Carbon Productivity and Mitigation: Evidence from Industrial Development and Urbanization in the Central and Western Regions of China

Round 1

Reviewer 1 Report

I did not find a response to my previous comments, which I have re-attached here. 

Comments for author File: Comments.pdf

Author Response

Dear reviewer

We highly appreciate your advice. All your suggestions are very important and they have significantly guided our paper writing and research! We have improved the revised manuscript to make it more academically readable. We have carefully considered all issues and advice mentioned above and answered the questions one by one（please find attached document）. We would like to further revise the paper, if necessary.

Author Response File: Author Response.pdf

Reviewer 2 Report

This manuscript focuses on a significant point. By considering the following issues, the quality of the paper can be improved.

1) The literature review part can be extended with recent and qualified papers 

2) The discussion can be strengthened by making more comparison with similar papers.

Author Response

Dear reviewer

We highly appreciate your advice. All your suggestions are very important and they have significantly guided our paper writing and research! We have improved the revised manuscript to make it more academically readable. We have carefully considered all issues and advice mentioned above and answered the questions one by one（please find attached document）. We would like to further revise the paper, if necessary.

Author Response File: Author Response.pdf

Reviewer 3 Report

In this work, the authors investigate carbon productivity (CP) from the perspectives of industrial development and urbanization to mitigate carbon emissions and propose a hybrid model that includes a spatial lag model (SLM) and a fixed regional panel model using data from the 17 provinces in the central and west regions of China from 2000-2018. Good empirical work. However, some issues should be addressed before publication.

1. The format of this paper is too poor, please further improve it. Also, the quality of some figures is too poor.
2. In the Introduction, the authors try to explain their contributions. However, it is not clear for readers, please further clearly state your contributions. Maybe a separate paragraph is needed.
3. In the end of the Introduction, the authors need to explain the manuscript’s structure. Maybe a separate paragraph is needed.
4. The literature about carbon emission and its factors is not comprehensive. I guess the authors may be benefitted from the following article: https://doi.org/10.1080/00036846.2021.1907289, https://doi.org/10.1111/twec.12898, and https://doi.org/10.1016/j.eneco.2021.105191.
5. The authors should add more contents to further discuss your main findings.
6. Please add more specific policy implications based on your main findings, which is very important for your readers.
7. A section of abbreviations should be added as a significant number of symbols and abbreviations have been used.

Author Response

Dear reviewer

We highly appreciate your advice. All your suggestions are very important and they have significantly guided our paper writing and research! We have improved the revised manuscript to make it more academically readable. We have carefully considered all issues and advice mentioned above and answered the questions one by one（please find attached document）. We would like to further revise the paper, if necessary.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Review Round 2:

“How does industrial development and urbanization affect spatially carbon emission efficiency: evidences from the central and west regions of China”

In the last round of revision, I did not find a response to my initial comments (full text in the appendix below), and neither is there a response in this round. While I continue to request a full response from the authors to those comments, I do notice and appreciate the changes they make.

Besides some remaining points from my initial comments, my main reservation about the current manuscript is on the interpretation of the results:

1. The authors are still using causal language to describe the regression results, but their research design does not warrant a causal interpretation. They should provide a clear caveat about this limitation in the introduction/conclusion and when discussing the results.
2. Policy implications (2) and (4) in line 783-797 are tenuously connected to the findings at best. I suggest removing them.

Appendix: Initial Comments

This paper studies the effects of industrial and development factors on carbon productivity (CP) in 17 provinces in China’s central and west regions during 2000-2017. The authors estimate a spatial lag model on provincial-level panel data to examine (1) spatial spillover of CP and (2) the effects of different factors on CP. These factors include the allocation efficiency of labor, fixed investment, and industrial structure, as well as the degree of urbanization. This paper has three main findings. First, there is a positive spatial correlation in CP. Second, urbanization and fixed investments negatively affect CP. Third, the authors conclude that differences in CP across provinces are largely due to different degrees of labor allocation inefficiency and that it should be the focus of effective climate-conscious industrial policy.

