Convergence in pollution terms of trade

ABSTRACT

To construct pollution terms of trade (PTT) on the basis of CO₂ emissions, we implement the world input–output tables for 40 countries by 35 industries to account for intermediate trade. We examine whether the PTTs have converged among the 40 countries between 1995 and 2009. The empirical evidence supports PTT convergence; PTT growth is negatively related to its initial level and this empirical result is robust to various control variables.

KEYWORDS:

• World input–output table
• international trade
• pollution haven hypothesis
• pollution terms of trade

JEL CLASSIFICATIONS:

• F18
• O13
• Q56

Acknowledgements

We thank the Editor Charles Van Marrewijk and two anonymous referees for constructive comments. We also thank Fumihiro Yamane, Keisaku Higashida and other participants at the Japan Economic Association Conference for their helpful comments on earlier versions of this paper circulated as ‘Testing the pollution haven effect: Accounting for international intermediate goods in trade’ and ‘Convergence in trade-associated pollution.’ Both authors are grateful to the financial support from JSPS KAKENHI Grant Number 25380346. Honma also acknowledges the financial support from JSPS KAKENHI Grant Number 16K03662 and Yoshida acknowledges the financial support from Research Grant of Faculty of Economics, Shiga University.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 We owe this argument to an anonymous referee for pointing this out to us.

2 The PTT is advocated in Antweiler (1996), whereas Muradian, O’Connor, and Martinez-Alier (2002) advocate the BEET. The BEET is the net of pollution emissions embedded in international trade, i.e. pollution incurred in the home country for foreign demand subtracted by pollution incurred in foreign countries for the home country's demand. A positive- (negative-) signed BEET for a country indicates a carbon surplus (deficit) for international trade in that country.

3 Intermediate trade is fully incorporated in the extended model in the next section.

4 Relatedly, the environmental Kuznets curve (EKC) hypothesis asserts that emissions per capita increase with increasing income per capita in the early stages of economic development before decreasing with increasing income per capita during the later stages of economic development.

5 Input–output analysis researchers have employed the multiregional input–output (MRIO) model to assess environmental impacts with intermediate trade (the MRIO is well surveyed by Wiedmann et al. 2007, 2011; Wiedmann 2009). The MRIO is applied to China (Pan, Phillips, and Chen 2008; He and Fu 2014), Japan–US (Ackerman, Ishikawa, and Suga 2007), Norway (Peters and Hertwich 2006), Spain (Serrano and Dietzenbacher 2010), the United Kingdom (McGregor, Swales, and Turner 2008; Druckman and Jackson 2009), and multiple countries (Peters et al. 2011; Nakano et al. 2009; Douglas and Nishioka 2012; Grether and Mathys 2013). In all the above studies except Grether and Mathys (2013) and He and Fu (2014), CO₂ emissions embodied in trade are calculated; however, none of the studies examined the PTT. Grether and Mathys (2013) analyze PTT of SO₂ for multiple countries, whereas He and Fu (2014) analyze PTT of CO₂ but only for China.

6 $\tilde{T}$ can also be calculated by using $\tilde{T}=Q-{\tilde{Q}}_D$, where ${\tilde{Q}}_D=\left(1-B\right)C$ is the implicit demand matrix (Johnson and Noguera 2012).

7 Reversal of emission intensities among countries and years is uncommon in our dataset. In other words, it is rarely observed that an industry k in year t is dirty in country i and clean in country j in year s.

8 Different from the growth rate of emission per capita that is taken as the explained variable in Ordás Criado, Valente, and Stengos (2011), the PTT depends not only on the home country's environmental regulations but also on those of the foreign country. Hence, we do not denote the negative relationship described in equation (12)(13) $\left(1/T\right)\log \left(PT{T}_{i,t}/PT{T}_{i,t-T}\right)=\alpha +\beta \log PT{T}_{i,t-T}+\mathit{\gamma }{\mathbf{z}}_{\mathrm{i},\mathrm{t}}+{\delta }_i+{\eta }_t+{\epsilon }_{i,t},$(13) as the defensive effect.

9 Data are publicly available on the website with the following URL: http://www.wiod.org/new_site/data.htm. Carbon emissions in the WIOD are computed based on the use of different energy sources in each industry. Therefore, emission intensities D_is in equation (1) are time-varying due to shifts in the use of different energy sources. Energy sources are the following: anthracite; lignite; coke oven and gas coke; gas and diesel oil; motor gasoline; jet kerosene; gas and diesel oil; residual fuel oil; naphtha; other oil; natural gas (dry); other gaseous fossil; municipal wastes (non-biomass fraction); and industrial wastes.

10 For most specifications, a Hausman test rejected the null that there is no correlation between regressors and the individual effects. Therefore, only fixed effects model results are presented herein.

11 One of the most recent studies on the Great Trade Collapse is Eaton et al. (2016), where the decline in world trade in 2009 is expounded.

12 The p-value for the initial PTT in the first subsample is only slightly greater than 10%, namely, 10.1%.

13 As we have discussed thoroughly in subsection 2.3, running a regression on PTT is not the same as running a regression on emissions as does the literature to which we refer to support the choice of explanatory variables. Therefore, it is no surprise that many explanatory variables are not statistically significant. We thank an anonymous referee for suggesting this to us.

Additional information

Funding

Both authors are grateful to the financial support from JSPS KAKENHI Grant Number 25380346. Honma also acknowledges the financial support from JSPS KAKENHI Grant Number 16K03662 and Yoshida acknowledges the financial support from Research Grant of Faculty of Economics, Shiga University.