Development and Validation of a Firm-Level Occupational Safety Risk Index from Public Regulatory Data

Occupational injuries impose enormous human and economic costs, yet researchers, investors, and regulators still lack a scalable, transparent way to answer a basic risk question: how dangerous is a given firm for the people who work there? Existing firm-level evidence relies on restricted government microdata or on firms' self-disclosed sustainability metrics. We develop and validate the Occupational Safety Risk Index (OSRI), an open-source, fully replicable composite measure built entirely from public regulatory data: OSHA’s establishment-level Injury Tracking Application (2016–2025) and mandatory Severe Injury Reports (2015–2025), linked to SEC registrants through a documented three-pass matching protocol with manual adjudication. OSRI aggregates four hazard components—injury frequency, days-away severity, case gravity, and severe-injury intensity—for 1,756 U.S. public firms (10,720 firm-years). The index passes a comprehensive validation battery: convergent validity across two independent reporting systems, known-groups validity across industries, discriminant validity against a separate reporting-opacity construct, test-retest reliability of 0.63, and strong predictive validity (ρ = 0.61 for next-year injury rates). Externally, firms with higher OSRI trade at significant within-industry valuation discounts and subsequently face higher borrowing costs, despite higher contemporaneous accounting profitability—a short-term margin, long-term penalty pattern. We release the index, crosswalk, and code for public use.