About the Paper

Download Paper

The underreporting of occupational injuries and illnesses to worker protection agencies has become a topic of great concern to researchers and policymakers. Although numerous studies have quantified the prevalence of the phenomenon, which specific types of injuries and establishments are most susceptible to underreporting is poorly understood. As a consequence, regulators have very little capacity to “red flag” employers that are likely to underreport the most injuries. The paper begins to fill this gap in existing literature in four interrelated ways. First, the researcher develops a simple theoretical model of the relationship between regulatory intensity, injury type, and underreporting.

The model yields a number of concrete predictions about how the frequency of injuries, and mix of injury types, will respond to changes in the frequency and/or stringency of audits. Secondly, the researcher proposes a scheme for classifying different types of injuries by their relative “detectability.” Third, using a dataset comprised of granular audit data obtained from the Occupational Safety and Health Administration (OSHA) and the Mine Safety and Health Administration (MSHA), the researcher tests the model’s predictions regarding which types of injuries will be underreported the most across regimes and over time.

Finally, the researcher explores whether any observable, establishment-level covariates – such as the percentage of injuries contained in regulatory filings that are highly detectable – could be used, in a manner akin to the IRS, to identify likely violators. Overall, the results provide considerable grounds for optimism that mining injury data in this fashion could provide useful insights. Not only do the researcher's findings bear out most predictions of the model, but they also suggest that empirical algorithms could be devised, based exclusively on observable firm- and injury-level characteristics, to help labor regulators identify employers that hide workplace injuries.

Research Gaps

• First, the theoretical model proposed here could be tested in other settings to explore whether it is generalizable to other time periods and enforcement regimes. Secondly, the injury classification scheme proposed, including the categorization of various injuries as severe, traumatic, and/or hard-to-attribute, could potentially be refined in ways that would enhance its predictive value. Finally and most importantly, it remains to be seen whether red flags similar to those developed here – such as the respective percentages of severe or easy-to-attribute injuries reported to regulatory officials – could become tools of practical value to DOL regulators. (pages 50-51)

Citation

Morantz, A. (2014). Stanford Law School. Filing Not Found: Which Injuries Go Unreported to Worker Protection Agencies, and Why? Chief Evaluation Office, U.S. Department of Labor.

Download Paper

This study was part of the Department of Labor Scholars Program, and was produced outside of CEO’s standard research development process.