GUBMENTPlain talk · policy frontier
Filings / Disability / Sources / §4 Findings: The work cliff and Medica
GBMT-12 · Research record · No. 12

§4 Findings: The work cliff and Medicaid/Medicare link

disability/research/ws04-work-cliff.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 anchors 7 and 9; adjudicates H4. Programs: SSI 1619(b) and SSDI cash-cliff / extended Medicare are separate rule stacks. Ticket take-up is cited only as a participation denominator for H4’s usability leg; population employment effects live in §5.

Search log (PRISMA-lite)

Field Record
Date 2026-08-11
Strings SSA Red Book TWP EPE SGA 1619(b); SSI ASR 1619(b) Table 40 43; BOND Final Evaluation Report Gubits 2018; Medicaid Buy-In workers with disabilities Gettens; Mathematica three E’s Medicaid Buy-In enrollment earnings
Sources searched SSA Red Book (SSDI/SSI employment supports; 2024 parameter page); SSI ASR 2023 §7 (1619); Abt/Mathematica BOND FER Vol. 1 + Abt 2019 brief; Washington MBI evaluation (Gettens et al. 2012); Mathematica MBI “three E’s” / Liu & Weathers earnings notes; BPC 2022 MBI data-gap brief; KFF buy-in state count
Inclusion Agency primary rules and published take-up tables; SSA-commissioned RCT/demo evaluations; peer-reviewed or Mathematica state/national MBI outcome studies with transparent comparison design
Exclusion Vendor “how to keep benefits” marketing; equating Ticket counseling with cliff redesign; treating BOND null as proof beneficiaries “don’t want to work”

1. Rule stack that binds (Red Book)

SSDI cash path (Title II):

Stage Rule (construct) 2024 parameters (Red Book)
Trial Work Period (TWP) 9 service months in a rolling 60-month window; full cash benefits continue regardless of earnings if disability continues Service-month threshold $1,110/mo
Extended Period of Eligibility (EPE) 36-month re-entitlement: cash paid when earnings < SGA; suspended when ≥ SGA (plus 3-month grace at first SGA month) Non-blind SGA 1, 550 * */mo; blind * *2,590
After EPE Cash terminates on a month of SGA; Expedited Reinstatement (EXR) is the safety net within 5 years
Extended Medicare Premium-free Part A continues ≥93 months after TWP ends if disability continues Separate from cash

The cash cliff is the step from full monthly DI benefit to zero once SGA persists past TWP/grace — not a gradual phase-down. Beneficiaries rationally treat SGA as a hard stop even when extended Medicare softens the health cliff for a time.

SSI path (Title XVI):

Medicaid Buy-In (state option): working people with disabilities pay premiums to keep Medicaid when earnings exceed ordinary ABD limits. KFF (2025/2026 surveys): 47 states offer a buy-in. National enrollment is stale: BPC (2022) cites ~193,000 enrollees across 35 states in 2011 as the last comprehensive public figure — a data gap for O4 scoring, not a claim that buy-ins are unused.

Confidence: strong on Red Book rule logic and 2024 parameters; strong on 1619(b) counts; moderate on current buy-in headcount (2011 last national).

2. Take-up (H4 first leg; KC3)

December 2023 (SSI ASR; Phase 0 lock):

Construct Count / share
1619(b) participants aged 18–64 108,825
Blind/disabled SSI recipients 18–64 4,148,143
1619(b) share 2.6%
Working blind/disabled (incl. 1619) 5.1% of blind/disabled

Ticket assignment rates in the Mathematica evaluation series and later GSA/OES summaries sit in the low single digits of the eligible pool (historical Phase-1 peaks ~1–1.4%; later program-wide summaries often ≤**~5%**). That is a usability signal for H4, not an employment effect size (those are §5).

KC3 remains unfired for 1619(b): denominators are published.

Confidence: strong.

