Return to Work of Disability Insurance Beneficiaries Who Do and Do Not Access State Vocational Rehabilitation Agency Services

Abstract

This study examines the relationship between services provided by state vocational rehabilitation agencies (SVRAs) and return-to-work outcomes of Social Security Disability Insurance (DI) beneficiaries. DI beneficiaries who enrolled in SVRA services were compared with matched and unmatched comparison groups of beneficiaries who did not enroll in these services. We examined the progression to substantial employment milestones for DI beneficiaries over a 10-year period beginning with their entry into the DI program. Employment outcomes of the SVRA enrollee group are substantially better than those of their matched and non-matched non-enrollee counterparts, and the timing of their employment outcomes is strongly associated with the timing of vocational rehabilitation (VR) enrollment. These findings indicate that differences in employment outcomes between DI beneficiaries who received VR services and those who did not are not simply due to observable differences.

Keywords

economics social security employment vocational rehabilitation

A critical concern of policymakers related to the growth of the Social Security Disability Insurance (DI) beneficiary population is the low exit rate for DI beneficiaries. Employment statistics based on cross-sectional data published by the Social Security Administration (SSA) indicate that benefits are terminated because of paid work for less than 1% of beneficiaries in a given year (Roessler, 2002; SSA, 2012). Recent longitudinal analyses of employment among DI beneficiaries by Liu and Stapleton (2011) found that 10 years after their entry into the DI program, 3.7% of beneficiaries entering in 1996 had their benefits terminated because of work. Although the longitudinal estimate of termination of DI benefits due to work is higher than the cross-sectional estimates, the overall exit rate is still relatively low.

The state–federal vocational rehabilitation (VR) program plays a role in facilitating return to work among DI beneficiaries who access its services, and thus it can potentially contribute to reducing the cost of the program while enhancing the economic independence of DI recipients. For example, among VR service recipients who were closed from program rolls in fiscal years 2009 to 2011, a total of 267,160 had reported receiving DI benefits at application, accounting for about 15% of the VR population during that period. The VR program serves approximately 1 million individuals a year and spends more than US$2.5 billion annually to support persons with disabilities as they achieve their independent living and employment goals (Dutta, Gervey, Chan, Chou, & Ditchman, 2008; Martin, West-Evans, & Connelly, 2010; U.S. Government Accountability Office [GAO], 2005). Currently, there are state vocational rehabilitation agencies (SVRAs) in all 50 states and U.S. territories.

In this article, we attempt to identify the relationship between SVRA service receipt and DI beneficiaries’ achievement of work milestones of trial work period (TWP) and suspension or termination from the DI program due to work (STW). This article makes the following key contributions:

We assess the relationship between SVRA services and the return-to-work outcomes for an annual cohort of all new DI awardees, a new approach in the literature. We compare outcomes for beneficiaries who enrolled for VR services and for a comparison group matched on observed beneficiary characteristics. Employment outcomes of the VR enrollee group are substantially better than those of their matched and non-matched non-enrollee counterparts, and the timing of their employment outcomes is strongly associated with the timing of VR enrollment.

Our results underscore the difficulty in identifying for DI beneficiaries who enrolled for SVRA services an equivalent comparison group of beneficiaries who did not enroll for such services—Even after matching enrollees and non-enrollees on observed characteristics, important unobserved differences remain. We provide new evidence that matching on observed characteristics may not fully address the issue of self-selection of DI beneficiaries who seek VR services and also engage in employment; consequently, identifying the causal effect of VR services on the likelihood of employment among DI beneficiaries remains challenging.

