Chen 2014 — Rejection from the Disability Insurance Program and Dependency on Social Support

Summary

Chen (2014) uses Survey of Income and Program Participation (SIPP) 1990–2008 linked to Social Security Administration (SSA) administrative records to trace rejected Disability Insurance (DI) applicants' long-run reliance on social support programs. Using a difference-in-differences (DiD) event-study design around the DI filing date, she finds that rejection does not push applicants back to work: labor supply falls $23$ percentage points (pp) below pre-filing levels at $t+10$ and earnings remain roughly $\$8{,}800$ below trend. Instead, rejected applicants shift onto other public programs (Temporary Assistance for Needy Families (TANF), food stamps, Unemployment Insurance (UI), general assistance) at rates that eventually match approved beneficiaries', consistent with a welfare-shifting rather than work-restoration effect.

Key Claims

• At $t=0$ (filing year), rejected applicants are $25$ pp more likely to be on social support relative to healthy comparison workers; the gap narrows but remains statistically significant at $t+10$: $+12$ pp for ages 30–49, $+7$ pp for ages 50–65.
• In the long run ($t+7$ to $t+10$), rejected applicants' social-support dependency is statistically indistinguishable from that of approved beneficiaries — a difference of at most $7$ pp that fades to zero.
• Labor supply is $23$ pp lower and earnings approximately $\$8{,}800$ below pre-filing at $t+10$, ruling out work-restoration as the dominant post-rejection path.
• Cohort heterogeneity: 2000s filers show larger effects than 1990s filers (younger cohort: $13.8\%$ vs. $4.0\%$ social-support probability gap), consistent with a more severely impaired and lower-education applicant pool entering during the 2000s expansion.
• Estimates are upper bounds on causal effects due to health selection: rejected applicants have worse unobserved health than healthy comparison workers (DiD comparison group), so part of the social-support increase reflects health deterioration, not the denial decision per se.
• Sample: males only, ages 30–65 at filing; $N = 193{,}051$ person-year observations from SIPP 1990–2008 matched to SSA administrative records.

Concepts Introduced or Extended

• DI Denied Population
• Program Spillovers
• Disability Insurance Program

Entities Mentioned

• Susan Chen

Quotes

"The results indicate that rejection from the DI program does not restore labor force participation for rejected applicants. Instead, rejected applicants increasingly depend on social support programs over time."

My Take

The welfare-shifting finding is striking: program stringency displaces costs rather than eliminating them. The long-run convergence between rejected and accepted applicants' social-support use directly challenges the policy narrative that denial encourages return to work. However, the male-only sample, DiD design (rather than instrumental variables (IV)), and upper-bound caveat limit precise causal interpretation — the social-support increase partly reflects health deterioration common to both rejected and approved groups. The cohort heterogeneity finding (larger effects in 2000s) is a valuable secondary result that links to the broader expansion-era applicant pool literature.