3. BOND as cliff evidence (anchor 9)

Abt/Mathematica BOND Final Evaluation Report (Gubits et al., Oct 2018) and the Abt 2019 synthesis brief lock the following:

Result Stage 1 (national sample) Stage 2 (volunteers)
Average earnings impact of 1 − for2 offset Null (not statistically significant) Null
Average SSDI benefits due +$143/year (~+1%) over 5 years +450–500/year (~+4%) over 4 years
Share with earnings above BYA (annualized SGA) +7% relative to control +23% relative
Enhanced work-incentives counseling (EWIC) vs standard WIC Virtually no incremental earnings/benefit impacts
Benefit-cost (Stage 1) Net social cost for full caseload Offset+WIC net benefit in volunteer subpopulation; EWIC net loss

Reading for the cliff wager: replacing the cash step-function with a 1 − for2 ramp did not raise average earnings in the nationally representative sample; it raised average benefits paid (windfall to those already at SGA under current law dominated induced above-BYA entry). Information intensification alone fails in the same experiment (EWIC ≈ WIC). Anchor 9 holds: BOND found limited (null average) earnings response to the offset.

That null constrains architecture #3 (cliff redesign scored against BOND): a pure DI cash offset is not a free fiscal lunch and may not move mean earnings. It does not rescue motivational programs; counseling enhancements also failed.

Confidence: strong — primary SSA-commissioned RCT/demo.

4. Coverage-cliff instruments vs information-only (≥2 evaluations)

H4’s second leg needs cliff/offset or coverage-separation designs that move earnings more than information-only interventions:

  1. Washington Medicaid Buy-In (Healthcare for Workers with Disabilities) — Gettens et al., Journal of Disability Policy Studies (2012): matched comparison of recent conventional-Medicaid enrollees who entered MBI vs matched stayers. Findings: MBI associated with more work, higher earnings, and lower food-stamp reliance while preserving medical coverage. This is a health-coverage cliff redesign, not a Ticket mailer.
  2. National / multi-state MBI outcome series (Mathematica) — Gimm, Davis, Andrews, Ireys & Liu (“three E’s,” 2008) and related Liu & Weathers earnings notes: among MBI participants with wages, on the order of ~40% increased earnings after enrollment (median real increase on the order of ~$2,600 in the national synthesis cited in secondary reviews). Design is pre/post among enrollees, weaker than the WA match, but directionally consistent and independent of Ticket counseling.

Contrast set (information / service vouchers): BOND EWIC null (§4.3); Ticket ITT employment/exit nulls (§5). Coverage continuity instruments show earnings movement where counseling- and voucher-intensity do not.

Band: Pure DI cash offset (BOND) did not move average earnings. The binding “cliff” in the supported reading is therefore health-coverage loss and rule complexity, not merely the existence of any phase-down schedule — and not work aversion cured by motivational messaging.

Confidence: moderate-to-strong on the WA matched study; moderate on national MBI descriptive earnings gains.

5. H4 adjudication

H4 — The cliff, not work aversion, binds return to work.

Criterion Result
1619(b)/Ticket take-up low relative to working-age rolls Met. 1619(b) 2.6%; Ticket assignment low-single-digit % of eligibles.
≥2 evaluations find cliff or offset / coverage-separation designs move earnings more than information-only Met via WA MBI matched evaluation + Mathematica MBI earnings series, contrasted with BOND EWIC and Ticket info/service nulls. Pure DI cash offset (BOND) itself is null on mean earnings — logged as architecture constraint, not a veto of the take-up leg.

Verdict: SUPPORTED, with the BOND null carried into scorecard O4 cells for cash-offset architectures.

Refutation path closed: take-up is not high, so “high take-up + offset non-response” does not fire even though BOND earnings non-response is real.

Confidence: strong on take-up; moderate-to-strong on the comparative evaluation leg once cash-offset vs coverage-cliff are distinguished.

Implications

← All Disability research documents Sources digest Read the whitepaper