, although other factors also might have played a role. Cumulative results for the third and fifth postaward years largely reflect the same trends.

The changes that the PRWORA brought about in 1996 appear to have permanently shifted the award-year and cumulative DI percentages upward for recipients aged 18–19 at SSI award (Chart 7). The award-year DI participation percentage for that age group continued to rise gradually through the 2001 cohort and remained fairly stable after that. Compared with that younger age group, recipients aged 20–39 at SSI award were substantially more likely to receive DI benefits in their award year and less likely to work their way onto the DI rolls as years passed. The shift between the 1996 and 1997 cohorts for the 20–39 age group is smaller than that for the 18–19 age group, possibly reflecting differential effects of PRWORA changes. Evidence of a business cycle effect appears for the older age group, as the unemployment rate (not shown) tracks a roughly parallel pattern ranging between about 4 and 6 percent; any such effect for the younger group may have been obscured by the size of the post-PRWORA shift. For the older group, the decrease in the award-year percentage of recipients on the DI rolls between the 1997 and 2000 cohorts is consistent with fewer existing DI beneficiaries receiving SSI awards during that period's economic expansion. Likewise, the increase in the percentage from the 2000 through 2004 cohorts might reflect increased DI entry during the recession among those who eventually qualified for SSI after spending down their resources.

Chart 8 compares the cumulative percentage of awardees achieving 1619(b) status across the 11 study cohorts. In the first postaward years, the percentages increase from the 1996 cohort through the 1999 cohort, during a period of economic expansion, and decline for subsequent cohorts following the recession of 2001. The series for the third and fifth postaward years indicate that cohorts with higher percentages in the first postaward year did not necessarily continue to outpace other cohorts in the later years. For example, the fifth postaward year percentage is lower for the 1999 cohort (as of 2004) than that of the 1996 cohort (as of 2001), presumably because the 5-year outcomes for the 1999 cohort were more affected by the 2001 recession than were those for the 1996 cohort. A cross-cohort comparison of the cumulative percentage achieving 1619(b) status and receiving no DI benefits reveals similar patterns, although at substantially lower levels.

There is a notable cross-cohort increase in the percentage of recipients achieving NSTW in at least 1 month within a given number of years (Chart 9). For recipients in their first postaward year, that increase between the 2001 and 2002 cohorts is surprising, given the economic recession during that period, and suggests that one or more other factors were involved. Possibly, special disability workload processing increased the number of working SSI recipients in NSTW because of the effect of the DI benefit award on the SSI payment. Cross-cohort differences in the proportion of awardees receiving DI benefits before they were first awarded SSI payments could also be a factor. Following the economic recession of 2001, the number of DI-only beneficiaries who spent their assets down and thus became SSI-eligible may have exceeded that of the prerecession cohorts. Indeed, Chart 6 appears to reflect such a trend. To the extent that these awardees were more likely to achieve NSTW than those who were awarded SSI before DI (or those who were awarded both simultaneously), any increase in the percentage of awardees in this group would increase the percentage of the cohort achieving NSTW, with other factors constant.

The cross-cohort increase in the percentage of recipients achieving NSTW while not in DI current-pay status was notably weaker than the respective increase among recipients overall. The difference implies that most of the observed cross-cohort increase in NSTW rates was among SSI recipients who received concurrent DI benefits and remained on the DI rolls when SSI payments ceased, or whose SSI payments ceased when they became disability insured and eligible for DI. Whatever the cause of the difference, the percentage of recipients that achieves NSTW without receipt of DI is a more accurate gauge of the extent to which SSI recipients forgo their payments because of work than is the percentage that achieves NSTW overall.

We also compare cumulative years in NSTW across award cohorts, beginning with the 2001 cohort (Chart 10). Each successive cohort had more NSTW years per 1,000 recipients, holding the years since award constant. This trend is largely consistent with the trend seen in Chart 9 for recipients with at least 1 month of NSTW as of various intervals. Cross-cohort growth in cumulative NSTW years while not in DI current-pay status was lower than cross-cohort growth in all cumulative NSTW years and, again, seems the more accurate gauge of the extent to which awardees completely forgo benefits because of work. For that group, successive cohorts exhibited substantial steady gains as of the third postaward year, from 24 years for the 2001 cohort (in 2004) to 30 for the 2004 cohort (in 2007). Possible explanations for those gains include the economic recovery and the TTW rollout. Changes in the composition of the cohorts (other than the age-sex distributions) also might contribute to the gains, but we have not examined that possibility further.

Employment and Earnings Statistics

Using data from earnings reported to the IRS and recorded in the MEF, Charts 11 and 12 present further evidence of employment success. Unlike the PCE statistics, MEF data contain information on earnings for awardees who have left the SSI rolls (for any reason) as well as for those who have not. In addition to filling a gap in the NSTW statistics that occurs because some SSI awardees work and eventually leave the SSI rolls entirely, these statistics reflect earnings reported to the IRS that are not contained in SSI records. For reasons described earlier, the employment and earnings statistics presented here start from the second postaward year for each cohort. Hence, the series for the 2001 cohort starts in 2003. We define employment as having annual earnings of at least $1,000 (adjusted to 2007 wage levels).

