Raut 2017 — Exits from the Disability Insurance Rolls Estimates from a Competing-Risks Model

disability-insurancemortalityrecoverycompeting-risksactuarialdurationexits

Summary

Using a competing-risks framework, Raut estimates cumulative incidence functions for three simultaneous disability insurance (DI) exit causes — death, recovery, and conversion to retirement — for 1981–2000 entrants drawn from the Social Security Administration's (SSA) Continuous Work History Sample. Over a 30-year horizon, approximately 38% of new entrants exit via death, 8% via recovery, and 39% via Old-Age and Survivors Insurance (OASI) conversion at the Full Retirement Age (FRA), with about 15% remaining on the rolls. Exit probabilities differ sharply by age at entitlement, sex, and impairment type. Comparing the 1980s and 1990s entrant cohorts, death exit rates declined for all groups while recovery rates were largely stagnant — providing direct empirical evidence that DI mortality improvement does not translate into improved work capacity.

Key Claims

Method

Overall 30-Year Exit Probabilities (All Impairments, All Ages)

Exit cause 30-year cumulative probability
Death ~38%
Recovery ~8%
OASI conversion ~39%
Still on rolls ~15%

Death is the modal single exit cause for the overall beneficiary population, slightly trailing OASI conversion only because young entrants (who are far from FRA) weight toward death while older entrants dominate the conversion category.

Select-Period Exit Probabilities (Table 1: All Impairments, All Ages)

Duration year Annual death prob. Annual recovery prob.
Year 0 0.0565 0.0013
Year 1 0.1173 0.0084
Year 2–9 Declining Declining

Year 1 death probability (0.1173) exceeds year 0 (0.0565) because year 0 is truncated to part of a calendar year for most entrants and because the sickest entrants die early and are no longer in the risk set by year 1. The select period captures the severe right-skew in entrant health: the most critically ill exit in years 0–2 and survivors demonstrate relative resilience.

9-Year Cumulative Incidence by Age at Entitlement (Table 2)

Age at entitlement Recovery (9-yr) Death (9-yr)
20–30 16.3% 12.9%
31–40 8.3% 21.0%
41–50 3.8% 29.8%
51–55 1.4% 32.1%

Young entrants are recovery-dominant; older entrants are death-dominant. The age gradient is steep: a 20-to-30-year-old entrant is 11× more likely than a 51-to-55-year-old entrant to recover within 9 years, but 60% less likely to die.

9-Year Cumulative Incidence by Impairment Type (Table 4)

Impairment Age Recovery (9-yr) Death (9-yr)
Musculoskeletal 20–30 32.4% 4.4%
Musculoskeletal 51–55 5.7% 12.6%
Mental disorders 20–30 22.1% 2.8%
Mental disorders 51–55 4.9% 10.4%
Circulatory 20–30 15.6% 29.5%
Circulatory 51–55 2.8% 55.1%
Neoplasms 20–30 15.1% 69.8%
Neoplasms 51–55 1.4% 89.6%

Musculoskeletal and mental disorders have the highest recovery rates — and by far the lowest death rates. Neoplasms are death-dominant at all ages (70–90% cumulative death within 9 years). This table explains why the compositional shift toward musculoskeletal/mental conditions mechanically reduces measured beneficiary mortality without implying improved work capacity: low death rate ≠ high recovery rate.

Sex Asymmetry (Table 5)

Impairment Age Men recovery (9-yr) Women recovery (9-yr)
Musculoskeletal 20–30 38.5% 20.3%
Mental disorders 20–30 ~29% ~16%

Women have both lower death exit rates AND lower recovery exit rates than men with the same impairment and age at entitlement. This double-low pattern means women who enter the rolls spend longer average durations on them, contributing to the female enrollment share growth documented in DI Growth Decomposition.

Cohort Comparison: 1980s vs. 1990s (Table 6)

Exit cause Direction (1980s→1990s) Affected groups
Death Declined for all groups All ages, both sexes, all impairment types
Recovery Largely stagnant Most subgroups; modest improvement for ages 41–50 only

This is the central finding for the morbidity-mortality distinction: death exit rates declined systematically across all DI subgroups between the 1980s and 1990s, but recovery exit rates did not improve correspondingly. Declining mortality on the rolls is driven by medical advances and the compositional shift toward lower-mortality conditions — not by improved work capacity among beneficiaries.

Concepts Introduced or Extended

Entities Mentioned

Quotes

"In this study, I use a competing-risks framework to estimate the probability that a DI beneficiary exits the rolls because of death, recovery of ability to work, or conversion to retired-worker status."

"The results suggest that during the first year after the waiting period, the death probability is higher for older and male beneficiaries; during the first 9 years on the rolls, the recovery probability is higher for younger and male beneficiaries."

My Take

Raut's key contribution relative to the prior Office of the Chief Actuary (OCACT) actuarial studies (Zayatz 2011, 2015) is the competing-risks framework, which corrects the independence assumption in conventional Kaplan-Meier estimates. The substantive results align with OCACT benchmarks but the competing-risks approach is more internally consistent for estimating cause-specific exit probabilities. The most policy-relevant finding — stagnant recovery rates despite declining death rates across the 1980s→1990s comparison — directly challenges narratives that attribute DI roll growth to easier entry standards (if recovery were rising, the easy-entry narrative would predict longer rolls; if recovery were falling, it would predict easier exit; instead, recovery is flat while rolls grow). The sex asymmetry finding (women lower on both death AND recovery) is underappreciated in the decomposition literature.