GUBMENTPlain talk · policy frontier
Filings / Disability / Sources / §5 Findings: Ticket to Work, VR, and m
GBMT-12 · Research record · No. 12

§5 Findings: Ticket to Work, VR, and measured return-to-work

disability/research/ws05-ticket-vr.md
This is a working research document from the disability filing, published as written — including the parts later corrected. It is the underlying record for Whitepaper No. 12, not a summary of it.

Date: 2026-08-11. Extends Phase 0 anchor 6; locks BOND cross-cite from §4 / anchor 9; adjudicates H5. Completes the Phase 0–owed <2pp employment-bar extract. Programs: Ticket covers DI and SSI disability beneficiaries; VR outcomes enter as the pre-Ticket and Ticket-era service channel. BOND is DI-only.

Search log (PRISMA-lite)

Field Record
Date 2026-08-11
Strings Ticket to Work Mathematica Stapleton Mamun Page STW; Wittenburg Fraker TTW service enrollment 0.4 percentage points; SSA Social Security Bulletin v83n1 Ticket ROI; BOND Final Evaluation earnings null; Hyde Stapleton Ticket participation 4.1%
Sources searched SSA SSB v83n1 (2023) synthesis of Mathematica TTW evaluation; Mamun & Page / Stapleton, Mamun & Page (IZA / Mathematica 2013–14); Wittenburg et al. JVR 2007 initial impacts; Thornton et al. evaluation reports; Abt/Mathematica BOND FER 2018; SSAB 2017 SSI work-incentives statement (participation cite); GSA OES TTW mailer RCT (~≤5% assignment)
Inclusion SSA-commissioned Mathematica technical evaluations with ITT designs; peer-reviewed extracts of those evaluations; BOND as the cliff-redesign comparator
Exclusion EN marketing success stories; participant-only outcome tables offered as population effects; advocacy “Ticket works” summaries without appendices

1. What Ticket is, and what “success” meant

Ticket to Work (TTW) issues eligible DI/SSI disability beneficiaries a “ticket” assignable to Employment Networks (ENs) or state VR agencies (SVRAs), with outcome-oriented provider payments aimed at benefit exit via work. Congress’s load-bearing hope was population-scale exits from cash benefits — not merely counseling contacts.

Mathematica’s evaluation (seven reports over ~11 years) therefore emphasizes intent-to-treat (ITT) effects on the eligible pool. Participant-only success rates are selected and cannot clear H5’s population bar.

2. Participation (eligible-pool fraction)

Source / vintage Participation construct Magnitude
Mathematica early rollout (Phase 1, ~2004–05) Tickets in use / ticket-eligible ~1.1–1.4% in longest-running Phase 1 states
SSB v83n1 citing Eimicke et al. Phase 1 participation by Dec 2005 1.8%
Hyde & Stapleton (via SSAB 2017) Assigned ticket to EN or VR, 2010 4.1% of SSDI+SSI recipients
GSA OES (FY2023 evaluation plan context) Assigned a ticket since inception ≤~5% of eligibles

SSB Table 4 (2006 stock cohort): of 11.1M beneficiaries on the rolls at some point in 2006, 1.2% enrolled in SVRA services that year and 0.04% in EN services. Among non-enrollees, only 3.2% ever reached suspension/termination for work (STW) by end-2016 — selected enrollees do better, which is selection, not an ITT effect.

Confidence: strong that population assignment/enrollment is low-single-digit.

3. Exact effect-size extract (H5 <2pp bar)

3a. Service enrollment (not employment)

Wittenburg, Fraker, Stapleton, Thornton, Gregory & Mamun (Journal of Vocational Rehabilitation, 2007) — longitudinal fixed-effects ITT on 4.7M beneficiaries, first two rollout years:

3b. Employment / benefit-exit (STW) — primary H5 object

Stapleton, Mamun & Page (2013/2014; IZA Journal of Labor Policy / Mathematica seventh-report study), as synthesized in SSA Social Security Bulletin v83n1 (2023):

Power / bound language owed for the pp bar. Stapleton, Mamun & Page state that methods cannot rule out impacts as small as ~5% relative on STW incidence (Thornton’s self-financing threshold ≈ 3,000 new STW cases/year). With annual SGA-exit rates near 0.5% of the worker stock, a 5% relative increase is on the order of **~0.025 percentage points** absolute — far below H5’s <2pp employment/exit bar for the eligible pool. The evaluation’s detectable “large” effects are ruled out; any residual undetectable effect sits orders of magnitude under 2pp.

Locked H5 extract: official SSA-commissioned ITT evaluations put Ticket’s population employment / benefit-exit effects at null, undetectable, or ≪2 percentage points. Service-enrollment impacts that are detected are themselves 0.1–0.4pp — also under the bar, and they are not employment.

Confidence: strong on the <2pp lock from primary evaluation text.

4. VR as channel, not a substitute lever

Pre-Ticket, SVRAs were the near-monopoly public VR channel under cost reimbursement. Under Ticket, most assignments still flow to SVRAs (early evaluation: ~90%+ of in-use tickets). SSB synthesis: combined EN+SVRA success is statistically indistinguishable from SVRA-only success at population scale — ENs did not deliver a new population-level exit technology. Provider payments per STW case may be lower for ENs than SVRAs (Thornton 2012), which is a unit-cost finding, not an employment effect size above 2pp.

5. BOND cross-cite (cliff redesign vs Ticket)

§4 locks BOND: 1 − for2 DI offset → null mean earnings; +1% / +4% benefits paid (Stage 1 / Stage 2); EWIC counseling enhancement null. Relative to Ticket, BOND tests the cliff instrument directly. Neither Ticket nor the BOND offset clears a large population employment effect. Cliff architectures still outrank Ticket redesign for O4 logic (they attack the wrong object Ticket cannot fix), but they inherit BOND’s null-mean-earnings constraint.

6. H5 adjudication

H5 — Ticket to Work is not the binding lever.

Criterion Result
Primary SSA-commissioned evaluations show employment or benefit-exit effects <2pp on the eligible pool Met. ITT STW/employment effects null or undetectable; any effects below detection sit ≪2pp absolute. Detected service-enrollment effects 0.1–0.4pp.
Refuted only if effects exceed the bar in ≥2 independent evaluations Not met. No equal-quality evaluation located that clears ≥2pp population employment/exit.

Verdict: SUPPORTED.

Confidence: strong.

Implications

← All Disability research documents Sources digest Read the whitepaper