ABSTRACT
Concerns about the effects and consequences of climate change have notably increased in recent decades. Despite large advances in the understanding of this phenomenon, further research into the determinants of gas emissions is necessary, to shed light on the responsibilities of producers and consumers, and their potential contribution to mitigation strategies. This paper studies the trajectories and determinants of carbon embodied in world trade during a period of 15 years. Our methodology relies on a multiregional input–output model, environmentally extended. Drawing on data from the World Input–Output Database, we estimate embodied emissions in bilateral flows. Then, we assess the determinants of CO2 emissions embodied in trade, combining input–output modelling with trade gravity panel data analysis. This paper offers a methodological approach that explains and quantifies the underlying factors of carbon trade, integrating the production and consumption perspectives and considering the geographical, structural and institutional context of countries.
Acknowledgements
We would like to thank the anonymous referees and Prof. Michael Lahr for his valuable feedback. This work has been partially supported by the Ministry of Science and Innovation of the Spanish Government (projects ECO2015-65582, ECO2016-74940 and ECO2016-75927-R) and the Department of Science, Technology and Universities of the Government of Aragon (Research Groups ‘Agrifood Economy, Globalization, Economic Development and Environment (19th–21st Century)’ and ‘Growth, Demand and Natural Resources’).
Disclosure statement
No potential conflict of interest was reported by the author(s).
ORCID
Rosa Duarte http://orcid.org/0000-0003-3113-1698
Vicente Pinilla https://orcid.org/0000-0003-2256-8898
Ana Serrano https://orcid.org/0000-0001-6991-7915
Notes
1 Note that WIOD CO2 emissions are disaggregated considering their energetic source, apart from the sector detail. Therefore, it is possible to examine the different energy mixes across countries and to evaluate their changes over time (see Genty et al., Citation2012). As an example, during the period 1995–2009 around 75% of direct CO2 emissions in the USA were generated by coal, gasoline and natural gas.
2 We calculate carbon embodied in trade for the 41 areas in the WIOTs, but in the econometric estimation we delete Taiwan and the region ‘Rest of the World’.
3 This procedure is used to adjust a matrix, with a minimum loss of information, to a required sum of columns and rows when positive and negative entries are present.
4 Note that when estimating using PPML, the dependent variable is defined in levels. But it is expressed in logarithms when using OLS as the estimation technique.
5 The larger importance of distance when measuring direct compared to embodied (direct and indirect) carbon trade is also shown in Table SI2, where the partial R2 is calculated using simple regressions. In this case, the partial R2 is 1.7 times larger when direct carbon trade is the dependent variable, which shows the robustness of our findings.
6 Note that, in this paper, RTAs do not consider environmental provisions.
7 See supplementary information for more detail on country classifications.