Recent Research on VR Services and Outcomes for DI Beneficiaries

The SVRAs and SSA have carved out unique administrative procedures to facilitate return to work for DI beneficiaries. SVRAs apply several time-consuming eligibility criteria before applicants can receive services (see Note 1). However, SVRAs exempt SSA disability beneficiaries from these criteria to facilitate timely access to services, a policy related to enhanced employment outcomes among DI recipients in the VR program (Honeycutt & Stapleton, 2013). In addition, SSA has created several administrative mechanisms to pay the SVRAs for services provided to DI beneficiaries (Schimmel, Stapleton, Mann, & Phelps, 2013). Under what is referred to as the “traditional cost reimbursement system,” SSA reimburses the SVRAs for the cost of services provided to people receiving DI benefits. For the agencies to receive reimbursement, DI beneficiaries must return to work for at least 9 continuous months at a substantial earnings level, referred to as substantial gainful activity (SGA). Within SSA’s Ticket to Work (TTW) program, SVRAs can also be compensated under an outcomes-only or milestone-outcomes payment system. In the outcomes-only system, SVRAs are eligible for payments only in months when the TTW participant does not receive DI cash benefits because of work (earning above SGA). Under the milestone-outcomes payment system, SVRAs can receive payments in the months when a DI beneficiary achieves intermediate employment outcomes that fall short of achieving the status of not receiving DI cash benefits because of work. It is also possible within the milestone-outcomes system for the SVRAs to receive payments in months when beneficiaries do not receive cash benefits because of work. SVRAs must choose one of the TTW payment systems. However, they can request payment under the traditional system rather than the TTW payment systems on a case-by-case basis.

Research findings regarding the complex relationship between receiving VR services and utilization of SSA disability benefits are varied and inconsistent. Substantial numbers of VR service recipients enter the DI program after being engaged with the SVRAs; a recent study found that more than 11% of first-time VR applicants in 2003 entered DI in the 5 years following their VR application (Stapleton & Martin, 2012). Another descriptive study (GAO, 2007) found that among Social Security disability program beneficiaries who completed VR services between 2000 and 2001, only about 10% left the disability rolls for at least 1 month within 3 years. However, many beneficiaries who leave the benefit rolls come back to them; recent estimates suggest that Social Security disability program beneficiaries who assigned their ticket to a SVRA and stopped receiving cash benefits because of work remained on non-payment status for about 5 months, on average, over an 18-month period following first achieving non-payment status (Schimmel et al., 2013). Finally, Dean and his associates also found mixed effects of VR services on employment and earnings among VR service recipients in Virginia. Although their fixed-effects estimates show positive association between VR services and labor market outcomes (Dean, Pepper, Schmidt, & Stern, 2014b), analyses of data for those with mental illness and cognitive impairment suggest that the effect of VR services varied by different types of services (Dean, Pepper, Schmidt, & Stern, 2013, 2014a).

This article contributes to the related literature in two ways:

We provide new information on the progression to DI milestones (TWP completion and STW) by VR enrollees that has not yet been previously analyzed.

We adjust the outcomes for observable characteristics by matching DI beneficiaries who enrolled in SVRA services with those who did not enroll for such services using propensity score matching. We first compare the employment outcomes for a “matched” group of DI beneficiaries. This adjustment allows us to assess variations in outcomes that might be due to observable characteristics that were important drivers of outcomes noted in previous studies, such as impairment.

In addition, we show outcomes for a group of non-matched beneficiaries who did not apply to receive VR services over a 10-year period to illustrate how the estimates vary without any adjustments for observable characteristics (i.e., the outcomes that might be observable in simple descriptive statistics); we also examine the relationship between the timing of VR enrollment and employment outcomes.

Data and Method

Data Sources

The data sources for this research included two administrative data sets that were matched prospectively: the 2010 Disability Analysis File (DAF), formerly known as the Ticket Research File, and the U.S. Department of Education, Rehabilitation Services Administration (RSA) Case Service Report (RSA-911) files. The DAF contains longitudinal and one-time data from SSA records on people who participated in the Supplemental Security Income (SSI) or DI programs while between the ages of 10 and 65 for at least 1 month from March 1996 through December 2010 (Hildebrand et al., 2011). Longitudinal data include variables such as state of residence, impairment codes, and benefit payments, whereas one-time data include variables such as Social Security number and date of birth. Data from SSI and DI sources are combined into an individual record for each beneficiary. SSA continues to update the DAF annually, adding the records of new beneficiaries and updating the records of those previously included.

The RSA-911 data matched with the DAF were drawn from the RSA case service reports (Form 911) compiled and made available each year. Each annual RSA-911 file includes records of demographic and programmatic data for each person who had enrolled in the RSA VR program and was closed from the program during a given fiscal year. The RSA files that were linked to the DAF contained selected variables from each annual RSA-911 file from 1998 through 2010.

