Firm behavior and pollution in small geographies

Abstract

We consider the relationship between the location choices of potentially polluting firms and local income. Unlike previous research in the area of environmental justice, we distinguish between pollution potential and actual releases of toxic substances in the locality. We explore the relationship between the profit maximizing behavior of potentially polluting firms in their choice of both location and expenditures to influence the likelihood of toxic releases and their expected financial costs. We proxy the expenditures on prudential behavior by observing the co-localization of waste remediation activities. Evidence supports the conclusion that firms behave rationally in managing risk of toxic release, which may result in disparities in exposure to toxic releases faced by certain population groups.

Introduction

Analysis of industrial localization and concentration has been of considerable interest since, at least, Marshall in 1920 and spawned a broad literature. However, there is a narrower question of whether the localization and distribution of polluting activities in the production of tradeables is the result of firms’ strategic decisions based on local demographic characteristics. This more nuanced question falls within the strand of literature in the realm of environmental justice as well as the broader context of industrial organization.

From the perspective of environmental justice, the question is, “Are low income areas or areas with larger fractions of minorities disproportionately affected by potential health risk associated with exposure to toxic releases?” A concomitant question is then, “Are polluting firms more likely to locate in lower income or higher minority areas?” On the other hand, there is the alternative question as to whether or not households self-select locally or regionally by income due, in part, to a correlation between housing prices and local environmental quality? If so, the correlation is ex post. Or, is there some combination of these circumstances at work to create the circumstances the literature reports?

This paper primarily analyzes the first question while showing that this effect is not the result of changes in demographic characteristics. Much of the literature has focused on actual or observed localized toxic releases and the demographic characteristics of the surrounding areas. In our view, the issue has two dimensions. The first dimension is whether demographic characteristics – in particular, income – of the local area influences potentially polluting firms’ location choices. Then, importantly, we consider the second dimension as to whether the potentially polluting firms’ choices on prudential expenditures that result in different likelihoods of toxic release are influenced by local demographic characteristics. That is, even though the previous literature has found correlation between toxic releases and lower income levels in the surrounding areas, it does not necessarily mean that potentially polluting firms are more likely to locate in lower income areas, but rather that potentially polluting firms in those lower income areas are more likely to realize that pollution potential because they take fewer costly precautions in terms of waste management.

Therefore, in this paper, we take a different approach, one that has the potential to add a critical understanding of location decisions of potentially polluting firms and their decisions that influence the likelihood of toxic releases. That is, we don’t restrict the analysis to reported toxic releases, but we consider the universe of firms in industries that are represented in the Environmental Protection Agency’s Toxic Release Inventory (TRI), the potentially polluting firms,¹ and analyze the demographic characteristics of the local area where these firms are located or locate. More importantly, we use the location decisions of firms belonging to the waste management and remediation industry as a proxy for the demand for prudential expenditure arising from firms concerned about the liability of toxic release in terms of their expected profits.

To clarify the relationship between the location and toxic waste management choices of potentially polluting firms and local income levels, we first provide a theoretical framework. A potentially polluting firm seeking to maximize profits will be concerned about the liability of toxic releases and the threat such releases pose to its financial results. As the firms’ exposure to pollution-related financial risk increases with income (due to higher property values or higher probability of collective actions by residents and other businesses), a prudential response is to manage that risk by limiting the release of hazardous waste in the environment. A potential channel through which firms can reduce their toxic releases is by employing more waste management and remediation services. Two otherwise identical firms, one in a high income neighborhood and the other in a low income neighborhood, would be expected to demand different levels of waste management and remediation services, positively correlated to the surrounding incomes.

Like any firm choosing a location, a firm engaged in a potentially polluting activity needs to consider the array of attributes of any particular location in terms of their importance for profitability,² including the financial risk of release and the necessary costs of managing the likelihood of toxic release. Recognizing that the incentives faced by a representative firm result in the realization of a localized aggregation of similar (potentially polluting) firms, we derive some conditions that can lead to either a positive or negative (or both) relationship between local income and the total level of potentially polluting activities in a neighborhood.

We then investigate empirically the predictions of our theoretical framework by looking at the relationship between the location choices of potentially polluting firms, waste management and remediation firms, pollution hazards and local income levels. In this analysis, we consider the demand for remediation or waste management services as a demand that arises from firms concerned about the impact of toxic release on their expected profits. Although not perfectly correlated with localized firms’ demand for environmental quality, waste management/remediation is the only clearly identified industry involved in pollution mitigation in the regional non-tradeables sector for which entry and employment data are available.

