DI Return-to-Work Milestones

Definition

The sequential pathway a Social Security Disability Insurance (DI) beneficiary may traverse to exit the rolls through employment: (1) enrollment for vocational services (state Vocational Rehabilitation [VR] agency or Employment Network under Ticket to Work), (2) Trial Work Period (TWP) start, (3) TWP completion (9 months of Substantial Gainful Activity (SGA)-level earnings within a rolling 60-month window), and (4) benefits suspended or terminated for work (STW). Achievement of each milestone is rare; most beneficiaries never progress beyond the initial award.

Key Ideas

Milestone rates are low: Within five years of award, only 7.5% enroll for services, 9.8% start TWP, 8.0% complete TWP, and 5.2% achieve STW; these fractions are roughly consistent with the 3.1% sustained work-exit rate found over ten years in Anand and Ben-Shalom (2016)
Age is the dominant predictor: Ages 18–24 reach STW at $5\times$ the rate of ages 50–57 (12.6% vs. 2.3%); the gradient is stable across all four milestones, appears within year 1, and survives full multivariate controls — it is not a proxy for impairment mix
Impairment type runs second: Sensory impairments $\to$ highest rates across all milestones; back/musculoskeletal $\to$ lowest; intellectual disability beneficiaries complete TWP at unexpectedly high rates (11.5%) but rarely achieve STW (4.5%), suggesting supported/sheltered work that does not cross the SGA threshold
Administrative Law Judge (ALJ)-award penalty: Beneficiaries awarded at the ALJ level after appeals are 3–4 percentage points (pp) less likely to complete TWP or achieve STW than those awarded at initial Disability Determination Services (DDS), consistent with the administrative delay $\to$ employment-attachment decay channel in DI Application Costs and Take-Up
Income effect of benefit level: Each $100 increase in monthly DI benefit reduces milestone probability by 1–3 pp across all four outcomes
State heterogeneity persists after controlling for beneficiary characteristics and unemployment; unresolved mix of state VR policy, labor market structure, and disability culture
SSI-as-child paradox: Among young disabled worker (DW) awardees (age 18–39), former Supplemental Security Income (SSI)-as-child disabled workers have the highest STW rates and earnings — $\approx 72\%$ earned more than $1,000 annually in $\geq 1$ year within five years vs. $\approx 40\%$ for DW with no SSI history; they are the most disadvantaged at entry yet most work-oriented in outcomes, likely because they are youngest and have the longest exposure to SSI work incentives and VR services (Ben-Shalom and Stapleton 2013)
Disabled Adult Children (DAC) have far lower work engagement than DW: Disabled Adult Children Program awardees achieve STW and earnings-above-$1,000 at roughly one-fifth the rate of comparable disabled workers; DAC net-suspended-or-terminated-for-work (NSTW) months per 1,000 awardees are trivially small

How It Works

The TWP allows DI beneficiaries to test their ability to work without immediately forfeiting benefits. Nine months of earnings above the TWP threshold (approximately $970/month in 2013; not necessarily consecutive) within a rolling 60-month window constitute completion. After TWP, the beneficiary enters a 36-month Extended Period of Eligibility (EPE): benefits are suspended in any EPE month in which SGA is achieved, and terminated if SGA is sustained beyond the EPE. STW is recorded at the first such suspension or termination. The Ticket to Work program (1999) expanded the menu of Employment Networks beyond state VR agencies, creating a second pathway to service enrollment.

Why It Matters

The strong age gradient implies that early-intervention programs targeting new awardees under 40 will have the highest return: this group is most responsive and, if not supported back to work, will remain on the rolls for decades. The ALJ-delay finding confirms that administrative processing time has a real employment cost beyond the income-support question. The milestone-rate data also reveal that most policy-relevant action (such as it is) occurs within years 1–3 after award — later interventions face steeper employment-detachment.