There is a long lag between the time an individual with a disability applies for SVRA services and when his or her data become available because data on SVRA participation are only reported after a case is closed by the agency. For example, if a participant entered the VR program in 2003 and exited in 2006, there would be no record of his or her participation in the 2003, 2004, or 2005 files. To account for the fact that it may take several years for a beneficiary’s VR case to be closed, we examine beneficiaries’ VR enrollment status annually for up to 5 years after their DI award (see Note 2). Beneficiaries who have multiple enrollments in VR services in a 5-year period were combined into the single-person-level record. Altogether, the data allow us to identify the timing of when beneficiaries enrolled in VR services as well as when they achieved an employment milestone. We then look at the beneficiaries’ return-to-work outcomes up to 9 years after they enrolled for VR services.

Analysis Sample

We matched the DI beneficiaries who enrolled in SVRA services with those who had never enrolled for such services. The enrollee and comparison groups were drawn from all DI beneficiaries who entered the DI program in 2000, were between 25 and 54 years of age at DI award, and did not die before 2010. The enrollee group consisted of 17,369 DI beneficiaries (a) who applied for VR services in Fiscal Year 1997 or later and within 5 years of DI award (i.e., before 2006), (b) who were closed from VR services after an individual plan for employment had been developed, and (c) whose time from first DI application date to last VR case closure was less than 5 years.

To further assess the relationship between the timing of service receipt and employment outcomes, the VR-service-enrolled group was further divided into six subgroups, depending on when individuals applied for VR services. The following six subgroups consisted of those individuals who had applied for SVRA services:

before receiving DI award,

less than 1 year following DI award,

between 1 and 2 years following DI award,

between 2 and 3 years following DI award,

between 3 and 4 years following DI award, and

between 4 and 5 years following DI award (see Figure 1).

Figure 1.

Flow diagram for the analysis sample.

These subgroups provide the equivalent of a phased rollout design; using them, we can see whether the act of applying for VR services systematically influences the rate at which DI beneficiaries achieve the employment milestones. In such a design, baseline performance is measured on several groups of participants who receive the intervention at different points in time.

We matched applicants with their nearest neighbor non-applicants using propensity scores based on the beneficiaries’ demographic characteristics as well as certain SSA programmatic attributes. The comparison group was drawn from the 167,891 DI awardees in 2000 who never had a closure from VR services between Fiscal Years 1997 and 2010 (shown as the SVRA non-applicant group in Figure 1). To identify the matched comparison group of beneficiaries, we matched the following demographic characteristics: gender, age in years (25–29, 30–34, 40–44, 45–49, and 50–54), race/ethnicity (non-Hispanic Black, non-Hispanic White, and Hispanic), education (0–11 years, 12 years, 13–15 years, and 16 years or more), and primary impairment (affective disorder, psychiatric disorder, intellectual disability, severe vision, severe hearing, back disorder, other musculoskeletal, HIV/AIDS, cerebral palsy, multiple sclerosis, other physical, and other impairment). Many demographics or individual characteristics have been shown to be related to employment outcomes for the general population of people with disabilities (Sevak, Houtenville, Brucker, & O’Neill, 2014), as well as for Social Security Disability beneficiaries (Ben-Shalom & Mamun, 2013; Mann, Mamun, & Hemmeter, in press). Programmatic attributes included in the matching process are the adjudication level for DI award decision (Disability Determination Services, the administrative law judge in the Office of Hearing and Appeals, or higher levels), whether the beneficiary received a TTW within 5 years of the DI award date, average indexed monthly earnings, and average benefit amount at award (see Table 1).

Table 1.

Comparison of Applicant and Non-Applicant DI Beneficiaries: Before and After Matching.

