The Trade Network Structure of the “One Belt and One Road” and Its Environmental Effects

Round 1

Reviewer 1 Report

Thank you for your responses. The authors addressed most of my main concerns. I think the updated version of the manuscript has improved. I think the manuscript is an important contribution to the CO2 emissions literature. 

Author Response

We sincerely thanks the reviewer for the insightful comments!

Reviewer 2 Report

Referee Report on “The Trade Network Structure of the “One Belt and One Road” and Its Environmental Effects_777175

 

This study uses three-stage least squares (3SLS) for analysis of the relationship among CO2 emissions and trade network effect (including Betweenness, Eigenvector, Clustering, Weighted indegree and Weighted outdegree), scale effect (per capita GDP), composition effect (the share of industry added value in GDP), technology effect (net inflows of FDI as a percentage of GDP) and the environmental Kuznets curve effect (GDP2). They find that the status of country nodes in the trade network has a significant impact on CO2 emissions, with differences between high- and low-income countries. The argument is clear and the results are interesting. However, I have the following specific concerns.

 

Major Concerns and Comments:

1. The authors define that countries with GNI per capita greater than or equal to US$4036 are high-income countries, and those with GNI percapita less than US$4036 are low-income countries. However, this is not consistent with the general classification of national income levels. At the same time, the critical value of income classification will affect the analysis results of this study. This is not the same as Aller et al. (2015). The author should explain how the amount of US $ 4036 is given. As far as the measurement method is concerned, the authors can use threshold regression to estimate and classify the countries with high and low income. It can also highlight the contribution of this article relative to Aller et al. (2015).

 

2. What is the reason for the panel data model with fixed effect used by the authors? Is it because of the recommendation of Housman test or other reasons? The author is suggested to explain.

 

3. When discussing the impact of trade network effect on CO2 emissions, the author only considers one effect at a time? Why not put the five effects at the same time? Is it due to insufficient data length? Or the authors can weight the five effects into a trade network effect index and consider the impact of the comprehensive effect.

 

4. The authors point out that the high-income-country sample, the sign of the coefficients of LNCloseness is consistent with the result of Aller et al. (2015). However, after adding the total trade volume (LNTrade), the CO2 emissions are positively and significantly affected by the closeness centrality (LNCloseness), which is contrary to the result of Aller et al. (2015). However, I check the results between this paper and Aller et al. (2015), the description of this paper is incorrect. The authors should check the results carefully.

 

5. This article proves that the increase in the trade volume of the Belt and Road will increase carbon emissions, which is not surprising. It is recommended that the author can put forward policy suggestions to improve the environmental impact to increase the contribution of this article.

 

Minor Concerns

1. The format and figures should be readjusted.

 

Reference

Aller, C.; Ductor, L.; Herrerias, M. J. The world trade network and the environment. Energy Economics 2015, 52, 55-68.

 

Evaluation:

For the above reasons, I believe that the current situation in this article is not suitable for publication in this journal. I encourage authors to make submissions after making appropriate corrections.

Comments for author File: Comments.pdf

Author Response

1. The authors define that countries with GNI per capita greater than or equal to US$4036 are high-income countries, and those with GNI per capita less than US$4036 are low-income countries. However, this is not consistent with the general classification of national income levels. At the same time, the critical value of income classification will affect the analysis results of this study. This is not the same as Aller et al. (2015). The author should explain how the amount of US $ 4036 is given. As far as the measurement method is concerned, the authors can use threshold regression to estimate and classify the countries with high and low income. It can also highlight the contribution of this article relative to Aller et al. (2015).

We thank the reviewer for the helpful comment. We used the income standard established by the World Bank in 2016 while the standard used by Aller et al. (2015) is the standard from World Bank in 2014. We have specified the standard we used in Section 2.3 of the paper. In addition, since the data we used in the paper is based on the monetary value in 2016, we believe it is reasonable to use the standard of income in 2016. 

 

2. What is the reason for the panel data model with fixed effect used by the authors? Is it because of the recommendation of Housman test or other reasons? The author is suggested to explain.

We thank the reviewer for the insightful comment. According to the book by Wooldridge (2010), “Econometric Analysis of Cross Section and Panel Data”,“…we can consistently estimate partial effects in the presence of time-constant omitted variables that can be arbitrarily related to the observables x_it. Therefore, FE analysis is more robust than random effects analysis.”[1] In addition, since we study the impact of OBOR trade network on the CO₂ emission, our goal is to study the difference between various country groups and the countries in the sample are fixed. Hence using data model with fixed effect is more appropriate.

3. When discussing the impact of trade network effect on CO2 emissions, the author only considers one effect at a time? Why not put the five effects at the same time? Is it due to insufficient data length? Or the authors can weight the five effects into a trade network effect index and consider the impact of the comprehensive effect.

The network structure indicators of the trade network are mainly used to measure the indirect trade effect discussed in the paper. Since different network structure indicators reflect different network structure characteristics, it is reasonable to study the individual effect. Furthermore, there usually exist strong correlations among network structure variables and hence, multicollinearity will be an major issue if building a model including all the network indicators.

