Tariffs and Carbon Emissions

ABSTRACT

International trade and the environment are uneasy partners. Both environmentalists and free traders worry that the pursuit of one goal may obstruct the other. Nowhere is this tension more evident than in the area of climate change. Does trade liberalization increase carbon emissions? Do efforts to reduce carbon emissions lead to protectionist pressures? This paper addresses these questions by examining the relationship between CO2 emissions and tariffs in 109 to 153 countries from 1988 to 2013. Using instrumental-variable regressions to address reciprocal causation, I find that emissions reductions led to higher tariffs on manufactured goods. This suggests that carbon-intensive industries responded to carbon restrictions by lobbying against trade liberalization. In contrast, emissions did not affect tariffs on less carbon-intensive primary products, and neither type of tariff affected CO2 emissions. My results imply that efforts to combat climate change may obstruct trade liberalization, but the latter should not hinder climate change mitigation.

KEYWORDS:

• Climate change
• environment
• trade

Acknowledgments

I thank the International Interactions editor, two anonymous referees, Tim McKeown and Gabriella Montinola for helpful comments, and Evan Sandlin for research assistance. Any errors or omissions are, of course, my own. Replication materials for this article can be found at http://dvn.iq.harvard.edu/dvn/dv/internationalinteractions.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Notes

¹ I emphasize industry pressures, as these are clearly salient to both trade and climate change politics. However, the argument could also incorporate citizen and environmental groups. For example, the Citizens’ Climate Lobby—a grassroots environmental movement—explicitly advocates border taxes along with carbon restrictions to prevent carbon leakage (https://citizensclimatelobby.org/laser-talks/border-tax-adjustment/?). Such civil-society pressures, if politically effective, would also create a negative relationship between carbon emissions and tariffs. It is also possible that mass public opinion matters, and may provide the main impetus for greenhouse gas restrictions. However, it is not clear whether and how public opinion would affect the link between such restrictions and tariffs. For example, empirical work by Bernauer and Nguyen (2015) shows that “greener” individuals are more supportive of trade liberalization, a set of preferences that could produce a positive relationship between emissions and tariffs.

² Data for CO₂ emissions are from the World Bank’s World Development Indicators. Data for other variables come from the same source, unless stated otherwise.

³ I employ import-weighted tariffs to ensure that the measured tariff rate is relevant, that is, imposed on goods and countries from which country i actually imports.

⁴ The following countries were coded zero before 1997 and one from 1997 onwards: Australia, Austria, Belgium, Bulgaria, Canada, Croatia, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Latvia, Lithuania, Luxembourg, Netherlands, New Zealand, Norway, Poland, Portugal, Romania, Russia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Ukraine, and the United Kingdom. All other countries were coded zero for all years. Belarus, Malta, and Turkey are also Annex 1 signatories but were coded as zero for all years because they do not have binding emissions targets.

⁵ Contiguity is measured using the Correlates of War database (http://cow.dss.ucdavis.edu/data-sets/direct-contiguity). States are coded as contiguous if they share a land border or are separated by 24 miles of water or less.

⁶ Data are from the U.S. Energy Information Administration (https://www.eia.gov/cfapps/ipdbproject/IEDIndex3.cfm?tid=3&pid=3&aid=6). I also tried neighboring oil reserves, but this instrument was weak, perhaps because contiguity matters less for oil than for gas, which requires pipelines.

⁷ Gas reserves—unlike gas production—should be unaffected by demand in country i, strengthening this instrument’s claim to exogeneity.

⁸ Treating EU members “as if” they were independent and including them in this analysis does not alter my results. For a discussion of this issue, see Weinberg (2016).

⁹ I log this variable for empirical rather than theoretical reasons: the logged variable is a much stronger instrument than the unlogged one. Neighbor tariffs are the simple average across contiguous neighbors.

¹⁰ I also tried including GDP per capita squared in case CO₂ emissions follow an environmental Kuznets curve. The quadratic term was insignificant and did not affect my results, so I omit it from the analyses presented here.

¹¹ As Roodman (2009) recommends, I limit the number of instruments by employing only two lags. Tests for serial correlation reveal none: p-values (H₀: no autocorrelation) range from .15 to .99, with an average across models of .45.

¹² Three of the remaining OECD members—Canada, Japan, and the United States—refused to ratify the Kyoto Protocol.

¹³ This concern applies less forcefully to the impact of emissions on tariffs because broad-based efforts to reduce emissions should affect all sectors, tradable and non-tradable alike.

¹⁴ Data on greenhouse gas policies are from Columbia University’s Sabin Center for Climate Change Law (http://web.law.columbia.edu/climate-change) and the London School of Economics’s Grantham Research Institute on Climate Change and the Environment (http://www.lse.ac.uk/GranthamInstitute/).

¹⁵ For the tariff outcomes, I estimated an endogenous treatment model using the previous instruments. The results for manufacturing tariffs were stronger than the ones presented here, but the primary-tariff model failed to converge. For the carbon policy outcomes, I estimated instrumental-variable probit models, but these also failed to converge.