	Before matching			After matching
Demographics	SVRA app^a	SVRA non-app^b	Test of diff. (p value)	SVRA app^a	SVRA non-app^a	Test of diff. (p value)
Female	44.1	50.1	.00	44.1	44.7	.29
Age (in years)
25–29	11.3	5.6	.00	11.3	11.7	.33
30–34	13.1	8.7	.00	13.1	13.3	.47
35–39	18.6	14.4	.00	18.6	18.9	.46
40–44	20.8	19.1	.00	20.8	20.8	.88
45–49	19.9	22.7	.00	19.9	18.9	.02
50–54	16.4	29.5	.00	16.4	16.4	.99
Race/ethnicity
Black (not Hispanic)	16.0	16.7	.04	16.0	16.4	.32
White (not Hispanic)	75.1	72.1	.00	75.1	73.7	.00
Other, missing/unknown	2.2	2.8	.00	2.2	2.4	.37
Hispanic	4.7	6.6	.00	4.7	5.2	.04
Education (in years)
0–11	12.2	17.0	.00	12.2	12.3	.91
12	41.0	37.6	.00	41.0	40.9	.85
13–15	16.6	13.0	.00	16.6	17.2	.10
16+	8.6	6.8	.00	8.6	8.5	.88
Missing	21.6	25.5	.00	21.6	21.1	.21
Primary impairment
Affective disorder	19.2	17.2	.00	19.2	19.4	.62
Psychiatric disorder	15.6	11.0	.00	15.6	15.4	.68
Intellectual disability	6.7	5.5	.00	6.7	6.4	.14
Severe vision loss	4.9	1.3	.00	4.9	4.8	.82
Severe hearing loss	1.8	0.4	.00	1.8	1.9	.69
Back disorder	12.4	17.0	.00	12.4	12.4	.97
Other musculoskeletal	7.8	11.0	.00	7.8	8.0	.59
HIV/AIDS	1.2	1.7	.00	1.2	1.2	1.00
Cerebral palsy	0.8	0.5	.00	0.8	1.0	.07
Multiple sclerosis	2.4	2.5	.76	2.4	2.6	.26
Other physical	27.0	31.8	.00	27.0	26.9	.73
Other impairment	0.2	0.2	.75	0.2	0.1	.36
Adjudicative level of DI award
Initial DDS	57.3	51.3	.00	57.3	56.4	.09
Reconsideration of DDS	19.6	21.3	.00	19.6	20.4	.06
Administration law judge in Office of Hearing and Appeals or higher	20.2	23.6	.00	20.2	20.3	.81
Missing	2.9	3.9	.00	2.9	2.9	.95
Received Ticket to Work within 5 years of DI award date	12.8	0.2	.00	12.8	12.7	.88
Average indexed monthly earnings (US$)	1,504	1,682	.00	1,504	1,488	.15
Average benefit amount at award date (US$)	776	824	.00	776	771	.10

Note. All categorical variables were tested using a chi-square test. The two continuous variables, indexed monthly earning and benefit amount, were tested using a t test. The matching also included a squared term for each of these continuous variables. SVRA app = state vocational rehabilitation agency applicant; SVRA non-app = state vocational rehabilitation agency non-applicant; DI = Social Security Disability Insurance; VR = vocational rehabilitation; DDS = Disability Determination Services.

n = 17,369. ^bn = 167,890.

Analysis

We constructed dichotomous dependent variables to measure two return-to-work outcomes using data from the DAF: TWP and suspension or termination from the DI program due to work (STW; see Note 3). We measured these outcomes at 1-year intervals following DI award; for example, we measured whether beneficiaries ever completed TWP within 1 year, within 2 years, and so forth, for up to 10 years after DI award.

We selected these two return-to-work outcomes because they capture two important employment support provisions intended to assist DI beneficiaries in their efforts to become self-sufficient through work (SSA, 2014). In addition, DI beneficiaries achieving these outcomes are directly related to SSA reimbursements to SVRAs for the cost of services provided, or milestone and/or outcome payments under the TTW program. The TWP allows the DI recipient to test his or her ability to work for at least 9 months. The TWP begins when a beneficiary starts working and earning above US$770 per month (in 2014) or if the person works more than 80 hr in self-employment in a month (see Note 4). The TWP continues until the beneficiary accumulates 9 TWP months (not necessarily consecutive) within a rolling 60-month period. During the TWP, beneficiaries receive full DI benefits, regardless of their earnings level, as long as the work activity is reported and the beneficiary continues to have a disabling impairment.

After the TWP is completed, benefits may be suspended or terminated, depending on the beneficiary’s level of employment and earnings. Following TWP completion, beneficiaries enter an Extended Period of Eligibility (EPE). During the EPE, the beneficiary is entitled to benefits in all months when not engaged in SGA—that is, in months that earnings or work activities are below the SGA level (in 2014, SGA was US$1,800/month for those who are blind and US$1,070/month for all others) after any allowed disregards, provided that he or she continues to have a disabling impairment. Benefits are suspended in the months of SGA, except in 3 grace period months. After a 36-month re-entitlement period and completion of any remaining grace period months (which may occur during or after the 36 months), the entitlement to benefits ends in the first month in which the beneficiary is engaged in SGA. Otherwise, the EPE continues, and the beneficiary may continue to work at below the SGA level indefinitely, without loss of benefits.

