Stapleton and Martin 2012 — Vocational Rehabilitation on the Road to Social Security Disability Longitudinal Statistics from Matched Administrative Data

vocational-rehabilitationdisability-insuranceSSDIearly-interventionlongitudinal-statisticsadministrative-dataSSImatched-datainstitutional-incentivesssd-entryVR-services

Summary

Stapleton and Martin (2012) use matched Rehabilitation Services Administration (RSA) 911 vocational rehabilitation (VR) closure records (1998–2009) linked to the Social Security Administration's (SSA) Ticket Research File (TRF), Master Earnings File (MEF), and Numident to document how many VR applicants enter Social Security Disability (SSD) in the 60 months following application — the first study to do so. The 2002 cohort (N=480,566) shows 17.8% entered SSD within 60 months, with Supplemental Security Income (SSI) recipients at VR application being the single strongest predictor (94% SSD entry). The paper shows that institutional incentives in the SSA cost-reimbursement payment system may perversely encourage VR agencies to accelerate, rather than prevent, SSD entry. The causal effect of VR service delivery on SSD entry is theoretically ambiguous and empirically unidentified.

Key Claims

Concepts Introduced or Extended

Entities Mentioned

Quotes

"VR services might accelerate the client's entry into SSD, perhaps by helping the client understand SSD rules and obtain a job that does not represent SGA."

"The state has a financial incentive to help the client obtain Medicare. The state pays a share of Medicaid expenditures, whereas Medicare is funded entirely by the federal government."

My Take

The paper's core descriptive finding — ~1 in 6 VR applicants enters SSD within 5 years, with SSI recipients at near-certain risk — is policy-relevant and actionable. The institutional incentive analysis is particularly insightful: the SSA cost-reimbursement structure and state Medicaid-to-Medicare fiscal incentives create a situation where VR agencies rationally accelerate SSD entry rather than preventing it. This is a systemic design flaw, not a behavioral failure. The causal question (do VR services cause more or less SSD entry?) remains open; the proposed instrumental variable (IV) (state-month duration percentile) is creative but the authors correctly acknowledge business cycle confounding as a serious threat. As a descriptive study, it is highly reliable — the statistics are population-level administrative data counts, not estimates. The main limitation is the absence of a comparison group to assess counterfactual SSD entry absent VR services.