Chart 11 shows cumulative percentages of the 2001 cohort that obtained any employment, by age group. By 2007, 18.6 percent of all the recipients in that cohort had worked in at least 1 year since the second postaward year. Not surprisingly, cumulative employment percentages for the two youngest groups were much higher than for all older groups: 39.4 percent and 27.1 percent of those aged 18–19 and 20–39 at award, respectively, had worked in at least 1 year by 2007, compared with 12.2 percent, 6.4 percent, and 4.2 percent for those aged 40–49, 50–61, and 62–64 at award, respectively.

Chart 12 presents inflation-adjusted average annual earnings for members of the 2001 SSI award cohort who had any earnings. Average earnings increased rapidly from the second year after award (2003) through 2007 for recipients in the three youngest age groups, but the increase for those in the two older groups was modest. The continued rise in average annual earnings among all recipients with positive earnings was likely due in part to recipients with higher earnings continuing to work longer than did those with lower earnings. Another possible factor is that a worker's earnings generally rise with age, particularly before age 50.

Summary and Conclusions

Our analysis illustrates that cumulative rates of employment and work-incentive use among SSI recipients over a multiyear period substantially exceed the annual or monthly cross-sectional estimates. For instance, we find that 19.4 percent of the 2001 SSI award cohort had had PCE in at least 1 year by the end of 2007. By contrast, only 7.9 percent of working-age SSI recipients had earnings from work in the month of December 2007, according to annual statistics published by SSA (2008). Similarly, we find that 9.8 percent of the 2001 SSI award cohort had attained NSTW in at least 1 month by the end of 2007, including the 8.4 percent who maintained their SSI eligibility under Section 1619(b). The latter figure exceeds the published cross-section figure for 1619(b) participation in December of every year from 2001 through 2007 and almost quadruples the largest single cross-section figure for that period (2.2 percent in December 2007).

The differences between the long-term cumulative statistics and the cross-sectional monthly statistics are not surprising because they describe beneficiary activities from two different perspectives. The cross-sectional statistics contain all current recipients in their denominators—including millions of recipients who have been on the rolls for many years and failed to work or use work incentives in the past—and count recipients with PCE or 1619(b) status in a single month only. The cumulative statistics include in their denominators only those recipients who have been on the SSI rolls for no more than 6 years and account for all recipients who achieved PCE or 1619(b) status during any month of the 6-year period. The higher numbers for the long-term cumulative statistics do not imply that more recipients work than the monthly statistics suggest. Instead, they offer a more complete picture—one that is important for understanding the dynamic process of finding work.

A substantial number of SSI awardees receive DI benefits, although their initial DI award might not coincide with their initial SSI award. (Our award cohorts exclude persons who receive SSI only during the 5-month DI waiting period; all others are included.) We find that more than 30 percent of the 2001 SSI award cohort received DI benefits during their award year, and others received DI benefits in subsequent years. We also find that nearly half of the recipients who had at least 1 month in NSTW or 1619(b) status were receiving DI benefits in those months.

Young recipients were much more likely to have their SSI payments suspended or terminated because of work for at least a month than were older awardees. In the 2001 award cohort, 19.2 percent of recipients aged 18–19 at award and 14.9 percent of those aged 20–39 at award achieved NSTW, compared with only 6.3 percent, 3.7 percent, and 1.8 percent of recipients aged 40–49, 50–61, and 62–64 at award, respectively. Although persons aged younger than 40 at award represented only 43.4 percent of the cohort, they accounted for about 73 percent of the NSTW months through 2007.

Only a small minority of persons who gave up SSI payments because of work enrolled in employment services from providers eligible for payment from SSA, and most of those who achieved 1619(b) status and NSTW had not enrolled in such services. Although the employment rates among service enrollees were well above the rates for the 1998 award cohort as a whole, the more favorable outcomes for service enrollees may reflect, at least to some extent, their relatively high interest in higher earnings—a factor that likely explains their enrollment in services in the first place. TTW might therefore have expanded use of these services by those who would have forgone payments for work without the additional help. Outcomes for such recipients may have offset some of the costs, but only if they were higher than the outcomes the same recipients would have achieved without TTW. Our analysis does not provide evidence about the extent to which SSA payments for services were offset by lower benefit expenditures.

