Low and Pistaferri 2020 — Disability Insurance Theoretical Trade-Offs and Empirical Evidence

Summary

A survey article in Fiscal Studies 41(1): 129–164 synthesizing two decades of theoretical and empirical research on disability insurance (DI) program design. The paper reframes the central policy question around screening error types: type I errors (false rejections of genuinely disabled applicants) are the dominant efficiency problem, not type II errors (false acceptances), because only 58% of severely work-limited US adults receive DI while applicant pools are dominated by the genuinely impaired. The authors organize the design space along five policy dimensions and use the Low-Pistaferri (2015) life-cycle structural model to show that increasing benefit generosity is welfare-improving on net given the catastrophic, permanent nature of disability shocks. Published April 2019 (submitted), journal issue 2020.

Key Claims

1. Type I errors dominate. Only 58% of severely work-limited adults in the US receive DI. The false rejection rate (type I) is 37–54% among applicants; the false acceptance rate (type II) is only 18–28%. The US DI program is calibrated too stringently relative to its coverage goal.
2. Insurance-incentive trade-off as organizing framework. The central design challenge balances insurance value (covering people who need it) against moral hazard (creating labor supply disincentives). DI insures against permanent, catastrophic shocks for which self-insurance is infeasible — unlike unemployment insurance (UI), which covers transitory shocks where precautionary savings can substitute.
3. Five policy dimensions. The design space has five axes: (1) stringency of the medical test, (2) application process and labor market attachment requirements, (3) categorical eligibility structure (all-or-nothing vs. partial disability), (4) benefit generosity and progressivity, and (5) reassessment frequency and standards.
4. Welfare effect of increasing generosity. Under the Low-Pistaferri (2015) life-cycle model with state-dependent utility, increasing DI generosity is welfare-improving: the marginal dollar of benefits goes to severely disabled workers for whom consumption-disability complementarity makes the marginal utility of consumption high.
5. Labor supply literature synthesis. The causal labor supply literature (Maestas, Mullen, and Strand [MMS] 2013: −28 percentage points [pp]; French-Song 2014: −26 pp; Autor, Maestas, Mullen, and Strand [AMMS] 2015: −48 pp corrected) overstates the effective work disincentive because the counterfactual labor force participation of denied applicants is low. The mechanical labor supply loss scales to 12–17 pp when denied applicants' realistic baseline participation rates are applied.
6. Insurance value of DI. DI substantially reduces financial distress: DI allowance reduces bankruptcy by 31% and foreclosure by 34% within 3 years (Deshpande, Gross, and Su 2019). Food security and consumption fall sharply upon rejection. Insurance value is not captured by labor supply estimates alone.
7. Labor market attachment requirement (US uniqueness). The US is unique among Organisation for Economic Co-operation and Development (OECD) countries in requiring substantial prior labor market attachment (20 quarters of covered earnings for workers age 31+) before application is permitted. This creates a coverage gap for workers with irregular employment histories.
8. State-dependent utility and the Netherlands model. If disabled workers have higher marginal utility of consumption (disability-consumption complementarity), standard welfare analysis understates the insurance value of DI. The empirical evidence on state-dependent utility is mixed. The Netherlands — with partial disability benefits, mandatory rehabilitation, and employer co-responsibility — is presented as a model for reducing both type I and type II errors simultaneously.
9. Baily-Chetty applied to DI. Meyer and Mok (2019) apply the Baily-Chetty sufficient-statistics framework to DI: optimal benefit generosity is a function of the consumption-smoothing value (large) relative to the labor supply distortion (moderate). Under plausible parameters, this supports higher generosity than current US levels.
10. Reassessment as policy lever. Continuing Disability Reviews (CDRs) reduce type II errors (removing improper recipients) and preserve labor market attachment. The optimal CDR frequency trades off fixed medical review costs against the dynamic benefit of returning recovered workers to employment.

Concepts Introduced or Extended

• DI Application Costs and Take-Up — type I/II error taxonomy as organizing framework; five policy dimensions; application process and labor market attachment requirements
• Causal Effects of DI Receipt — synthesis of labor supply literature with realistic counterfactual labor force participation (LFP); insurance value dimension

Entities Mentioned

• Hamish Low
• Luigi Pistaferri
• Nicole Maestas
• Kathleen Mullen
• Alexander Strand
• Eric French
• Jae Song
• David Autor
• Manasi Deshpande
• Tal Gross

Quotes

"We argue that the most pressing concern is about Type I errors: the US seems to be missing out a substantial fraction of those who should be insured."

"DI insures against permanent and catastrophic shocks—unlike UI, which covers transitory shocks for which precautionary savings can substitute."

My Take

The paper's most important contribution is the type I/type II error reframing: it shifts the analytical presumption from "DI is too generous" (the standard efficiency critique) to "DI is too stringent" (the coverage gap critique). The 37–54% false rejection estimate is striking, though it depends on operationalizing "deserving" via the severely work-limited population — a broader standard than SSA's formal medical criteria. The five-dimensional policy taxonomy usefully organizes the disparate empirical literature. The Netherlands model is appealing in principle, but mandatory employer co-responsibility may not translate readily to the US institutional context of at-will employment and weaker labor-management coordination.