Contreary et al 2017 — Employment Patterns Before Applying for Disability Insurance

Summary

Uses Survey of Income and Program Participation (SIPP) panels (1996, 2001, 2004) linked to Social Security Administration (SSA) administrative records to describe employment patterns of Disability Insurance (DI) applicants ages 25–55 in the 6–24 months before application. Introduces a Type 1/Type 2 taxonomy: Type 1 applicants ($\approx 53\%$) were consistently employed or ceased employment shortly before filing; Type 2 ($\approx 47\%$) had intermittent or no prior employment. Type 2 applicants have substantially lower allowance rates, higher public program participation, and are outside the reach of employer-focused early-intervention proposals.

Key Claims

• At the 6-month pre-application window: $28\%$ consistently employed, $25\%$ ceased employment, $14\%$ intermittent employment, $33\%$ no employment — roughly $53\%$ Type 1 and $47\%$ Type 2.
• Type 1 allowance rates: $\approx 47$–$50\%$; Type 2 allowance rates: $\approx 32$–$37\%$ — a gap of $10$–$15$ percentage points (pp) that persists across the 6-, 12-, 18-, and 24-month windows.
• Intermittent employment reduces allowance by approximately $10$ pp vs. consistently employed (regression-adjusted, statistically significant); ceased employment is statistically indistinguishable from consistent employment.
• Age 51–55 vs. ages 25–30: $+30$ pp allowance rate — by far the largest predictor, confirming vocational grid age-threshold effects.
• Workers' compensation (WC) receipt at the 6-month window: $-13$ pp allowance rate (regression-adjusted).
• Type 2 demographics: more female ($53$–$57\%$ vs. $\approx 47\%$ Type 1), more Black ($22$–$23\%$ vs. $\approx 13\%$ Type 1), lower education, higher Medicaid participation ($27$–$31\%$) and Supplemental Nutrition Assistance Program (SNAP) participation ($27$–$29\%$), lower average household income ($\$46{,}178$ vs. $\$62{,}986$ for Type 1).
• Employer-focused early-intervention and worker-retention proposals — aimed at preventing DI application among recently employed workers — miss $\approx 47\%$ of the applicant pool. Type 2 applicants are most reachable through the public programs (Medicaid, SNAP) they already participate in.

Concepts Introduced or Extended

• Conditional DI Applicants
• DI Application Costs and Take-Up
• SSA Sequential Determination Process
• Vocational Grid
• DI Denied Population

Entities Mentioned

• Kara Contreary
• Todd Honeycutt
• Michelle Stegman Bailey
• Joseph Mastrianni
• Social Security Administration

Quotes

"Proposals to divert DI applicants toward employment often target workers who have recently exited employment or are at risk of exiting employment.... [However,] approximately half of DI applicants had no employment or intermittent employment during the 24 months before their application."

My Take

The key contribution is the Type 1/Type 2 taxonomy, which reframes the policy debate about DI diversion. Earlier SSA proposals (Benefit Offset National Demonstration, employer early-intervention programs) implicitly assume applicants are recently workforce-attached — true for Type 1 ($\approx 53\%$) but not for Type 2 ($\approx 47\%$). The allowance rate differentials confirm the vocational grid's age effect dominates all other applicant-pool characteristics: age swamps employment recency as a predictor of award ($+30$ pp for 51–55 vs. 25–30). The $-13$ pp for WC receipt is notable and plausible given the vocational grid's treatment of claimants who have engaged in recent work activity. Design limitations include a relatively small sample ($N \approx 500$–$1{,}360$ depending on window length) restricted to ages 25–55, which excludes the large 55+ population where grid effects are most pronounced. The purely descriptive design precludes causal inference. The core policy conclusion — that roughly half the applicant pool requires a public-program-based rather than employer-based intervention — is well-supported by the data.