SSI Parallel (Ben-Shalom and Stapleton 2012; Ben-Shalom et al. 2012)

SSI has a different work-incentive structure (Section 1619(b) instead of TWP/EPE), but the longitudinal dynamics are qualitatively similar. For the 2001 SSI adult award cohort tracked six years: 9.8% had benefits suspended or terminated for work in at least one month vs. $\approx 2\%$ cross-sectional rate — a $4\times$ gap driven by the fact that most spells are brief. The age gradient is even steeper than DI: ages 18–19: 19.2% STW, ages 50–61: 3.7%. Complication: 46.9% of 1619(b) months involved simultaneous DI receipt, so only 4.5% of the cohort achieved 1619(b) purely as SSI beneficiaries. About 78% who achieved 1619(b) had never enrolled in Social Security Administration (SSA)-funded services, paralleling the $\approx 80\%$ no-service figure for DI.

Extended 11-year tracking of the 1996 SSI cohort (Ben-Shalom et al. 2012 full report) reveals a $6\times$ longitudinal-to-cross-sectional gap: 11.9% ever in 1619(b) vs. 1.9% cross-sectional. Only 7.4% achieved 1619(b) without concurrent DI payments. $\approx 40\%$ of the 1996 cohort had received DI benefits by year 11. Under-40 awardees accounted for 78.1% of all 1619(b) months. Employment was 28.1% over 11 years — nearly identical to the DI 1996 cohort (28.0%). 77% of eventual 1619(b) activity for the 1996 cohort occurred within the first five post-award years, mirroring the DI figure exactly.

Cross-cohort: 2001–2003 award cohorts showed notably lower cumulative employment and 1619(b) stats than prior cohorts at the same post-award age, consistent with the 2001 recession. The gross NSTW rate rose for successive cohorts from 2001 onward, but the DI-overlap-free NSTW showed a weaker trend, suggesting the rise was partly driven by an increasing share of concurrent DI/SSI beneficiaries rather than genuine work exit. Cross-state variation is large (NSTW 6.4% WV to 19.9% AK) and closely correlated with DI cross-state rankings (ND, SD, MN, WY high; AL, KY, WV low), with service enrollment rate a strong predictor ( $r = 0.75$ ).

CDR Cessation and Post-Exit Earnings (Hemmeter and Bailey 2016)

Continuing Disability Reviews (CDRs) represent a different pathway off DI rolls than voluntary work return through the TWP: involuntary removal following a Full Medical Review (FMR) determination that the beneficiary no longer meets disability standards. Hemmeter and Bailey (2016) track post-FMR earnings of 2,124,835 DI workers (CDR Waterfall file, 1998–2008 FMRs) using an upper-bound comparison methodology analogous to Bound (1989)'s denied-applicant approach:

Overall cessation rate: 5.7% after all appeals; gradient: 13.4% (ages 18–30) to 2.8% (ages 50–59); 17.6% for neoplasms vs. <4% for congenital/endocrine diagnoses; only 21.5% of CDR-selected beneficiaries ever reach FMR stage
~70% of ceased beneficiaries have any post-FMR earnings (vs. ~27% of continued), but mean earnings are only ~$13,000/year — about 70% of pre-DI levels; ~25% have no post-cessation earnings at all
Only 37% have earnings in all follow-up years; only ~20% sustain earnings above SGA or the poverty threshold in every post-FMR year
The upper bound on the cessation effect is ~43 pp probability of earnings above SGA/poverty/minimum wage — substantially larger than the ~28 pp employment response found at the entry margin (Maestas, Mullen, and Strand 2013), because FMR cessation is pre-selected for high likelihood of medical improvement
Expanding CDRs to marginal cases (Medical Improvement Not Expected [MINE] diary types, low CDR profile scores, mailer recipients) would yield lower average post-cessation earnings — those groups already show worse labor market outcomes within the current cessation population
About 20% of former DI workers return to the program within 8 years (Hemmeter and Stegman 2013)

These findings complement the voluntary-work-return milestone data above: combining the ~5.7% of beneficiaries involuntarily removed via CDRs with the ~5% who achieve STW via TWP, roughly 10% of the DI stock transitions off rolls in a given period — the remainder stay until old-age conversion at Full Retirement Age (FRA).

Open Questions

Why do intellectual disability beneficiaries plateau after TWP completion? Sheltered employment, employer accommodation limits, benefit-loss aversion around Medicaid, or measurement issues with the SGA threshold?
How much of the state fixed effect is traceable to VR system capacity, labor demand composition, or cultural/physician norms around disability certification?
Does the positive Black coefficient on return-to-work persist when controlling for urban/rural status, industry, or local VR agency quality?