Bolduc Fortin Labrecque and Lanoie 2002 — Workers' Compensation Moral Hazard and the Composition of Workplace Injuries

Summary

This paper provides theoretical and empirical evidence that workers' compensation (WC) insurance affects not only the frequency but also the composition of reported workplace injuries. The authors distinguish between easy-to-diagnose injuries (fractures, contusions) and difficult-to-diagnose injuries (back injuries, bursitis, spinal disorders), and show that a $1\%$ increase in WC wage replacement raises the proportion of difficult-to-diagnose injuries by $0.13$–$0.4$ percentage points (pp) — consistent with both ex ante (reduced prevention) and ex post (false or exaggerated claims) moral hazard. Using a panel logit kernel (hybrid multinomial probit) on $9{,}800$ Quebec construction workers monthly from 1977–86, the paper confirms that WC benefit elasticities are systematically higher for hard-to-diagnose than easy-to-diagnose injuries across all specifications and subperiods.

Key Claims

• WC insurance affects the composition of reported injuries, not only total injury incidence — the defining finding of the paper.
• $1\%$ increase in WC benefits → $0.13$–$0.4$ pp increase in proportion of difficult-to-diagnose accidents ($0.4$ pp before 1979; $0.13$ pp post-1979 experience-rating reform).
• WC elasticity for difficult-to-diagnose injuries ($2.53$–$4.01$ before 1979; $0.93$–$2.62$ after) consistently exceeds that for easy-to-diagnose injuries ($1.15$–$2.17$ before 1979; $0.71$–$2.20$ after) across all four model specifications.
• Unemployment Insurance (UI) replacement ratio is negatively correlated with injury reporting — workers in Quebec's seasonal construction industry substitute WC for UI in winter (lay-off season).
• Winter seasonality (Q1: November–January) significantly predicts difficult-to-diagnose injury reporting, consistent with WC-as-UI-substitute hypothesis.
• 1979 experience-rating reform weakened moral hazard: WC elasticities fell markedly in the post-1979 subperiod, consistent with firms increasing safety investment when bearing more of the cost.
• Older workers and higher-skilled workers report fewer injuries of both types — consistent with risk-aversion and lower accident rates.

Concepts Introduced or Extended

• Workers' Compensation Moral Hazard

Entities Mentioned

• Denis Bolduc
• Bernard Fortin
• France Labrecque
• Paul Lanoie
• Richard J. Butler

Quotes

"a 1 percent increase in WC benefits would cause a 0.4 percentage point increase in the proportion of difficult-to-diagnose accidents, before 1979 (0.13 for 1979-86)"

"the WC insurance affects not only the incidence but also the composition of workplace injuries"

My Take

This paper's core methodological innovation is the composition test for ex post moral hazard: if WC benefits only reduced prevention effort (ex ante), we'd expect proportional increases in both injury types; a shift toward hard-to-diagnose cases specifically implicates fraudulent or exaggerated reporting. The results are internally consistent — the moral hazard signature weakens after 1979 experience-rating reform, which is exactly what the theory predicts if employer prevention incentives strengthened. The Quebec construction industry context is well-chosen: seasonal employment creates a strong structural motive for WC-UI substitution, physician choice by workers (unique to Quebec among North American jurisdictions at the time) amplifies ex post fraud opportunities, and the long panel (10 years, monthly) allows clean random effects estimation. Main limitation is external validity — the construction/seasonal/physician-choice features of Quebec may overstate the composition effect for broader labor markets.