Strand and Trenkamp 2015 — When Impairments Cause a Change in Occupation

Summary

Strand and Trenkamp (2015) study Disability Insurance (DI) claimants who were denied at Step 5 of the sequential determination process because they retained the capacity to perform some job in the national economy — the "own-occupation impairment" population. Using 37,110 claimants denied at Step 5 in 2005 who did not appeal and did not reapply (approximately 14% of all Step 5 denials), they document substantial post-denial earnings losses: employment fell 21.8 percentage points (pp) and median earnings fell to 77 cents on the dollar relative to pre-onset levels. Mental disorders produce the largest employment and earnings declines; sensory and certain cardiovascular/neurological impairments produce the smallest.

Key Claims

• The study population — Step 5 denials who did not appeal — is the administrative indicator of "own-occupation" impairment. Their impairments are severe enough to prevent past-occupation work (they passed Step 4) but allow some work in the national economy (denied at Step 5). Because they did not reapply, they had no ongoing Social Security Administration (SSA) involvement after denial.
• Employment fell from 85.0% (2000, before onset) to 63.2% (2008, after decision) — a 21.8 pp decline. Median earnings among positive earners fell from $19,875 to $17,171 (ratio 0.77). Maximum earnings over the full 2006–2011 period averaged $25,362 vs. $33,095 pre-onset (also ratio 0.77).
• 78.1% of the study group worked at some point during the post-denial period — comparable to recipients of vocational rehabilitation services, and far above the employment rate of allowed DI beneficiaries.
• Earnings declines are regressive across the pre-onset earnings distribution:

Earnings decile	Employment before	Employment after	Earnings ratio (medians)
1st	67.6%	48.6%	0.92 (medians rose slightly)
5th	97.0%	74.8%	0.74
9th	99.1%	73.3%	0.58
10th	99.5%	67.9%	0.52

The top decile fell from $87,123 to $45,374 median earnings (−48%); the employment drop was also steepest (−31.7 pp vs. the overall −21.8 pp). For the lowest decile, median conditional earnings actually rose slightly ($9,943 → $12,700), reflecting selection bias: workers with very low pre-onset earnings who had large losses would have qualified for DI (fallen below substantial gainful activity (SGA)) and been excluded from the sample.

• By education: college graduates earned 67% of pre-onset maximum earnings, vs. 77% for the full sample — mirroring the earnings-decile pattern.
• Mental disorders are associated with the largest employment and earnings declines of any diagnostic group. The top diagnoses by severity of employment decline: chronic liver disease (−30.1 pp), organic mental disorders (−26.8 pp), anxiety disorders (−25.7 pp), essential hypertension (−25.0 pp).
• Diagnoses with the smallest employment declines, hence most promising for vocational rehabilitation: blindness and low vision (−16.7 pp), carpal tunnel syndrome (−17.3 pp), chronic ischemic heart disease (−17.4 pp), late effects of cerebrovascular disease (−18.1 pp), fractures of upper limb (−18.5 pp).
• Back disorders (the most common diagnosis, n=9,197) produce exactly average outcomes: −22.0 pp employment, 0.78 median earnings ratio.
• The 37,110 sample is only 14% of 267,821 Step 5 denials; the majority were excluded because they appealed, reapplied, lived in prototype states (expedited step 5), or had prior claims. The sample represents the purest own-occupation population.
• Own-occupation impairment risk in the population may be large: Rank and Hirschl (2014) find more than twice as many household heads reported some impairment as a severe work impairment; Wixon and Strand (2013) show ~25% of DI claimants appear to have own-occupation impairments at evaluation.

Concepts Introduced or Extended

• DI Denied Population
• SSA Sequential Determination Process
• Vocational Grid
• Substantial Gainful Activity

Entities Mentioned

• Alexander Strand
• Brad Trenkamp
• Bernard Wixon
• Javier Meseguer

Quotes

"We find that about one-quarter to one-third of high earners were not employed 3 years after the initial determination. Among those who were employed, earnings decreased by one-third to one-half of their pre-onset levels. These results provide a benchmark for one aspect of the generosity of the DI program."

"Our sample population had return-to-employment rates that were similar to groups that had received vocational rehabilitation services."

My Take

A clean administrative-data descriptive paper, not a causal identification study. Its primary value is documenting the coverage gap: workers whose impairments cause genuine earnings losses but who do not meet the any-occupation standard. The 77-cent earnings ratio and 22 pp employment decline are the numbers worth preserving. The diagnosis-level breakdown is useful for vocational rehabilitation (VR) policy targeting: mental disorders are problematic (large employment declines even for this higher-work-capacity group) while sensory and selected cardiovascular/neurological impairments are more tractable. The selection bias caveat at lower deciles is real: those estimates understate losses for the full own-occupation population because the most-affected low earners qualified for DI and exit the sample. No causal identification; the pre/post comparison conflates the occupation change effect with business cycle and aging over 2000–2008.