Autor Duggan and Gruber 2014 — Moral Hazard and Claims Deterrence in Private Disability Insurance

Summary

Autor, Duggan, and Gruber (2014) use proprietary administrative data from a large US private long-term disability (LTD) insurer (2000–2006) to estimate behavioral responses to three plan design parameters: the Elimination Period (waiting period), the Replacement Rate (benefit as share of pre-disability earnings), and the Maximum Monthly Benefit cap. The paper documents that both more generous replacement rates and shorter elimination periods substantially increase LTD claiming, and decomposes the elimination period effect into a mechanical censoring component ( $\approx 60\%$ ) and a genuine behavioral deterrence component ( $\approx 40\%$ ). The forward-looking deterrence mechanism is inconsistent with liquidity constraints but consistent with rational forward-looking moral hazard, as deterrence effects are income-invariant. These results support the Autor-Duggan (2003) hypothesis that marginal Social Security Disability Insurance (SSDI) enrollees have meaningful work capacity.

Key Claims

Replacement rate elasticity: A $10\%$ increase in the replacement rate raises LTD accessions by $\approx 40\%$ (elasticity $\approx 4.0$ ). Cap-adjusted (instrumenting away workers above the benefit cap) elasticity $\approx 1.4$ , still large relative to SSDI estimates.
Elimination period effect: Reducing the Elimination Period (EP) from $180$ to $90$ days approximately doubles LTD accessions. Roughly $\approx 60\%$ of this effect is mechanical censoring (short spells that would end before a 180-day EP never make it to the accession count), and $\approx 40\%$ is behavioral deterrence — workers with impairments shorter than the EP choose not to file.
Who is deterred: Deterred spells are the least severe — they return to work rather than transition to SSDI. Workers not deterred by a longer EP are disproportionately future SSDI recipients; $\approx 60\%$ of the replacement rate effect flows through future SSDI claimants.
Forward-looking moral hazard mechanism: Deterrence effects are income-invariant (do not vary with worker income), ruling out liquidity constraints as the primary mechanism and supporting forward-looking cost-benefit calculation by workers.
Longer EP → longer spells: A longer EP selects in more severe spells, raising average realized LTD spell duration conditional on accession.
LTD vs. SSDI comparison: LTD accession rate is $\approx 35\%$ lower than SSDI conditional on reported disability. LTD exit rate $\approx 12.7\%$ /quarter vs. SSDI exit rate $\approx 2\%$ /quarter — LTD beneficiaries return to work at roughly $\approx 6\times$ the rate of SSDI beneficiaries. $\approx 41\%$ of LTD spells co-occur with an SSDI award; the LTD plan offsets SSDI dollar-for-dollar.
Policy implication: Results support the Autor-Duggan (2010) proposal to use LTD as a transition-state screening mechanism before SSDI entry; marginal SSDI enrollees have high work capacity and respond to incentives.

Concepts Introduced or Extended

Entities Mentioned

Quotes

"Our results indicate that LTD beneficiaries are far more responsive to financial incentives than are SSDI beneficiaries."

"The income-invariance of deterrence effects is inconsistent with a liquidity constraints explanation and instead supports a forward-looking moral hazard mechanism."

My Take

The paper's main methodological contribution is the decomposition of the EP effect into censoring vs. deterrence using a structural spell-duration model, which is non-trivial and requires assumptions about the shape of the disability duration distribution. The income-invariance test for liquidity constraints is clean and persuasive. The data are unusually rich (proprietary insurer records) but the sample is limited to large employers offering list-billed LTD — generalizability to individual-market LTD or SSDI is not guaranteed. The $\approx 41\%$ SSDI overlap finding is striking and has direct policy relevance for integrated disability insurance (DI) reform proposals.