We used logistic regression to estimate the relationship between receipt of VR services and achievement of an employment-related outcome over the 10 years following DI award. The regression adjustment allows us to further adjust for any chance differences that exist after matching (of which there are few) and characteristics not included in the match, such as state. In addition, the regression adjustment increases the precision of the estimated relationship between VR services and achievement of the employment milestones. We estimated a series of 10 logistic regressions for each of the two employment milestones—TWP completion and STW—analyzing the achievement of the milestone at every year after DI award for 10 years. In each regression, we compared the five enrollee subgroups with the comparison group. We used individual characteristics, state of residence, and the propensity score as controls in each regression model.

Results

Propensity Score Matching

Propensity score matching resulted in a comparison group where there were few statistical differences between it and the VR enrollees group. As shown in Table 1, prior to matching, the differences in means between DI beneficiaries who did and did not enroll in SVRA services were statistically significant at the 5% level for all but 2 of the 36 variables we tested. For instance, enrollees were more likely than non-enrollees to be female, better educated, and younger, and to have received a TTW within 5 years of DI award. In addition, enrollees received lower benefit amounts than did non-enrollees. After matching, the enrollee and matched comparison groups were not statistically different on 33 of the 36 variables at the 5% level (see Table 1). Even for the three variables that showed statistically significant differences in means—indicators for ages 45 to 49, Hispanic ethnicity, and non-Hispanic White—the differences are substantively small in absolute terms (between 0.5 and 1.5 percentage points).

Regression-Adjusted Estimates on TWP Before/After Beneficiaries Applied for SVRA Services

Figure 2 shows that the estimated rates of TWP completion among the enrollee subgroups were greater than those for the matched and unmatched comparison groups in the years before and after the beneficiaries applied for SVRA services. Each panel in Figure 2 graphically shows the estimated mean TWP completion rates for an enrollee subgroup and the comparison groups for each year after their DI award; the vertical distance between the line for an enrollee subgroup and that for the matched comparison group represents the mean differences shown in Appendix Table A1. Each of the Panels b through f in Figure 2 indicates the timing of the SVRA application with a black vertical line; for Panel a, we cannot indicate the timing because SVRA applications occurred before DI award for this enrollee subgroup. By comparing the part of the graphs to the left of the vertical line in Panels b through f, we can see that for each enrollee subgroup, the differences between enrollees and the matched comparison group are smaller than the differences between enrollees and the unmatched comparison group (see Note 5). Thus, identifying a matched comparison group using propensity score matching helped reduce the underlying observed differences between the enrollees and the non-applicants. However, the differences in TWP completion remain between enrollees and the matched comparison group before the beneficiaries applied for SVRA services.

Figure 2.

Estimated rates of TWP completions among SVRA enrollees, matched and unmatched comparison groups by SVRA application time relative to DI award.

More specifically, the estimated differences in TWP completion before the beneficiaries applied for SVRA services vary across the five enrollee subgroups (shown in Panels b–f in Figure 2). The largest absolute differences (between 3 and 8 percentage points) appear for the enrollee subgroup whose members applied for SVRA services 4 to 5 years after their DI benefit award. For the other four enrollee subgroups, the differences during the period before application range between 2 and 5 percentage points. However, the relative size of the differences in TWP completion among enrollees relative to the matched comparison group ranges between 50% and 80% across all enrollee subgroups. Almost all these differences are statistically significant at least at the 5% level and are also substantively important relative to the TWP completion rate observed for the matched comparison group (see Note 6).

For every year after the beneficiaries applied for SVRA services, the estimated differences between the enrollee subgroups and the matched comparison group in TWP completion rates are positive, substantial, and statistically significant. In Figure 2, the differences between the enrollee subgroups and the matched comparison group are shown by the gap between the chart lines for each group to the right of the black vertical line in each panel. As shown in the figures, the proportion of DI beneficiaries completing TWP has a consistent pattern of increasing over time after applying for VR services; the size of the difference between enrollees and matched comparison group increases in the first few years after SVRA application and levels off in the later years.

