Causal Inference–Based Study of Key Contributors to Industrial Accidents

Abstract

This paper aims to study the causal relationships among several prevailing factors and the three main dimensions of the consequences of industrial accidents that involve hazardous materials. Relying on data of the Major Accident Reporting System for the period between 1979 and 2019, regression, matching and inverse probability weighting based techniques are employed to quantify the average treatment effect of many scenarios by using the standardized mean difference to assess the covariate balance. The obtained results highlight the important influence of explosions and contractors’ involvement on human casualties, major releases of hazardous materials on environmental damages, and other causes and explosions on material losses. Although human and organizational causes are involved in the three dimensions of consequences, the contribution of plant/equipment causes seems more important in the environmental and material contexts. Such results may provide a better general understanding of the causal structure in such a critical field or at least help initiate a different vision for taking advantage of the lessons learned from the occurrence of major industrial accidents.