Lindner (2013) uses Survey of Income and Program Participation (SIPP) data 1990–2004 matched to Social Security Administration (SSA) administrative records to document the employment transitions that precede disability insurance (DI) applications. Most future applicants are working a year before applying but not working at the time of application; job loss explains nearly all of the income decline preceding application. Although health shocks (illness/injury, work limitation onset) trigger most job separations, applicants typically wait several months before applying — and a meaningful share engage in job search and claim unemployment insurance (UI) benefits in the interim. Classifying non-employment spells by their beginning event reveals sharp heterogeneity: illness/injury applicants apply quickly and bypass alternatives, while job-loss applicants wait the longest and most resemble workers genuinely searching for re-employment.
Employment decline is gradual then steep: Employment rate among future applicants runs at ~80% thirty months before application, falls to ~60% at ten months out, then plunges to ~30% by the application month. Two-thirds of the employment decline occurs in the final ten months.
Job loss explains the income decline: A decomposition of monthly income changes shows that changes in employment (job loss) track the overall pre-application income decline almost perfectly; changes in wages and non-employment income are negligible. DI application is primarily a response to employment exits, not to earnings cuts while working.
Applicants do not apply immediately after job loss: DI application hazard peaks at 22.4% in month 2 of non-employment but remains above 10% through month 12. Over 30% of applicants are out of work 6+ months before applying.
Health shocks dominate beginning events: Using SIPP 1996–2004, ~49% of non-employment spells begin with an illness or injury; ~19% begin with work limitation onset (narrow def). Together, health shocks account for the majority of beginning events. Job loss (layoff/laid off/discharged), quit, and retirement together account for ~30–40% as an upper bound.
Sharp heterogeneity in spell characteristics by beginning event (reference category = job loss):
| Beginning event | Spell duration | UI participation | Job search | Notes |
|---|---|---|---|---|
| Illness/injury | ~4.8 months (−57%) | 11% | Very low | Apply fastest, short waiting time |
| Work limitation | ~6.2 months (−32%) | 32% | Moderate | Intermediate profile |
| Job loss (ref) | ~9.2 months | 71% | High | Longest wait, most alternatives considered |
| Quit | ~7.9 months | 47% | Low | Not significantly different from job loss in duration |
Illness/injury applicants have shorter processing times: Waiting time for application determination is ~25% shorter for illness/injury than for job loss applicants, consistent with clearer medical eligibility. Application success rate is numerically higher (69% vs. 60% for job loss) but not statistically significant.
Work limitation wording matters: The SIPP's broad work limitation question ("limits the kind or amount of work") produces a much less dramatic employment drop at work limitation onset (~10 percentage points [pp]) compared to the Health and Retirement Study's (HRS) restrictive definition ("prevents work altogether," ~45 pp drop). Findings on employment transitions are sensitive to survey instrument design.
Policy implications of heterogeneity: Illness/injury applicants (~49% of SIPP 1996–2004 spells) apply quickly regardless of alternatives — pre-emptive re-employment policy is unlikely to reach them. The remaining ~51% (work limitation, job loss, quit) demonstrate prolonged non-employment with job search and UI use before applying — this group is the plausible target for re-employment support, short-term income transfers, and work accommodation programs.
"Most applicants for DI stop working before they apply for DI benefits. These employment transitions account for most of the observed income decline prior to application."
"Applicants generally do not apply immediately after job loss and that those not directly affected by an illness or injury tend to consider other options before applying for benefits. Social policies aimed at increasing re-employment opportunities for such individuals could therefore be effective in reducing application rates."
The paper's main contribution is establishing the beginning-event typology and the heterogeneity in spell characteristics — a clean descriptive result using well-matched administrative data that has direct policy relevance. The finding that ~49% of applicants begin non-employment with an acute illness or injury (implying limited responsiveness to re-employment policy) and ~51% do not is a useful empirical anchor for calibrating the scope of early-intervention programs.
The descriptive regression in Table 5 is carefully executed but limited: it controls for observable demographics, not for unobserved selection into beginning-event categories. Workers who quit rather than get laid off may differ systematically from the job-loss group on motivation, health trajectory, or employer relationships, and the regression does not resolve this. The key result — illness/injury applicants apply faster and use fewer alternatives — is compelling on its face but does not allow causal statements about whether removing one beginning event type would change application patterns.
The paper predates the Mueller-Rothstein-von Wachter (2016) null result on UI-DI interaction and the Contreary et al. (2017) Type 1/Type 2 applicant classification. Contreary et al. reach a similar two-type conclusion from a different angle (pre-application employment history rather than beginning events), providing convergent validation.