Overall, I find the topic quite important, not only because China’s carbon abatement strategy is crucial to the global response to climate change, but it can also provide valuable lessons on how to balance the climate imperative with development needs. However, in my opinion, this paper does not offer a satisfactory answer to the questions it raises. There are many confusing aspects of the conceptual framework and research design, which are not justified or explained. As such, I am not convinced by the authors’ interpretation of their results. The clarity of the writing also needs to be significantly improved for this paper to be publishable. Below are my main comments.

1. There seems to be a deep confusion regarding how CP should be interpreted and used in evaluating carbon abatement policy. The authors take issues with “disparity” and “inequality” in CP across regions and imply that the current policy is inadequate. However, even the first-best policy is not characterized by equal CP everywhere. CP certainly reflects energy efficiency in the production process, but it largely depends on sectoral composition. If a province – because of its resource endowments – has a comparative advantage in capital-intensive and heavy industries, its CP will naturally be lower. Even when carbon emissions are properly priced in the first-best scenario, these provinces will still have a lower CP. Therefore, a lower CP in itself does not imply inadequate policies. I suggest the authors lay out a clearer conceptual framework to guide their empirical analysis.
2. Several key elements in the research design are not well motivated:
   1. The spatial weight matrix in equation (5): I want to understand more about the choice to use a gravity model here as it does not appear to be as natural as in a trade context. Moreover, why divide the geographic distance by the difference in per capita GDP? This seems to be an ad hoc choice that is not explained.
   2. As labor and capital are both factors of production, it is odd that they are included in different forms (equation (3)). The authors choose to capture the “allocation efficiency” of labor using a Theil Index while directly including the average productivity of capital. Why? It is also unclear that the Theil Index represents allocative efficiency. Efficient allocation of labor requires the marginal productivity of labor to equalize across industries, but this index is based on average
   3. The TS measure is also very confusing. Some more intuition on what it measures should be given.

3. I find the interpretation of some findings, including the main ones, to be quite sloppy. For example, the authors test for spatial spillover of CP, but it is unclear what mechanism they had in mind when they propose to do so. Is it due to technology transfer, labor mobility, or simply correlated comparative advantage? Without the mechanism, it is difficult to understand the meaning of such spatial correlations. And yet, the authors claim a causal interpretation from these results, stating that “a higher CP … can more highly improve the CP of the surrounding provincials…” and a lower CP reduces the CP of surrounding provinces (line 412-416).

Similarly, it is unclear whether the results from the spatial lag model can be interpreted as causal. One particular concern I have is that the industrial and urbanization factors could be simply picking up the sectoral composition in the province, and the results could be driven by strong spatial correlation in comparative advantage. These concerns could be partly addressed if the authors can (a) spell out the identification assumptions more clearly and (b) provide a sense of what the causal mechanism is (e.g. how does capital intensity in Shaanxi affect CP in Gansu, and why is it as important as the capital intensity in Gansu itself?).

The authors make several statements comparing the effect of different factors, such as “improving TL is more efficient than improving TK to enhance CP” (line 475-476). Given that we don’t know how the costs of improving TL, TK, and TS by 1% compare with each other, I don’t think it is possible to state which is more efficient.

Minor comments:

4. How are standard errors calculated for correct statistical inference? I couldn’t find the explanation anywhere.
5. The regression tables need to be labeled more clearly: use meaningful labels of the variables instead of variable names in the statistical software.
6. The equations should be explained more thoroughly, including the meaning of the subscripts and all variables.
7. This paper would also benefit from extensive copy-editing to correct grammatical errors (in the title as well) and improve readability.
8. It would be helpful to provide a summary statistics table so that readers can have a better idea of variation in the key variables.
9. This paper is also related to the broad literature on environmental Kuznets curve. The authors can consider citing it in the lit review section.

Author Response

Dear reviewer,

We highly appreciate your advice. All your suggestions are very important and they have significantly guided our paper writing and research! We have improved the revised manuscript to make it more academically readable. We have carefully considered all issues and advice mentioned above and answered the questions one by one. We would like to further revise the paper, if necessary.

