This study is the first to examine if mandatory pollution disclosure programs, exemplified by the Toxics Release Inventory (TRI) program, affect workers’ chemical exposure. Several authors (Porter, 1991; Ambec and Lanoie, 2008) have argued that external pressure can prompt plants to learn about their inputs and production processes, leading to source reduction. Indeed, according to plant managers, the most important benefit from the TRI program is their discovery of opportunities for source reduction as they prepare their inventories of inputs and production processes in order to estimate their TRI emissions (Lynn and Kartez, 1994; Kraft et al., 2011). Plants’ source reduction can potentially reduce both emissions into the environment and worker exposure to chemicals. As noted in Roelofs and Ellenbecker (2003), source reduction, motivated by environmental protection goals, “… have had a significant impact on … reducing workers’ hazardous chemical exposure in the work environment.”

Ultimately, the impact of TRI program on worker exposure is an empirical question. We examine actual personal air contaminant exposure data collected during 1,644 plant-level inspections between 1984 and 2009 for 29 states. We focus on the US chemical manufacturing sector, which ranks among the top sectors in air emissions reported to the TRI program (EPA, 1998) and violations of worker exposure limits to air contaminants (OSHA, 2011).

Our findings are two-fold. First, we find that worker exposure has declined at a time coinciding with the TRI program implementation, accounting for the time trend and other confounding factors such as plants’ inspection and compliance history, and state fixed effects. The counts of chemical test results that exceed the legal exposure limits per inspection declined by 31% from 1.02 to 0.70 per inspection in the post-TRI program period, while the maximum ratio of test results to the legal limits declined by 10% in the post-TRI program period, both estimates of the decline in worker exposure are statistically significant at the 10% level. These results present the first evidence for a decline in plant-level pollution-related measures at a time coinciding with the inception of the TRI program (Hamilton, 2005), as no other such data exist in the pre-TRI program period (Gray, pers. comm.). Our description of the decline in actual worker exposure, measured with personal sampling devices and recorded by the regulator, is an important innovation over studies that are forced to rely on plants’ self-reported emissions to describe plants’ responses to environmental policies (Gamper-Rabindran, 2006).

Second, while our results indicate that worker exposure declined at a time coinciding with the inception of the TRI program, we are not able to attribute this decline to the TRI program per se. To definitively attribute the decline to the TRI program we believe the evidence should indicate that worker exposure declined to a greater extent in plants that were affected to a greater degree by the program. However, we do not find statistically significant evidence of greater declines in worker exposure either in plants with greater emission reductions or in plants operating in industries with greater emission reductions.