By and large, the potentially polluting activities considered here result from industrial activities whose output is not dependent on the local market, i.e., production of tradeables. Firms in these activities are free to choose any location, subject to zoning restrictions. One might naturally think of household demand for localized environmental quality to be expressed collectively through the political process and reflected in a regulatory or statutory framework that restricts the nature, location and technologies of productive activities. By restricting our analysis to a single state, Texas, we control for an otherwise heterogeneous regulatory framework.³ There are few local environmental restrictions imposed in the State of Texas, beyond local zoning laws, and the state itself takes a relatively light hand to regulation. Thus, jurisdictions in the state are largely subject to a practically identical regulatory environment. Differences in local demand for waste management services must arise from the consumers of those environmental services, i.e., local firms responding to localized conditions, and we largely eliminate any localized version of the pollution haven hypothesis due to regulatory heterogeneity.

In addition to the analysis of potentially polluting industry localization and household income, we consider the probability of entry of a potential polluter in and across the given geographies. By also focusing on entry, we avoid the question of inter-jurisdictional population sorting that might occur in the years following a potentially polluting firm’s entry. Residential mobility and regional sorting by income due to the presence or absence of an environmental hazard, if they occur, would already be reflected in observed household incomes in the areas proximate or distant to the pre-existing industrial concentrations (demographic characteristics are given at the time of entry). We conduct a similar analysis of likelihood of entry of waste management/remediation firms into those geographies while controlling for the presence of potentially polluting firms. In both cases, we control for agglomeration economies that might serve to attract firms into existing industrial concentrations (Glaeser, Kallal, Scheinkman, Schliefer, 1992, Henderson, Kuncoro, Turner, 1995, Combes, 2000, Rosenthal, Strange, 2003).

Working at the census tract-level, we find that locations and entry probabilities of potentially polluting firms are positively correlated with local income over only the lower range of income. These firms’ demands for measures to reduce the likelihood of toxic release, as proxied by the presence and entry of waste management/remediation firms, show a similar, but amplified, pattern and are positively correlated with the presence of potentially polluting firms. We also find very persuasive evidence that the relative frequency of toxic release, i.e., the ratio of toxic releases to the number of potentially polluting firms, is negatively correlated with proximate household income. Taken together, our results lead us to conclude that the inverse U-shaped relationship between income and toxic release is, at least partially, the product of potentially polluting firms seeking to maximize their expected profits by balancing the financial risk associated with a toxic release and the costs of waste management and remediation services.

The firm-level profit-maximization approach used in this paper has been common to economists investigating firm siting choices in relation to environmental issues. For example, List and Co (2000) used a conditional logit model to investigate in which states multinational firms make investments, leveraging variation in state environmental policies to consider whether foreign direct investment is, in part, driven by environmental standards. Likewise, Keller and Levinson (2002) investigated the number of new foreign-owned plants as a function of abatement costs to determine the effect of environmental-related compliance costs on the location of foreign direct investments. List et al. (2003) analyzed plant relocation choices made by firms, and found that differences in environmental factors (in this case, air quality regulations) significantly alter location choices. In our paper, by restricting the analysis to a single state, Texas, we control for an otherwise heterogeneous regulatory framework and focus on non-regulatory environmental factors and how these factors affect firms’ siting and location choices.

Our empirical finding at small geographical scale is novel in the field of environmental economics and has clear application to the literature on environmental justice. The early environmental justice literature has primarily focused on the relationship between local income and pollution exposure (Arora, Cason, 1999, Brooks, Sethi, 1997). While these papers find some empirical evidence of an inverted U-shaped curve, the theoretical relationship between levels of undesirable localized emissions and regional or local income or the role of the remediation industry have not been investigated.

Understanding the mechanisms leading to the correlations between income, race and pollution is crucial to draw policy implications. Several papers in the environmental justice literature try to identify the relative importance of the various causal mechanisms. Banzhaf and Walsh (2008) and Depro et al. (2015) show that residential mobility and sorting by income can explain the observed correlations. On the other hand, Been and Gupta (1997) and Pastor et al. (2001) find that disproportionate siting by polluting firms seems to matter more than residential mobility. Wolverton (2009) contributes to this discussion by matching TRI firms’ location decisions with tract demographic characteristics at the time of siting for two MSAs in Texas.

Modelling entry decisions in all tracts in Texas and using more economic factors as control variables, De Silva et al. (2016) find evidence that polluting firms choose to locate disproportionately in poor or high minority areas. However, in their analysis, they consider only firms in the TRI which have actually reported a toxic release. While these results might well portray an effective reality, we find that limiting the analysis to TRI firms that have a release on record is overly narrow and may miss a useful, broader picture. Our paper contributes to the environmental justice literature by considering potentially polluting firms and exploring how their location and waste management decisions, proxied by the presence of waste management and remediation firms, can explain the disparities in pollution exposure in different household income localities.

The structure of the article is as follows: in the next section, we present our theoretical framework. We then develop our empirical approach. In the section that follows, we explain the patterns of entry and exit in the remediation industry. Lastly, we summarize and conclude.