For those who applied for SVRA services 1 to 2 years after DI award (Panel c in Figure 2), the average percentage point difference in TWP completion compared with the matched comparison group increases from 3 percentage points in the year before the SVRA application to 7 percentage points in the year after SVRA application, then increases rapidly to 15 percentage points 5 years after application (i.e., 7 years after DI award), and levels off at 16 percentage points in the last 2 years of the observation period. Relative to the matched comparison group, these positive differences translate to between an 82% and 120% increase in the fraction achieving TWP completion. A very similar pattern appears for the enrollee subgroups whose members applied for SVRA services 2 to 3 years after DI award (Panel d), 3 to 4 years after DI award (Panel e), and 4 to 5 years after DI award (Panel f).

Regression-Adjusted Estimates on STW Before/After Beneficiaries Applied for SVRA Services

A similar picture appears when we look at the estimated differences between the enrollee subgroups and the comparison groups in the achievement of STW before and after the beneficiaries applied for SVRA services (see Figure 3). The differences in the achievement of STW between enrollees and the matched comparison group before applying for SVRA are smaller than the differences between enrollees and the unmatched comparison group, which suggests that propensity score matching helped reduce the underlying unobserved differences between the enrollees and the non-applicants. The differences between enrollees and the matched comparison group are not statistically significant for most enrollee subgroups, with the exception of those who applied between 2 and 3 years (Panel d in Figure 3) and 4 to 5 years after DI award (Panel f in Figure 3; see Note 7).

Figure 3.

Estimated rates of STW achievement among SVRA enrollees, matched and unmatched comparison groups by SVRA application time relative to DI award.

Differences in achievement of STW between the enrollee subgroup and the matched comparison group widen steadily in the first few years following the year of application and then levels off in the later part of the observation period (see Figure 3). For those who applied for SVRA services 1 to 2 years after DI award (Panel c in Figure 3), the average percentage point difference in achievement of STW compared with the matched comparison group increases from −1 percentage point in the year before the SVRA application to 2 percentage points in the year after SVRA application, then increases rapidly to 7 percentage points 5 years after application, and levels off at 7 percentage points in the remainder of the observation period. Relative to the matched comparison group, these positive differences translate to between a 46% and 100% increase in the fraction achieving STW. The pattern is similar for the other enrollee subgroups, namely, those who applied for SVRA services 2 to 3 years after DI award (Panel d), 3 to 4 years after DI award (Panel e), and 4 to 5 years after DI award (Panel f).

Although the above analyses control for observed differences in beneficiary characteristics, an important caveat is that the findings are not impacts. For example, although we made every attempt to minimize differences in characteristics between enrollees and non-enrollees, we still observe small differences in the rate enrollees achieved TWP completion and STW before they applied for SVRA services, suggesting that beneficiaries might have returned to work even in the absence of VR service.

Although the sharp increases in the achievement of TWP completion and STW for each enrollee subgroup between the year before and after SVRA application are encouraging, they may not indicate the impact of SVRA services. Indeed, they may still reflect beneficiaries self-selecting the timing of their SVRA application, which may in turn reflect unobserved changes in the ability or preferences for engaging in paid employment. Consequently, the sharp increases in differences in return-to-work outcomes between the enrollee and the matched comparison group immediately after SVRA application do not necessarily reflect solely the effect of VR services on the outcome. The estimated differences in return-to-work outcomes may be confounded with the selection effect. Even for the subsequent years after SVRA application, we cannot rule out that much of the differences reflect self-selection bias, even though the estimated differences in outcomes are found to grow over time before leveling off in the later years.

Discussion

In this article, we identified 17,369 DI beneficiaries between the ages of 25 and 54 in the 2000 cohort of new DI awardees who had applied for and enrolled in SVRA services. Using propensity score matching, we also created a matched comparison group from the 167,890 DI beneficiaries who did not apply for SVRA services. Our results illustrate the trajectory of employment outcomes of SVRA enrollees and how they compared with the matched comparison group. Results for the period after SVRA enrollment suggest that DI beneficiaries in the enrollee group had a higher rate of completing TWP and achieving STW than did the matched comparison group. More specifically, we found an increase in the rate of achievement of these outcomes following SVRA application that continued to increase before leveling off a few years following the application. Thus, SVRA services are positively associated with higher rates of TWP completion and achievement of STW among working-age DI beneficiaries in our sample, after controlling for their characteristics at DI entry.