4. The authors point out that the high-income-country sample, the sign of the coefficients of LNCloseness is consistent with the result of Aller et al. (2015). However, after adding the total trade volume (LNTrade), the CO2 emissions are positively and significantly affected by the closeness centrality (LNCloseness), which is contrary to the result of Aller et al. (2015). However, I check the results between this paper and Aller et al. (2015), the description of this paper is incorrect. The authors should check the results carefully.

In the high-income-country sample, the coefficient for LNCloseness is 0.557 (1^st column in Table 7) while the corresponding coefficient in Aller et al. (2015) is 1.218 (significant under 5%). After introducing LNTrade, the coefficient of LNCloseness of our paper is 2.489 (2^nd column of Table 7) (significant under 10%) while the same coefficient in Aller et al. (2015) is -0.569. Therefore, the sign of the coefficient of LNCloseness (before introducing LNTrade) is the same with the one in Aller et al. (2015). However, the corresponding coefficient has the opposite sign with the one in Aller et al. (2015) after introducing LNTrade. Based on the above, our discussion related to the coefficient of LNCloseness is accurate.

5. This article proves that the increase in the trade volume of the Belt and Road will increase carbon emissions, which is not surprising. It is recommended that the author can put forward policy suggestions to improve the environmental impact to increase the contribution of this article.

We added the following policy recommendations in the paper at the end of the paper. “Developing countries should implement strict environment control policies and actively promote trade liberalization of environmental friendly products, through which energy consumption and carbon emissions will be reduced to improve the environmental quality.”

 

[1] Wooldridge, J. M. . (2010). Econometric analysis of cross section and panel data, 2nd edition. MIT press books, 1(2), 206-209.

Reviewer 3 Report

I have some recommendations towards the authors:

• my suggestion is to promote more recent studies in relevant journals (not only from China but as well international and most important not older than 10 years);
• this will also widen the list of references;
• my concern is also the timeframe of data sets (only until 2016 or in some cases only 2014), is there a possibility to add more recent data until for example 2018?

Author Response

1. my suggestion is to promote more recent studies in relevant journals (not only from China but as well international and most important not older than 10 years); this will also widen the list of references.

We added the following references:

11. Essandoh, O. K., Islam, M., Kakinaka, M. Linking international trade and foreign direct investment to CO2 emissions: Any differences between developed and developing countries? Science of The Total Environment 2020, 712, 136437.
12. Fernández-Amador, O., Francois, F., Tomberger, P. Carbon dioxide emissions and international trade at the turn of the millennium. Ecological Economics 2016, 125, 14-26.
13. my concern is also the timeframe of data sets (only until 2016 or in some cases only 2014), is there a possibility to add more recent data until for example 2018?

The data of this paper is from WDI database. Up to date, 2014 is the latest CO₂ emission data can be obtained from the database. This is the reason why our paper only includes the data till 2014.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

This is a replication exercise of Aller et al. (2015) using bilateral trade data of the “One Belt and One Road” countries. The topic is very important since Co2 emissions, specially in these economies, are increasing over time. However, the paper is underdeveloped, the contribution of the paper is not very clear and the 2SLS methodology is not explained. Specifically, I have the following general comments:

1) The authors should add a theoretical or conceptual framework where they motivate why studying the indirect effects are important. For example, the authors add additional network variables, compared to Aller et al. (2015), but they do not motivate their inclussion. Why the clustering coefficient can affect Co2 emissions? What are the mechanisms through which this network variable affect Co2 emissions?

2) They should also explain in details their 2SLS approach. Are the authors using Instrumental Variables (IV)? The endogenous variables should be clearly stated in the text as well as the instrumental variable. The validity of the instruments should also be discussed in the paper and statistical tests should be added (e.g. Hansen J-test and Cragg Donald test). If they are not using IV. The authors should account for the endogeneity of trade and network. They could simultaneously estimate the environmental, trade, income, and network equations using a three-stage least square procedure and trade partners' population and trade partners' partners population as IV for trade and network. Perhaps, the results presented in model 2 differ from Aller et al. (2015) because the authors assume that trade and network are exogenous. I urge the authors to use a 3SLS and an IV strategy to estimate the trade and network effects.

Specific comments:

3) In all the models, the authors are using the logarithm transformation. How did they treat the observations with 0 value? Is a country with 0 centrality omitted from the sample? I suggest a log(x+1) transformation or the hyperbolic transformation to deal with the 0 values. It is clear from Table 5 that the values of the variables are very low as most of the logs are negative.

4) Equations 6 and 7 are redundant. I would only present equations 8 and 9. More importantly, these specifications are not correct, since the authors do not estimate all the network variables simultaneously.

5) In explaining the results, they refer to Model (1) or Model (2), which is confusing, since the models are given by equations 8 and 9, although as I stated in comment 4 these equations are not correct, the equations should only have a network variable. I would say “the estimation results presented in column (1) show…” instead of “the estimation results of Model (1)…”.

6) References, they are not in alphabetical order.

 

References

Aller, C., Ductor, L., & Herrerias, M. J. (2015). The world trade network and the environment. Energy Economics, 52, 55-68.