Workers' Compensation and DI

Definition

Workers' compensation (WC) and Social Security Disability Insurance (DI) are the two largest social insurance programs for people with disabilities in the United States. Both serve workers unable to earn due to disabling conditions, but they differ in scope: WC covers only work-related injuries and illnesses with any severity, while DI covers total disability from any cause expected to last $12+$ months. Despite their overlapping populations, the programs are administered independently and interact through a legally mandated offset provision. Understanding their interaction — how workplace injuries feed DI enrollment, whether workers substitute between them, and how each shapes the other's incentives — is central to understanding the total social insurance architecture for disability in the US.

Key Ideas

WC paid benefits to $\approx 4$ million workers in 2008 at $\approx \$60$ billion/year (equally split cash/medical); DI served $\approx 7.4$ million disabled workers at $\$95$ billion in cash plus $\$54$ billion in Medicare.
WC cash benefits grew $+65\%$ (1987–2008); DI cash benefits grew $+403\%$ — divergent growth despite serving the same underlying population.
A lost-time workplace injury roughly doubles the 10-year probability of DI receipt (O'Leary et al. 2012).
WC replacement rates in practice fall well below policy targets, making DI a de facto backup insurer for the long-term consequences of workplace injuries.
The WC-DI offset caps combined benefits at $80\%$ of preinjury earnings, but only $\approx 11\%$ of DI beneficiaries are affected.
Aggregate inverse trend between WC and DI is not reproduced within states — individual-level substitution is not the mechanism.
Employers do not bear the full cost of workplace injuries; the DI program, funded by non-risk-adjusted payroll taxes, absorbs a share, weakening prevention incentives.

How It Works

The Occupational Injury Pathway into DI

O'Leary et al. (2012) provide the first matched-data estimate of this pathway using New Mexico (NM) Workers' Compensation Administration (WCA) records (1994–2000) linked to Social Security Administration (SSA) admin data and Internal Revenue Service (IRS) earnings ( $N=98{,}148$ ; followed through 2009). Medical-only cases ( $\leq 7$ days off work, no significant long-term impairment expected) serve as the counterfactual.

Main result: A lost-time injury doubles 10-year DI receipt probability ( $6\% \to 12\%$ ). This holds across all age groups.

The "10-year aging effect": Within each 10-year age group, lost-time cases match the DI hazard of medical-only cases in the next older group. A lost-time injury in the 25–34 cohort creates a DI trajectory that matches the medical-only trajectory for 35–44 year olds. Equivalently, a workplace injury adds $\approx 10$ years of disability risk.

15-year cumulative DI receipt by age group:

Age group	Medical-only	Lost-time
15–24	$3.0\%$	$4.9\%$
25–34	$5.5\%$	$10.1\%$
35–44	$10.9\%$	$20.0\%$
45–54	$20.3\%$	$34.4\%$

Severity gradient: Higher WC severity $\to$ higher DI probability. permanent total disability (PTD) $>$ permanent partial disability (PPD) $>$ temporary disability (TD) ( $\geq 8$ wk) $>$ TD ( $<8$ wk) $>$ medical-only. Crucially, even TD ( $<8$ wk) cases have substantially higher DI risk than medical-only — suggesting that even short absences have long-term consequences through career disruption, reduced competitiveness, or cumulative health effects.

Why the gap between WC PTD ( $0.5\%$ of lost-time cases) and DI ( $12\%$ over 10 years): WC disability classifications are made within 1–2 years of injury; DI enrollment often occurs years later as delayed health effects, changing labor market conditions, and cumulative functional limitations compound. Many cases settled as PPD via lump sum cannot be reopened; injured workers may not know they can request additional WC benefits if long-term losses exceed projections.

Covariate Gradients (Cox Proportional Hazards, Stratified by Age)

Earnings gradient (preinjury, 2007 dollars):

Earnings	Medical-only hazard ratio (HR)	Lost-time HR
$<\$10\text{K}$	$2.40$	$1.56$
$\$10$ – $\$20\text{K}$	$2.38$	$1.45$
$\$20$ – $\$30\text{K}$	$1.79$	$1.83$
$\$30$ – $\$40\text{K}$	$1.43$	$1.18$
$\$40$ – $\$50\text{K}$	$1.26$	$1.15$
$\$50\text{K}+$	$1.00$ (ref)	$1.00$ (ref)

Low earnings $\to$ much higher DI hazard, consistent with DI's progressive benefit formula (higher replacement rates for low earners) and lower job quality/career resilience.