More robust outcome measures could be considered for future research. In this study, STW achievement was noted when it was achieved for the first time, producing a cumulative measure over time that never gets smaller and does not account for people coming on and off the DI rolls. A more robust measure that accounts for DI beneficiaries churning on and off the rolls would be to count the number of months per year that DI beneficiaries achieve STW. Another set of robust employment outcome measures would be longitudinal earnings and taxes paid. Such data would allow researchers to look at earnings before, during, and after DI enrollment and SVRA services. In addition, tracking federal income, Social Security, and Medicare taxes paid would allow researchers and policymakers to understand the marginal contributions to society made by work-oriented beneficiaries who do work but do not achieve STW on a consistent basis.

Footnotes

Appendix

Table A2.

Adjusted Mean Differences for STW (Marginal Effects From Logistics Regressions).

SVRA application	(1) STW by Year 1	(2) STW by Year 2	(3) STW by Year 3	(4) STW by Year 4	(5) STW by Year 5	(6) STW by Year 6	(7) STW by Year 7	(8) STW by Year 8	(9) STW by Year 9	(10) STW by Year 10
Before DI	0.011** [0.008, 0.013]	0.018** [0.014, 0.022]	0.026** [0.021, 0.031]	0.033** [0.027, 0.039]	0.037** [0.030, 0.044]	0.038** [0.030, 0.047]	0.041** [0.031, 0.050]	0.041** [0.032, 0.050]	0.041** [0.032, 0.049]	0.040** [0.031, 0.049]
<1 year after DI	0.004* [0.000, 0.008]	0.021** [0.016, 0.027]	0.044** [0.037, 0.051]	0.061** [0.052, 0.070]	0.073** [0.064, 0.083]	0.079** [0.068, 0.090]	0.084** [0.072, 0.097]	0.086** [0.073, 0.099]	0.086** [0.073, 0.099]	0.087** [0.073, 0.100]
1–2 years after DI	−0.009* [−0.017, −0.000]	−0.008 [−0.020, 0.004]	0.018* [0.004, 0.031]	0.041** [0.029, 0.053]	0.054** [0.039, 0.068]	0.062** [0.048, 0.076]	0.068** [0.052, 0.084]	0.069** [0.054, 0.085]	0.073** [0.057, 0.089]	0.073** [0.057, 0.089]
2–3 years after DI	0.004 [−0.002, 0.010]	0.005 [−0.005, 0.015]	0.017** [0.006, 0.029]	0.039** [0.023, 0.054]	0.049** [0.035, 0.064]	0.063** [0.047, 0.079]	0.070** [0.052, 0.089]	0.078** [0.059, 0.096]	0.078** [0.059, 0.097]	0.078** [0.058, 0.098]
3–4 years after DI	0.005 [−0.002, 0.012]	0.008 [−0.002, 0.018]	0.004 [−0.008, 0.016]	0.010 [-0.008, 0.029]	0.028** [0.012, 0.045]	0.043** [0.028, 0.058]	0.053** [0.037, 0.070]	0.067** [0.048, 0.086]	0.069** [0.049, 0.089]	0.068** [0.048, 0.088]
4–5 years after DI	0.008* [0.001, 0.016]	0.017** [0.007, 0.028]	0.019** [0.006, 0.031]	0.021** [0.006, 0.035]	0.019 [-0.002, 0.040]	0.038** [0.019, 0.057]	0.064** [0.040, 0.088]	0.072** [0.050, 0.094]	0.075** [0.052, 0.098]	0.076** [0.053, 0.098]
Observations	34,684	34,726	34,738	34,738	34,738	34,738	34,738	34,738	34,738	34,738

Note. Estimates are adjusted for demographic characteristics, propensity score, and state. Confidence intervals are shown in parentheses; they are adjusted for clustering at the state level and also robust to heteroskedasticity. The shaded cells show differences in means before beneficiaries applied for SVRA services. STW = suspension or termination of benefits due to work; SVRA = state vocational rehabilitation agency; DI = Social Security Disability Insurance.

p < .05. **p < .01.

Acknowledgements

The authors would like to thank David Stapleton and David Wittenburg for their insightful comments on an earlier version of the manuscript, and Xiao Barry for very helpful research assistance.

Authors’ Note

The contents of this report do not necessarily represent the policy of the U.S. Department of Education or any other federal agency (Education Department General Administrative Regulations, 75.620 [b]). The authors are solely responsible for all views expressed and any errors or omissions.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the National Institute on Disability and Rehabilitation Research, U.S. Department of Education, through its Rehabilitation Research and Training Center on Disability Statistics and Demographics grant to Kessler Foundation, under cooperative agreement H133B12006.