Round 1 

Point 1: There seems to be a deep confusion regarding how CP should be interpreted and used in evaluating carbon abatement policy. The authors take issues with “disparity” and “inequality” in CP across regions and imply that the current policy is inadequate. However, even the first-best policy is not characterized by equal CP everywhere. CP certainly reflects energy efficiency in the production process, but it largely depends on sectoral composition. If a province – because of its resource endowments – has a comparative advantage in capital intensive and heavy industries, its CP will naturally be lower. Even when carbon emissions are properly priced in the first-best scenario, these provinces will still have a lower CP. Therefore, a lower CP in itself does not imply inadequate policies. I suggest the authors lay out a clearer conceptual framework to guide their empirical analysis. 


Response 1: Thank you for your valuable suggestions. We have greatly improved the manuscript based on your suggestions.

Yes, you are right. Regions would prefer to develop intensive capital and heavy industrial when the region has some comparative resource endowment in terms of energy and other resources, generally resulting in a lower CP.

Carbon productivity (CP) is an important indicator to measure the low-carbon economic development for a region (Kaya and Yokobori, 1999). Thus, CP is generally used to measure the level of low-carbon economy. Furthermore, numerous studies have verified that CP, and the correlation between the studied factors and CP show spatial-temporal differences. Thus, in this study, we measure CP, analyze CP changes and then investigate the correlation between the studied factors and CP through a SLM model in the central and west regions of China, providing specific suggestions from the energy consumption side including production and likelihood to reach carbon emission mitigation.

We do not think the current policy from China’s central government is inadequate. On the contrary, we believe the policy from China’s central government for carbon emission mitigation is good, but local governments are the policy executors and their decisions are affected by their self-interest when implementing policies enacted by the Chinese central government [42]. In addition, the more highly-developed regions tend to quickly promote CP by importing high-carbon products from less-developed regions or by transferring their high-pollution industries to less-developed regions in the short term [25]. In fact, this trade in goods consequently transfers carbon emissions from highly-developed regions to less-developed regions, and the total amount of carbon emissions in the country is not reduced. On the contrary, the real and final result is sometimes off from the goal of China’s central government.  

Point 2: The spatial weight matrix in equation (5): I want to understand more about the choice to use a gravity model here as it does not appear to be as natural as in a trade context. Moreover, why divide the geographic distance by the difference in per capita GDP? This seems to be an ad hoc choice that is not explained.

Response 2: You are right. By your suggestions, we have re-edited equation (5) and improved its description. According to the general pattern of the gravity model, we divide the geographic distance by the difference value of per capita GDP as economic geographic distance between two regions (as shown in line 312-319,on page 9-10), considering source flows and interactions between regions driven by the wealth gaps of urban residents.

Point 3: As labor and capital are both factors of production, it is odd that they are included in different forms (equation (3)). The authors choose to capture the “allocation efficiency” of labor using a Theil Index while directly including the average productivity of capital. Why? It is also unclear that the Theil Index represents allocative efficiency. Efficient allocation of labor requires the marginal productivity of labor to equalize across industries, but this index is based on average productivity.

Response 3: Thanks a lot for your professional suggestions. By your suggestions, we have re-edited the variable descriptions.

(1)  The estimation forms calculating TL and TK, are different, mainly resulting from the different attributes of these two factors. On the one hand, in this study, TL is used to represent the level of resource allocation, in which workforce can quickly flow from one industry to another. On the other hand, TK, represents utilization efficiency of physical capital, which cannot make a quick transfer between industries in the short run, and some social physical capital in a region, as infrastructure investments, have the characteristics of quasi-public goods.  Therefore, to simplify analysis, we use the equation () to measure the productivity of physical capital inputs.

(2) Yes, you are right. In theory, the optimal and ideal state is that the marginal output of per-unit workforce across industries is the same according to production theory. But in fact, developers/producers generally produce outputs reaching a state located in the range between the average productivity of factors and the marginal productivity of factors (Hal R. Varian, 1999). Therefore, to simplify analysis, we use the average productivity of per-unit workforce to investigate the level of workforce allocation.

Point 4: The TS measure is also very confusing. Some more intuition on what it measures should be given.

