Autor and Duggan 2006 — The Growth in the Social Security Disability Rolls A Fiscal Crisis Unfolding

Summary

Autor and Duggan synthesize 20 years of post-1984 disability insurance (DI) expansion into a three-cause framework: (1) the 1984 Reform Act's shift from medical to functional eligibility criteria, (2) a rising DI earnings replacement rate driven by the progressive benefit formula interacting with stagnant low-skill wages, and (3) growing female insured rates. They project this growth toward a steady-state enrollment of $7\%$ of nonelderly adults — a trajectory (subsequently not realized) that would have required roughly tripling the payroll tax share dedicated to DI. The paper also formalizes the "unemployability insurance" characterization: four of five determination steps evaluate labor market capacity rather than medical status, and DI application behavior tracks the unemployment rate closely.

Key Claims

• Three-cause framework: 1984 Reform Act liberalization + rising replacement rate + rising female labor force participation (LFP) jointly account for the DI expansion. Population aging and health trends are comparatively minor — aging explains at most $6\%$ of growth; population health improved during the same period (mortality among 50–64-year-olds fell $17$–$29\%$, 1981–2001).
• 1984 Reform Act effects: Medical-only awards fell from $82\%$ (pre-reform) to $58\%$ at initial determination and $\approx 40\%$ after appeal. Mental and musculoskeletal disorder awards rose $+323\%$ (1983–2003), reaching $52\%$ of all awards — a massive diagnostic shift toward subjective, chronic conditions with lower mortality and longer spell durations.
• Two-channel replacement rate mechanism:
  • Bracket effect: Primary insurance amount (PIA) bend points are indexed to the Social Security Administration (SSA) national average wage index (AWI). As low-skill wages stagnate below the AWI trend, a larger share of low earners' average indexed monthly earnings (AIME) falls into the $90\%$ and $32\%$ replacement brackets, mechanically raising the benefit-to-earnings ratio without any legislative change.
  • Earnings history effect: AIME itself is indexed to average wage growth via the AWI. A worker whose real earnings have stagnated sees their indexed AIME rise relative to current earnings — further inflating their replacement rate.
  • Net effect (Table 2): 10th-percentile male aged 50–61 cash replacement rate rose from $48\%$ (1984) to $64\%$ (2002); including Medicare and net fringe benefits foregone, from $68\%$ to $86\%$. 90th-percentile male: negligible change. Medicare expenditures per DI recipient doubled: $\$3{,}259 \to \$7{,}700$ (1979–2005, real dollars).
• Female LFP channel: Share of women insured for DI grew from $61\%$ (1984) to $76\%$ (2004), expanding the pool of covered women. This explains roughly $1/6$ of the $151\%$ rise in female DI receipt over the same period — the remaining $5/6$ requires behavioral and screening explanations.
• "Unemployability insurance" characterization: Four of the five SSA determination steps evaluate the claimant's ability to engage in substantial gainful activity given their age, education, and work experience — not medical status alone. DI application rates track the unemployment rate. The 1996 Drug Addiction and Alcoholism (DA&A) termination of $\approx 130{,}000$ beneficiaries is informative: approximately two-thirds subsequently requalified for DI under different impairments, consistent with DI functioning as long-term income support for displaced workers rather than a medically targeted program.
• Screening breakdown: The administrative law judge (ALJ) appeals process has grown increasingly favorable to claimants. Attorney representation of claimants rose from $37\%$ (1977) to $70\%$ (2000); the ALJ allowance share of all DI awards rose from $12\%$ (1979) to $27\%$ (2002); SSA lacks its own ALJ representation and loses approximately $75\%$ of contested appeals.
• Fiscal trajectory: DI spending as a share of total Social Security outlays rose from $10\%$ (1985) to $17\%$ (2005); the DI payroll tax increased from $1.0\%$ to $1.8\%$ of covered wages. Average new DI award present value (cash + Medicare) $= \$245{,}000$. Steady-state enrollment projection: $7\%$ of nonelderly adults — $71\%$ above the $4.1\%$ level in 2005. (This projection was not realized; enrollment peaked $\approx 2014$–$2015$ at $\approx 8.9$ million and then declined; see DI Growth Decomposition.)
• Reform proposals: (1) independent SSA medical evaluations to offset the asymmetric incentive from claimant-funded expert testimony; (2) SSA ALJ representation to counter attorney advocacy; (3) universal health insurance to remove the insurance motive that causes workers to use DI as a health-insurance route; (4) Veterans Affairs (VA)-style graduated disability scale ($0$–$100\%$) enabling partial benefits and preserving work incentives.

Concepts Introduced or Extended

• DI Replacement Rate — two-channel mechanism (bracket effect + earnings history effect); Table 2 figures (1984–2002); Medicare cost doubling
• DI Growth Decomposition — three-cause framework; diagnostic composition shift (Table 1); aging/health ruled out as primary drivers
• Conditional DI Applicants — "unemployability insurance" characterization; DA&A termination requalification evidence
• DI Application Costs and Take-Up — ALJ appeals statistics (attorney representation trend, ALJ award share trend, SSA appeals loss rate)
• Disability Insurance Program — legislative history (1956 founding → 1984 Reform Act); fiscal trajectory (1985–2005)

Entities Mentioned

• David Autor
• Mark Duggan
• Disability Insurance Program
• Social Security Administration

Quotes

"We calculate that the DI program is currently on a trajectory that, if unchanged, will move it from 4.1 percent of the non-elderly adult population in 2005 to roughly 7 percent of that population in coming decades." (p. 72)

"Although DI is formally a health insurance program, four of the five steps at which DI claims are evaluated concern the claimant's ability to perform work-like activities and to obtain employment rather than his or her medical condition per se." (p. 86)

"Of the 130,000 beneficiaries terminated from DI in 1997 due to drug addiction and alcoholism, approximately two-thirds subsequently returned to the DI rolls on the basis of other impairments." (p. 88)

My Take

This is Autor and Duggan's most accessible and synthetic treatment of DI growth, presenting the first unified three-cause framework and making the "unemployability insurance" analogy explicit. The paper's most durable contribution is its diagnostic composition data (the $+323\%$ mental/musculoskeletal shift) and its description of the ALJ appeals breakdown; both remain useful stylized facts. Its central limitation — shared with the 2003 paper — is treating the post-1984 trajectory as monotonically continuing toward a $7\%$ steady state, when in fact enrollment peaked $\approx 2014$–$2015$ and declined. The paper underweights the baby boom's natural demographic ceiling: enrollment would inevitably peak as boomers aged through peak disability years and converted to Old Age and Survivors Insurance (OASI). The "unemployability insurance" framing was validated ex post (labor demand recovery drove $73\%$ of the post-2010 application decline per Deshpande et al. 2025), but its fiscal doom scenario was not.