Notes

References

Ben-Shalom

Mamun

(2013). Return-to-work outcomes among Social Security Disability Insurance program beneficiaries (Mathematica working paper). Washington, DC: Mathematica Policy Research.

Dean

Pepper

Schmidt

Stern

(2013). The effects of vocational rehabilitation for people with cognitive disabilities. Unpublished manuscript.

Dean

Pepper

Schmidt

Stern

(2014a, January). The effects of vocational rehabilitation for people with mental illness. Unpublished manuscript.

Dean

Pepper

Schmidt

Stern

(2014b). State vocational rehabilitation programs and federal disability insurance: An analysis of Virginia’s vocational rehabilitation program. IZA Journal of Labor Policy, 3(7), 1–19.

Ditchman

M. Y.

Chan

Fitzgerald

(2013). Vocational rehabilitation. In Strauser

(Ed.), Career development, employment, and disability in rehabilitation: From theory to practice (pp. 343–360). New York, NY: Springer.

Dutta

Gervey

Chan

Chou

C. C.

Ditchman

(2008). Vocational rehabilitation services and employment outcomes of people with disabilities: A United States study. Journal of Occupational Rehabilitation, 18, 326–334.

Guo

Fraser

M. W.

(2010). Propensity score analysis: Statistical methods and analysis. Thousand Oaks, CA: SAGE.

Hildebrand

Kosar

Page

Barry

Loewenberg

Fischer

. . .Hazelwood

(2011). User’s guide for the Ticket Research File (TRF10). Washington, DC: Mathematica Policy Research.

Honeycutt

Stapleton

(2013). Striking while the iron is hot: The effect of vocational rehabilitation service wait times on employment outcomes for applicants receiving Social Security benefits. Journal of Vocational Rehabilitation, 39, 137–152.

10.

Liu

Stapleton

(2011). Longitudinal statistics on work activity and use of employment supports for new Social Security disability insurance beneficiaries. Social Security Bulletin, 71(3), 35–60.

11.

Mann

Mamun

Hemmeter

(in press). Employment, earnings, and primary impairments among beneficiaries of Social Security disability programs. Social Security Bulletin.

12.

Martin

West-Evans

Connelly

(2010, Summer). Vocational rehabilitation: Celebrating 90 years of careers and independence. American Rehabilitation (Special Edition), pp. 15–18.

13.

Roessler

(2002). TWWIIA initiatives and work incentives: Return to work implications. Journal of Rehabilitation, 68, 11–16.

14.

Schimmel

Stapleton

Mann

Phelps

(2013, July). Participant and provider outcomes since the inception of Ticket to Work and the effects of the 2008 regulatory changes (Report submitted to the Social Security Administration). Washington, DC: Mathematica Policy Research, Center for Studying Disability Policy.

15.

Sevak

Houtenville

Brucker

O’Neill

(2014, April). The influence of age, gender, race and education on employment, earnings and hours worked. Paper presented at the State-of-the-Science Conference: Advancing Evidence-Based Practices and Policies to Close the Employment Gap, Bethesda, MD.

16.

Social Security Administration. (2012). Annual statistical report on the Social Security Disability Insurance Program, 2011. Baltimore, MD: Office of Retirement and Disability Policy.

17.

Social Security Administration. (2014). 2014 red book: A summary guide to employment supports for persons with disabilities under the Social Security Disability Insurance and Supplemental Security Income programs. Baltimore, MD: Office of Retirement and Disability Policy.

18.

Stapleton

Martin

(2012, September). Vocational rehabilitation on the road to Social Security disability: Longitudinal statistics from matched administrative data (Michigan Retirement Research Center working paper 2012-269). Ann Arbor, MI: Michigan Retirement Research Center.

19.

U.S. Government Accountability Office. (2005). Vocational rehabilitation: Better measures and monitoring could improve the performance of the VR program (GAO-05-865). Washington, DC: Author.

20.

U.S. Government Accountability Office. (2007). Vocational rehabilitation: Earnings increased for many SSA beneficiaries after completing VR services, but few earned enough to leave SSA’s disability rolls (GAO-07-332). Washington, DC: Author.