Response 4: The method of TS measurement in this study refers to Moore (1978), in which the method proposed by Moore is based on the fact that the structure of output in any period can be described by a vector whose coordinates are the quantities of outputs which form the basis for calculating the index numbers. These vectors are assumed to lie on the production possibility surfaces: economic growth corresponds to an outward shift of those surfaces, and the extent of that growth is estimated by index numbers calculated by weighting the elements of the output vectors. This method is desirable because measuring structural change by the angle between the two vectors can be done very simply using the formula from linear algebra for the cosine of the angle between vectors. 

Point 5: I find the interpretation of some findings, including the main ones, to be quite sloppy. For example, the authors test for spatial spillover of CP, but it is unclear what mechanism they had in mind when they propose to do so. Is it due to technology transfer, labor mobility, or simply correlated comparative advantage? Without the mechanism, it is difficult to understand the meaning of such spatial correlations. And yet, the authors claim a causal interpretation from these results, stating that “a higher CP … can more highly improve the CP of the surrounding provincials…” and a lower CP reduces the CP of surrounding

provinces (line 412-416).

Similarly, it is unclear whether the results from the spatial lag model can be interpreted as causal. One particular concern I have is that the industrial and urbanization factors could be simply picking up the sectoral composition in the province, and the results could be driven by strong spatial correlation in comparative advantage. These concerns could be partly addressed if the authors can (a) spell out the identification assumptions more clearly and (b) provide a sense of what the causal mechanism is (e.g. how does capital intensity in Shaanxi affect CP in Gansu, and why is it as important as the capital intensity in Gansu itself?). The authors make several statements comparing the effect of different factors, such as “improving TL is more efficient than improving TK to enhance CP” (line 475-476). Given that we don’t know how the costs of improving TL, TK, and TS by 1% compare with each other, I don’t think it is possible to state which is more efficient.

Response 5: Thanks for your suggestions. By your suggestions, we have made the revised manuscript more academically readable.

(1) Accompanied with rapid technology progress, spatial interactions between regions have continuously increased along with growing economic and social mutual exchanges. Meanwhile, numerous studies have proved that CP has significant spatial effects (Anselin, 2001). What’s more, in this study, before the spatial econometric model is proposed, we conduct an autocorrelation test on CP (in the section “4.1 Spatial autocorrelation test” ), and the results show that CP has a significant spill-over effect. Then, based on the results of scientific and statistic tests, and the results of data statistic analysis in the studied regions, we finally propose the SLM model.

(2) Thanks for your comments.

Yes, you are right. Our further work is to introduce the inherent differences of comparative advantages between regions to further investigate the inherent mechanism causing the difference in CP values between regions.

In this study, we focused on the spatial and temporal evolution of CP in the west and central regions, and the correlation between studied factors (energy consumption side: industrial development and urbanization) and CP, in order to provide specific suggestions for mitigating carbon emissions regarding consumption.

Our studied results verify that CP has a significant spatial effect, and that there are correlations between the studied factors and CP. In the revised “Discussion”, we improved the section to focus on discussing what caused these phenomena, and then provided specific suggestions for mitigating carbon emissions ( pages 24-25).

(3) The estimated results show that the intensity of marginal effects of TL on CP is higher than of TK on CP. According to the economic theory, it shows that the sensitivity of CP to TL is higher than to TS and thus we provide specific policy implications of paying more attention to TL than TS at current stage in the west and central regions. In this study, TL means the level of resource allocation, and TS, to some extent, represents technological progress. In the long run, whether the region is less-developed or highly-developed, increasing TS should be the most effective way to mitigate carbon emissions. However, due to the characteristics of economic development in this stage in the central and west regions, our findings show that improving resource allocation efficiency is more effective than increasing TS. In another words, results show that the resource allocation in these three industries in the central and west region is inefficient and should be improved.

Point 6: How are standard errors calculated for correct statistical inference? I couldn’t find the explanation anywhere.

Response 6: Thanks for your comments. By your suggestion, we have added the t-statistic description in tables 4-5.

Point 7 : The regression tables need to be labeled more clearly: use meaningful labels of the variables instead of variable names in the statistical software.

Response 7: Thanks lot for your help.  

The labels have been re-edited in the tables.

Point 8 : The equations should be explained more thoroughly, including the meaning of the subscripts and all variables.

