GUBMENTPlain talk · policy frontier
Filings / Disability / Sources / §1 Findings: Baseline — rolls, awards,
GBMT-12 · Research record · No. 12

§1 Findings: Baseline — rolls, awards, and composition

disability/research/ws01-baseline.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 1 and 8; adjudicates H1. Programs: SSDI (Title II DI) primary; SSI cited only where concurrent caution requires it. Awards ≠ stock.

Search log (PRISMA-lite)

Field Record
Date 2026-08-11
Strings SSDI growth decomposition aging women insured; Liebman disability insurance JEP 2015; Pattison Waldron disability incidence; Trustees Report Figure V.C3 incidence; CRR DI rolls drop 2015–2019; Duggan Autor disability rolls skyrocketing; DI ASR 2023 beneficiaries awards
Sources searched SSA DI ASR 2023; SSA OACT Trustees Fig. V.C3 + Long-Range Disability Assumptions; SSA briefing paper bp2019-01; Liebman JEP 2015 (and NBER summary); CRR IB 23-18 (Liu/Quinby 2023); Duggan growth chapter (Wharton PDF); CRS R45419
Inclusion Primary SSA statistical / actuarial series; peer-reviewed or SSA-commissioned decompositions with transparent methods; independent demographic reconstructions that cite SSA admin denominators
Exclusion Unlabeled “disability rolls” charts mixing SSDI+SSI; advocacy fraud tallies; raw beneficiary counts offered as incidence

1. Stocks and awards (construct locks)

December 2023 stocks (DI ASR):

Construct Count
Disabled workers in current-pay 7,365,987
All DI disabled beneficiaries (workers + widow(er)s + adult children) 8,709,006
Disabled adult children 1,142,375
Disabled widow(er)s 200,644

SSI (do not sum with SSDI casually): blind and disabled 6,264,542; under-18 983,169; aged 18–64 SSI (all categories) 4,039,319. Concurrent aged 18–64 on both Social Security disability and SSI: 949,971.

Awards: 2023 disabled-worker awards 523,834 (DI ASR Chart 10 family). Primary diagnosis among awards: musculoskeletal 34.0%, neoplasms 13.6%, mental disorders 12.7% — awards ≠ worker stock mental share ≈ 28.6% (Phase 0 / anchor 5).

Age composition of the worker stock (Dec 2023): the modal band is 60–64 (~34% of men and women workers). That is prevalence geometry, not proof of a new incidence surge.

Confidence: strong — SSA DI ASR / SSI ASR primary tables.

2. Age-sex-adjusted incidence (H1 recent-decade leg)

Trustees Report Figure V.C3 — awards per thousand disability-exposed, age-sex-adjusted:

Year Age-sex-adjusted incidence
2010 (peak) 6.4
2019 4.0
2022–23 2.9
2024 3.3

Gross (unadjusted) rates sit higher after 2000 as boomers enter peak-disability ages, then fall with the same trough. The adjusted series is the right object for H1: it removes the age-sex composition channel that raw roll counts absorb.

OACT’s 2025 Long-Range Disability Assumptions narrate the same post-2010 steep decline, pandemic trough, and an ultimate intermediate assumption of 4.6/1,000 — still below the 2010 peak. Absolute disabled-worker stock peaked near 8.95M (2014) and fell to 7.37M (2023) even while the “explosion” framing still circulates.

Confidence: strong — primary Trustees data table.

3. Two-method growth decomposition

Method A — SSA actuarial / published agency decomposition lineage

SSA OACT publishes gross and age-sex-adjusted incidence as the standing decomposition of awards from population and insured-status growth (Trustees Fig. V.C3; Goss presentations summarized in SSA briefing paper bp2019-01). Pattison and Waldron (2013), as summarized by that briefing paper, further split insured-worker growth into working-age population growth vs. rising disability-insured rates (primarily women). The agency method’s load-bearing claim for H1 is not a single %-of-GDP share; it is that composition-adjusted incidence, not raw rolls, is the integrity object — and that series fell after 2010.

Method B — independent demographic reconstruction (Liebman 2015; CRR check)

Liebman (Journal of Economic Perspectives, 2015) decomposes DI receipt with an age-by-year stock model (awards = incidence × exposed; exits via death/recovery). For the period since the early 1990s:

Male age- and cycle-adjusted incidence is described as flat since the early 1990s; female incidence rose toward male levels. The 1980s are a different story: policy-driven incidence collapse then rebound dominate that decade’s swing (1980 amendments / 1984 amendments).

CRR IB 23-18 (Liu & Quinby, 2023) is the second independent leg for the post-peak decade: the 2010→2019 incidence drop (0.69% → 0.44% in their framing) is attributed mainly to the business cycle (−0.14 pp) and residual ALJ/stringency (−0.12 pp), with population aging adding slight upward pressure (+0.02 pp). That is not a fraud-explosion story; it is a tightening-plus-recovery story after the peak.

Band / conflict (Autor–Duggan lineage)

Autor and Duggan (and Duggan’s growth decomposition through ~2003) assign larger shares to replacement rates and medical-eligibility criteria for the mid-1980s→early-2000s window. That lineage is equal-quality for that window; it does not supply a recent-decade age-adjusted rise. Report the band: **policy/composition shares for 1984–~2003 are contested; post-early-1990s majority demography (Liebman) and post-2010 incidence decline (Trustees + CRR) are not.**

4. H1 adjudication

H1 — The "explosion" is mostly composition and demography, not a mystery surge in fraud.

Criterion Result
Decomposition attributes a majority of long-run growth to aging / women’s insured status / related composition Met for the post-early-1990s growth that dominates the circulating narrative (Liebman: ~⅔). Full 1980s–2010s pooled share is period-dependent; 1980s swing is incidence/policy.
Age-adjusted award rates flat or down in the most recent decade Met. 6.4 (2010) → 4.0 (2019) → 2.9 (2022–23) → 3.3 (2024).

Verdict: SUPPORTED. The circulating raw-roll “explosion” fails both the SSA actuarial lens and the independent Liebman reconstruction once awards are age-sex-adjusted. Fraud is not a measured driver in either decomposition. Residual dispute is about how large the 1984-era screening change was for an earlier window — not about a post-2010 age-adjusted surge.

Confidence: strong on the recent-decade incidence leg; moderate-to-strong on the majority-demography claim once periodized as Liebman requires.

Implications

← All Disability research documents Sources digest Read the whitepaper