The last twenty years have seen two major trends in firm environmental regulation. First, states increasingly encourage voluntary environmental management efforts as an alternative to command-and control regulation. As a result, the level and distribution of pollution – and the associated environmental risks – are increasingly a function of firms’ discretionary decisions. Second, there has been growing concern about equitable distribution of environmental risks. Research consistently shows that poor and minority communities are disproportionately exposed to environmental pollution, but has not conclusively identified the reasons for this disproportionate exposure. Scholars have proposed three hypotheses – first, that firms choose to locate in poor or minority communities; second, that neighborhoods transition to include more poor and minority members once a polluting facility locates nearby; and third, that state regulators implement and enforce environmental regulations in discriminatory ways. In empirical studies, however, none of these hypotheses consistently explains the inequitable distribution of pollution.

In this paper we explore a fourth possibility – that firms manage environmental risk at least in part based on neighborhood characteristics. We model firm environmental management decisions to identify the factors that best explain firms’ environmental management decisions, including the characteristics of the surrounding community. This inquiry has two important implications. First, if a firm’s environmental management decisions contribute to inequitable distribution of pollution, increased reliance on voluntary firm efforts may exacerbate environmental inequities. Second, if environmental inequities are related to the environmental management decisions of private firms, the movement of environmental decision making from the public sphere (command and control regulations) to the private sphere (voluntary efforts) may limit communities’ ability to influence these decisions or seek redress.

We measure firm environmental management decisions using the fate of toxicity weighted TRI chemicals from 1990 – 2007. Whereas previous studies have largely focused on volume of releases, this project uses a hazard score that more precisely reflects potential risk. We use a geographic information system to build a “neighborhood” profile for each TRI reporting facility, based on socioeconomic, race, and ethnicity census data. We also control for facility and firm level characteristics and state economic, employment, and regulatory characteristics.

Using a static model, we test the proposition that environmental performance in one year is related to community characteristics in the same year.   We also estimate a dynamic model that explains changes in firm-level environmental performance as a function of changes in neighborhood characteristics, by completing within-unit differencing on both the dependent and community characteristic variables. We estimate both models using two samples: one containing all TRI reporting facilities, and a second containing only facilities that are members of multi-facility firms. The larger sample maximizes external validity, but at the cost of threats to internal validity stemming from firm-specific differences. The smaller sample more effectively controls for endogeneity by matching on parent firms, but at the cost of limited generalizability. Estimating static and dynamic models for each sample should allow us to derive robust conclusions regarding the effect of neighborhood characteristics on facility environmental management decision.