Response 8: Yes, you are right. We have re-edited descriptions of all equations and subscripts and variables in the revised manuscript.

Point 9 : This paper would also benefit from extensive copy-editing to correct grammatical errors (in the title as well) and improve readability.

Response 9: Yes. The revised manuscript has been re-edited professionally for English readability.

Point 10 : It would be helpful to provide a summary statistics table so that readers can have a better idea of variation in the key variables.

Response 10: Thank for your professional comments. By your suggestion, summary statistics (Shown in Table 2) for key variables have been added to the revised manuscript.

Point 11 : This paper is also related to the broad literature on environmental Kuznets curve. The authors can consider citing it in the lit review section.

Response 11: Yes, your comment is very useful.

In the revised manuscript, we have cited some literature about environmental Kuznets curve (EKC)  in section “2. Literature review” (shown in page 3, lines 124-129).

Round 2

Point 1 : The authors are still using causal language to describe the regression results, but their research design does not warrant a causal interpretation. They should provide a clear caveat about this limitation in the introduction/conclusion and when discussing the results.

Response 1: Thank for your professional suggestions. By your suggestion, we have added a caveat about this limitation in section “6. Conclusions and policy suggestions”. (shown in page 25, lines 721-729).

Point 2 : Policy implications (2) and (4) in line 783-797 are tenuously connected to the findings at best. I suggest removing them.

Response 2: Thank for your professional suggestions. By your suggestion, we have removed them.

Author Response File: Author Response.pdf

Reviewer 3 Report

I believe the authors have answered my questions well, thus I recommend to publish this manuscript.

Author Response

Dear reviewer,

Thanks for your invaluable comments. By your suggestion, the revised paper has been professional edited by a native English speaker.

Round 3

Reviewer 1 Report

I now recommend this paper for publication. Thank you for your effort to improve the draft. 

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 paper explores the effects of industrial development and urbanization on carbon productivity in China using a spatial lagging model and a fixed regional panel model. There are several findings. First, the authors find that industrial development and urbanization significantly impact carbon productivity, with urbanization and fixed investment having adverse effects on carbon productivity. In addition, carbon productivity has an apparent spatial spillover effect across the 17 central and western provinces. Moreover, while the impacts of industrial development and urbanization are similar across the regions with relatively small disparities, the differences in industrial workforce allocation are evident. To achieve the goal of carbon emission reduction in China, the authors suggest that policymakers in China should promote industrial technology, improve workforce allocation, and increase urbanization in the central and western regions.

While this paper examines a timely topic on carbon mitigation in China – the largest emitter globally – it does not yet seem to be sufficiently finished, both in terms of the analysis and the exposition. Hence, I recommend rejection for the paper. My comments are expressed in the following.

First, the paper links industrial development, urbanization, and carbon productivity using statistical models. The results seem to be correlations between variables, which lacks the power to suggest that industrial development and urbanization causally impact carbon productivity. Hence, the discussion of policy implications regarding reducing carbon emissions from promoting industrial technology, improving workforce allocation, and increasing urbanization is not convincing.

Second, the use of a spatial lagging model and a fixed regional panel model is interesting. Still, the authors should make more effort to state the motivation to use these two models.

Also, as for the audiences of Sustainability, people will be interested in learning how the studied factors correlating with carbon productivity lead to deep decarbonization, which the paper does not render. Notably, China has made a promise the achieve carbon neutrality in 2060. The paper could extend their discussions on how their work can contribute to this mitigation goal.

Lastly, the organization of the paper can be improved in several ways. The current version has part of the results (Tables 1-4) shown in the model description section (Sections 3 & 4), which are then discussed in the results section (Section 5). This makes the paper look convoluted. In addition, Sections 3 & 4 can be combined as they are both methodology sections. While the paper presents some key takeaways and results, it would benefit a lot from revision by a native English speaker.

Reviewer 2 Report

I do not think this paper warrants publication at Sustainability. The research design is lacking in motivation and rigor. The findings are hard to interpret and the authors make unsubstantiated claims throughout the paper. The writing needs to be improved a lot. I discuss these issues in more detail in my comments to the authors.

Reviewer 3 Report

Please find the attached document.

Comments for author File: Comments.pdf