Do random workplace inspections reduce worker injuries?

Occupational safety and health inspectorates have a challenging task. The number of workplaces far exceeds the capacity of the inspectors to visit each one. Most inspectorates manage the challenge by allocating resources to programmed interventions and responsive work arising because of serious workplace incidents or complaints. Random, unannounced inspections are often used as part of the targeting procedure but do random inspections actually reduce worker injuries?

A new study [ David I. Levine, Micheal W. Toffel, Matthew S. Johnson, “Randomized Government Safety Inspections Reduce Worker Injuries with No Detectable Job Loss, Science, Vol 336, 18 May 2012] using California data provides a well-researched answer. Michael Toffel, an environmental management expert (Harvard Business School), along with economists David Levine (University of California, Berkeley) and Matthew Johnson (Boston University) created 409 matched pairs of inspected and uninspected workplaces from Cal/OSHA data. Sectors were representative of the higher risk industries in the state and include construction ( general building, special trade contractors), wholesale trade (durable and non-durable goods), metal fabrication (doors, car parts, aerospace products), wood and lumber products, transportation and others. Firms included in the study had at least 10 employees and some had more than 500. The researchers looked at the records for up to four years before and after the year of inspection.

Their study found 9% fewer injuries following the inspection and 26% lower injury costs but no negative impacts on economic factors such as employment, total earnings, or company survival. Importantly, the random-inspection effects endured at statistically significant levels even four years after the year of the inspection.

The study certainly supports the idea that random workplace inspections reduce both minor and major injuries to workers and costs for employers without economic harm to the enterprise or reductions in the labour force.

All studies like this have limitations. There may be other factors peculiar to California that come into play in this study. The selected firms were single establishment firms in high-hazard industries in a specific region. The study does not look at the effects a percieved risk of inspection might have on uninspected firms. Random inspections within a targeted high-risk sector—particularly if well publicized and supported with actual inspections—may increase the perception of detection among all participants in the sector. This may increase the incentives towards improved attention to safety and health. Perhaps a future study will examine this.

The alternatives to random inspections in a sector include targeting firms based on injury rates, geography, severe incidents, and complaints. Each of these strategies has its advantages and disadvantages. Well-publicized blitzes announced in advance for regions, specific industries or equipment theoretically have an impact. Targeting firms with high rates of injury also makes sense but it is hard to know the true injury rate for small firms. On the other hand, firms outside the announced blitz domain and those with lower than average reported injuries may perceive a lower risk of detection of regulatory violations and, for a few, lower incentives to achieve and maintain safety and health in the workplace.

This may not be the last word on the issue but this research offers practical information for the consideration of policy makers and inspectorates.