We have not produced statistics on another interesting dimension of the extent to which SSI recipients forgo their payments for work: the number of months in which they remain in nonpayment status following the initial suspension or termination of payments for work. Schimmel and others (2013) provided such statistics for persons who experienced their first NSTW month in 2001—separately for those in DI nonpayment status (or otherwise ineligible for DI benefits) and for SSI-only recipients or those receiving concurrent SSI and DI benefits.²² They found that 45 percent of beneficiaries with concurrent benefits were in NSTW 12 months later, gradually declining to 30 percent at 60 months and 20 percent at 96 months. They also found that SSI-only recipients typically returned to current-pay status more quickly; slightly less than half were in nonpayment status for more than 1 month and 27 percent were in that status at 12 months, as were 15 percent at 60 months and 12 percent at 96 months.

Comparison with Long-Term Cumulative Statistics for DI Award Cohorts

In qualitative terms, the long-term cumulative statistics presented here are similar to the cumulative statistics produced by Liu and Stapleton (2011) for DI award cohorts. Quantitative comparisons are more difficult to make, however, because of differences between SSI's and DI's work incentives and differences in beneficiary characteristics. For example, although more SSI recipients had their payments suspended or terminated because of work in at least 1 month than did DI beneficiaries, that might simply reflect the fact that new SSI awardees tend to be younger than DI awardees, and the earnings levels that trigger the suspension of benefits differ between SSI and DI.

Both sets of statistics show that the percentage of awardees that eventually forgoes payments because of work (a long-term cumulative statistic) is far higher than the percentage that forgoes payments because of work in a given month (a monthly statistic), and that relatively young awardees account for a large portion of those who forgo payments because of work. For both DI and SSI, the long-term statistics show that many recipients work without having their payments suspended, even temporarily. For instance, although 19.4 percent of the 2001 SSI award cohort had PCE in at least 1 month by 2007, only 9.8 percent had their payments suspended or terminated because of work for at least a month. Functional limitations and declining health might have prevented recipients from earning enough to stop receiving payments, but perhaps many of those recipients would have done so if more assistance or stronger work incentives had been available.

In addition, both sets of statistics show that a large majority of those awardees who find work and earn enough to forgo all or (in the case of SSI) part of their payments do so in the first 5 years after their award. For example, for the 1996 SSI award cohort, the cumulative percentage with at least 1 month in 1619(b) status by the end of the fifth year after award, 8.9 percent, is equal to 77 percent of the corresponding percentage by the end of 2006, the 10th full year after award. Stapleton and others (2010) found that 4.8 percent of the 1996 DI award cohort had at least 1 month of suspended benefits for work by the end of the fifth year after award, also equal to 77 percent of the comparable figure at the end of 2006.

Policy Implications

Cumulative statistics on employment and work-incentive use by cohorts of SSI and DI beneficiaries paint a substantially different picture than do annual or monthly statistics—not because more beneficiaries are working than the short-term statistics suggest, but because the long-term statistics provide richer information on their work efforts. Compared with the cross-section statistics, the cumulative statistics show that many more awardees are working and on the margin of earning enough to give up some (in the case of SSI) or all of their cash payments at some point after they enter the DI or SSI rolls; that they are much more likely to be on that margin in the first 5 years after award; that those who are young at entry are also much more likely to be on that margin; and that many who work and forgo payments do not use SSA-financed employment services. Cumulative statistics also show that the share of recipients that uses a program work incentive at some point after program entry is much larger than the very small share that is using the incentive in a given month or year; that is, those who use these program features are a substantial, rather than tiny, minority of all recipients. Of course this does not mean that the work incentives had their intended effect of helping recipients earn enough to give up their payments; we do not know what the employment and earnings of recipients would have been in the absence of the incentives, or under alternative work incentives.

Compared with the cross-section statistics, the cumulative statistics also reveal a major challenge to efforts to reduce program costs by making the work incentives more attractive, such as those implemented as a result of the Ticket Act. Such initiatives are likely to increase the use and cost of work incentives to the considerable number of recipients who already earn enough to forgo some or all of their payments. Findings reported by Stapleton, Mamun, and Page (2014) illustrate this point: The introduction of TTW induced more young DI-only beneficiaries to use employment services but had little or no impact on the number of months in which they gave up their benefits for work.²³ Similarly, SSA is testing the Benefit Offset National Demonstration, an initiative that will allow many DI beneficiaries who would otherwise earn enough to give up their benefits for work to retain a portion of those benefits without having to reduce their earnings. Proposals to raise the SSI income and asset disregard thresholds, which have not been adjusted for inflation since the program's 1974 inception, might make work more attractive for many SSI recipients, but would also increase program costs for all recipients who already use the earned-income exclusion.²⁴ Targeting work-incentive enhancements toward individuals who are unlikely to forgo their payments for work under the current incentives would improve the prospects for reductions in benefit costs, but such targeting is likely to encounter administrative and other challenges.

There are, of course, other important reasons to consider making the work incentives more attractive to disability program participants. Most notably, policymakers may wish to support the efforts of people with disabilities to be productive members of society; to help them share the fruits of the American economy; and to enable them to escape from the economic hardships that many (especially those receiving SSI) will otherwise experience.²⁵ Our findings imply that achieving these policy objectives without increasing disability program costs will be challenging.