Industry: Mining has the highest hazard ratio (medical-only: $1.42$ ; lost-time: $1.27$ ). Construction and health sectors elevated for medical-only. Most other industries not significantly different from government (reference) for lost-time cases.

Gender: Women have lower DI hazard (HR $\approx 0.82$ medical-only, $0.90$ lost-time vs. men), even controlling for earnings and industry.

The WC-DI Offset

Congress created the offset in the original 1956 DI program, rescinded it in 1958, and reestablished it in 1965. The offset limits combined WC and DI benefits to $80\%$ of the worker's pre-injury average earnings. In most states, SSA reduces DI to meet this cap; in 15 "reverse offset" states, WC reduces instead.

$\approx 11\%$ of DI beneficiaries (initial entitlement 2003–2004) also receive WC or public disability benefits (PDB).
Their DI is on average $6\%$ lower than for beneficiaries without WC/PDB ( $\$916$ vs. $\$983$ /month), because the offset only applies when the combined total would exceed $80\%$ .
$\approx 85\%$ of WC/PDB recipients have state WC benefits specifically.

DI as Backup Insurance

WC replacement standards call for two-thirds of pretax earnings for temporary disability and $55$ – $65\%$ for PPD (National Commission 1972; Council of State Governments 1974). In practice, studies consistently find WC replacing well under half of long-term lost earnings for TD and PPD cases (Boden and Galizzi 1999; Reville 1999; Reville et al. 2001). Delayed effects — injuries compounding over time, job loss years after the original settlement — are never captured by WC's front-loaded determination process.

DI fills part of this gap for the most severely affected workers, functioning as a backup insurer for the long-tail consequences that WC systematically underestimates and under-compensates. This is welfare-improving for injured workers but creates a moral hazard: employers do not bear the full cost of their safety choices, reducing prevention incentives. The DI program is funded by non-risk-adjusted payroll taxes, so the subsidy is diffuse and invisible.

The Fiscal Scale

From NM data: $\approx 7\%$ of all new DI awards attributable to WC lost-time injuries. Extrapolated nationally:

$\approx 70{,}000$ new DI awardees in 2010 attributable to WC injuries
First-year SSA cost: $\approx \$889$ million
Present value (PV) per cohort: $\approx \$12$ billion (at average starting age $47$ , expected $13.5$ years on rolls)
Adding Medicare approximately doubles the PV figure

The Substitution Debate

Aggregate evidence: Sengupta, Reno, and Burton (2011) document that WC and DI cash benefits per $\$100$ of covered wages moved as near-mirror opposites 1980–2007: WC rose then fell; DI fell then rose. Guo and Burton (2008) found that declines in WC statutory benefit levels and more restrictive WC eligibility were associated with increased DI applications 1985–1999, and that reduced employer safety incentives contributed to this shift.

Within-state evidence: McInerney and Simon (2012) test the substitution hypothesis directly using state-level panel variation. Their finding: the overall inverse relationship does not hold within states over time. The increases in DI occurred in states other than those with reductions in WC. The aggregate inverse trend reflects between-state compositional differences, not individual-level program substitution.

Implication: The divergent growth of WC ( $+65\%$ ) and DI ( $+403\%$ ) since 1987 is not primarily explained by injured workers switching from one program to the other. Instead, WC and DI serve parallel but partially non-overlapping populations, and their growth is driven by different structural forces (employer cost-containment and benefit cuts for WC; demographic change, benefit formula, and liberalized eligibility for DI).

Why It Matters

Program Architecture

DI and WC were not designed as a coordinated system. The offset provision is a patch; the programs have different eligibility criteria, waiting periods, benefit formulas, and administrative structures. The result is that an injured worker's benefit trajectory depends heavily on idiosyncratic state WC policy choices, the severity of injury, whether their claim is settled before long-term effects manifest, and whether they have sufficient work credits for DI. Coordinating these programs — including redesigning employer incentives to bear more of the long-term DI costs attributable to workplace injuries — could improve both prevention incentives and benefit adequacy.

