Pollution havens and the trade in toxic chemicals: Evidence from U.S. trade flows

Abstract

National registries of toxic chemical emissions and facilities are increasingly used to raise public awareness of potential health hazards in local areas, but an unintended consequence may be the offshoring of production to less regulated countries. Using disaggregated U.S. trade data, this study examines the impact of registry listing on subsequent bilateral trade flows. Estimates from a difference-in-differences model indicate a significant shift toward imports from poorer countries following registry listing. Assuming that environmental protection is a normal good, this result suggests the emergence of pollution havens due to more stringent U.S. environmental regulation.

Introduction

Do changes in environmental regulation affect where pollution-intensive goods are produced? Differences in environmental policy coupled with trade liberalization may lead to the emergence of pollution havens, with polluting activity relocating to areas with less stringent regulation (Jaffe et al., 1995, Copeland and Taylor, 2004, Brunnermeier and Levinson, 2004, Taylor, 2004). Less clear is whether changes to environmental regulation itself have a similar impact on trade patterns, with production moving off-shore in response to mandated restrictions, heightened scrutiny, and financial considerations.

This paper examines American trade in toxic chemicals, specifically those designated as toxic by the U.S. Environmental Protection Agency's Toxics Release Inventory (TRI) program. Following the 1984 industrial accident at a Union Carbide plant in Bhopal, India that killed nearly four thousand people within days and poisoned an estimated half a million in the following years, the United States Congress passed the Emergency Planning and Community Right-to-Know Act (EPCRA) two years later (Broughton, 2005). The act required domestic industrial facilities to report to TRI the quantity of releases and transfers of certain toxic chemicals. These data are made available to the public under the premise that this information creates incentives for companies to improve their chemical management and reduce toxic releases. TRI data collection began in 1988 with 332 chemicals listed as toxic and has increased coverage to the current 683 chemicals and chemical categories.¹

This study examines whether the implementation of the toxic chemicals registry program affects trade flows, and if so, how. While the intuition behind this question is straightforward in that increased domestic regulation in a globalized market can lead to increased imports as substitutes for domestic output, identifying a causal relationship from actual trade patterns is much less so. This owes to difficulties in mapping between regulation and economic activity, which are usually measured differently, and isolating the regulatory effect from other confounding factors like location- or time-specific trends.

Besides seeing whether trade patterns change, there is the narrower issue of whether toxic chemical imports are disproportionately sourced from less regulated jurisdictions, a phenomenon commonly known as the pollution haven hypothesis. This issue is controversial in part because it is not obvious that poorer countries, given their factor endowments, would have a comparative advantage in producing capital intensive goods like chemicals despite the common perception that such countries are more likely to host dirty industries given lax environmental standards. Furthermore, environmental regulation may induce technological improvements to domestic manufacturers and allow them to produce more efficiently, thus mitigating adverse impacts from increased costs and scrutiny.

To analyze the trade of TRI chemicals over the past two decades, this paper uses bilateral trade data derived from records collected by U.S. Customs and Border Protection and processed by the Foreign Trade Division of the U.S. Census Bureau. These data are disaggregated at the Harmonized Schedule ten-digit level and comprise all trade in chemicals between the years 1989 and 2006 for 180 trade partners. The highly detailed nature of these data provide a methodological advantage in that, unlike existing studies of trans-national pollution havens, economy activity is observed at the commodity level instead of industry or subsector and thus corresponds to the individual chemicals listed on the TRI. This mitigates concerns regarding the composition of an industry's output, some of which may be pollution-intensive and subject to regulation while others are not.

Commodity disaggregation also allows comparison of the listed chemicals to those not subject to regulatory change, which can be used to control for pre-existing trends and comparability between chemicals. The analysis uses a difference-in-differences least squares regression model to identify an average treatment effect on imports before and after a chemical is listed on TRI, with the identification strategy further sharpened by variation across a panel of chemicals and different years of registry listing. Differences among trade partners, such as distance, trade barriers, and regulatory stringency, are accounted for by direct measures of shipping costs, paid duties, and per capita income, respectively.

The results from the regression model indicate that overall imports of chemicals listed on TRI do not significantly change compared to all other chemical imports; however, they are disproportionately sourced from poorer countries after listing. At the same time, I find a statistically significant fall in exports of listed chemicals, which may proxy for domestic output, thus suggesting that the TRI program may increase foreign production among developing economies at the expense of domestic manufacturers. In other words, this points to the creation of pollution havens from environmental regulation.

The remainder of the paper is outlined as follows. Section 2 reviews existing research relating to trade and environmental regulation as well as the historical background of the TRI program. Section 3 describes the hypotheses, data, and empirical framework used for analysis. Section 4 presents the results, and Section 5 concludes with a discussion of the findings.