Coe and Rutledge 2013 — How Does the Composition of Disability Insurance Applicants Change across Business Cycles

Summary

Using Health and Retirement Study (HRS) and Survey of Income and Program Participation (SIPP)–Social Security Administration (SSA) linked data across three economic periods (2001–03 recession, 2004–06 expansion, 2008–10 Great Recession (GR)), Coe and Rutledge show that recession-era Disability Insurance (DI) applicants are younger, better educated, higher earning, and more recently employed — the demographic signature of Conditional DI Applicants. Critically, observable compositional shifts explain less than $40\%$ of the rise in application rates and less than $25\%$ of the rise in allowance rates, and the award rate rose when it should have fallen — implying a structural change in both application behavior and SSA evaluation practices unexplained by standard economic and demographic predictors.

Key Claims

• Applicant composition shifts cyclically: GR applicants (vs. expansion applicants) are younger, better educated, higher income, and more likely to have held recent full-time employment — consistent with the conditional applicant profile of workers who prefer employment and apply only after job loss.
• Mixed health evidence: SIPP data show GR applicants were $5.1$ percentage points (pp) less likely to have had pre-recession work limitations ($p < 0.001$), consistent with healthier marginal applicants. HRS data show no statistically significant differences in activities of daily living / instrumental activities of daily living (ADL/IADL) limitations, Center for Epidemiologic Studies Depression (CESD) scale scores, or mobility measures between expansion and GR applicants.
• Blinder-Oaxaca decomposition (Fairlie 2005 nonlinear extension): Observable characteristics explain fewer than $40\%$ of the $0.7$ pp rise in application rates ($2.9\% \to 3.6\%$, expansion→GR). Observable characteristics explain fewer than $25\%$ of the $9.1$ pp rise in allowance rates ($40.5\% \to 49.6\%$). The majority of both increases is unexplained by observables.
• Award rate paradox: Given declining pre-recession work limitations (SIPP) and the severity of the GR unemployment shock, the award rate should have fallen. It rose from $1.2\%$ to $1.8\%$. This cannot be reconciled with a purely compositional explanation — it implies a structural shift in SSA evaluation behavior or policies during this period.
• No programmatic changes: The authors document no changes to SSA adjudication rules or benefit parameters during the 2000s that could explain the award rate increase, reinforcing the conclusion that the shift is behavioral and structural, not policy-driven.
• SSI applicants: Similar composition patterns hold for Supplemental Security Income (SSI) applicants, though sample sizes are smaller.

Concepts Introduced or Extended

• Conditional DI Applicants — directly confirms the profile: recession applicants are higher-income, better-educated, more recently employed; supports the hypothesis that conditional applicants are activated by recessions
• DI Countercyclicality — adds composition-level evidence and an award-rate paradox that complicates simple countercyclical narratives; suggests SSA evaluation behavior also shifts in ways unexplained by policy
• Blinder-Oaxaca Decomposition — applies Fairlie (2005) nonlinear extension to DI application and allowance rates; large unexplained residuals are the core contribution

Entities Mentioned

• Norma B. Coe
• Matthew S. Rutledge
• Health and Retirement Study
• Survey of Income and Program Participation
• Social Security Administration

Quotes

"The composition of applicants in the Great Recession looks more like the conditional DI applicant — healthier, better educated, recently employed — than those in prior recessions."

"The observable characteristics of applicants explain less than 40 percent of the change in application rates between the expansion and the Great Recession."

"The award rate increased during the Great Recession, despite the composition of applicants appearing healthier — a result that is at odds with a purely mechanical compositional story."

My Take

The paper makes two distinct contributions of unequal strength. The composition finding is well-documented: recession applicants fitting the conditional-applicant profile (younger, better educated, recently employed) is exactly what Autor and Duggan predicted and what Lindner et al. (2017) later confirm with far larger administrative data. The stronger claim — the award-rate paradox and its implication of structural behavioral change — is intriguing but underidentified. The Blinder-Oaxaca residual is by construction unexplained, and the paper cannot rule out that unobserved health differences (not captured by HRS/SIPP measures) or processing-time dynamics explain the award rise. Still, the cross-validation of composition findings across two independent datasets (HRS and SIPP) is methodologically valuable, and the award-rate anomaly is a real empirical puzzle that motivates further research.