Prevention Incentives

Employers pay WC premiums (often experience-rated), but not the DI costs of their injuries. Because DI costs are financed through flat payroll taxes and employers see them as a program for "all disabled workers," not as a consequence of specific workplace conditions, they have no incentive to prevent the delayed, long-term disability that ultimately reaches DI. The O'Leary et al. finding that even minor lost-time injuries (TD $<8$ wk) substantially raise DI hazard implies that employer prevention efforts that focus only on catastrophic injuries — the cases that WC classifies as PTD — miss most of the DI-cost-generating tail.

Research Limitations

The O'Leary et al. study is single-state (NM, 1994–2000), so external validity is uncertain. The medical-only-as-control assumption is imperfect (these workers may have pre-existing health differences). Education, preinjury health status, and post-injury labor market conditions are unobserved. The DI receipt measure is right-censored at 2009, missing long-term outcomes for the latest injury cohorts. The $\approx 7\%$ NM share extrapolated nationally is illustrative, not causal.

Disability Management and Post-Injury Employment Stability

Butler, Johnson, and Baldwin (1995) challenge the standard WC outcome metric — the first return-to-work rate — using longitudinal data from the Survey of Ontario Workers with Permanent Impairments ( $n=1{,}850$ ; injuries 1974–1987). While $85\%$ of injured workers initially returned to work, $40\%$ of those who returned were not employed due to injury effects at the 1990 interview date. Only about half of all injured workers achieved stable post-injury employment.

Their multinomial logit model identifies four mutually exclusive post-injury employment patterns: single absence/successful ( $39\%$ ), single absence/unsuccessful ( $29\%$ ), multiple absences/successful ( $21\%$ ), multiple absences/unsuccessful ( $11\%$ ). The dominant predictors of the best outcome vs. the worst are employer accommodations, not individual worker characteristics: reduced hours reduce the worst-pattern probability by $61\%$ , light-duty work by $55\%$ , modified equipment by $38\%$ .

The WC replacement rate discourages initial return to work, but has no significant effect on post-return stability — suggesting policy focused exclusively on reducing benefits misses the dominant driver of post-return failure.

Back injuries disproportionately produce multiple-absence patterns, especially for older, lower-skilled workers — the same population most vulnerable to eventual DI entry via the occupational injury pathway.

See Disability Management and Return to Work for full detail.

WC Moral Hazard: Ex Ante and Ex Post

Beyond the WC→DI pathway, WC insurance itself is subject to two forms of moral hazard that affect the composition of injuries reported. Bolduc, Fortin, Labrecque, and Lanoie (2002) test for both using a three-alternative panel logit kernel on 9,800 Quebec construction workers monthly (1977–86):

Ex ante: Higher WC wage replacement ratios reduce workers' prevention effort, raising true injury incidence.
Ex post: Workers falsely report or exaggerate claims, especially for difficult-to-diagnose injuries (back disorders, bursitis, spinal problems) where the insurer cannot easily verify the claim's validity.

The key empirical test: if WC benefits disproportionately increase the proportion of hard-to-diagnose relative to easy-to-diagnose injuries, this specifically implicates ex post fraud. Their finding: 1% increase in WC benefits → 0.13–0.4 pp increase in the proportion of difficult-to-diagnose accidents. WC elasticities for hard-to-diagnose injuries (2.62–4.01) consistently exceed those for easy-to-diagnose (0.71–2.20). The effect weakened after 1979 experience-rating reform, consistent with increased employer prevention effort.

See Workers' Compensation Moral Hazard for full detail.

Open Questions

Does the WC $\to$ DI pathway differ substantially across states with different WC benefit levels and eligibility rules?
How much of the "10-year aging effect" operates through health channels (injury compound effects) vs. labor market channels (job loss, career disruption) vs. behavioral channels (knowledge of DI program)?
What would experience-rating of DI payroll taxes by industry or employer do to prevention incentives and WC-DI overlap?
Is the individual-level substitution finding from McInerney and Simon (2012) robust to more recent periods when WC cuts have been more severe?
How does the Americans with Disabilities Act (ADA) (accommodation mandates) interact with the WC $\to$ DI pathway?