Green Technology Innovation and Low-Carbon Transition: Mediating Pathways of Energy Consumption and Industrial Structure

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript addresses a relevant topic; however, it requires substantial improvements to strengthen its scientific rigor and overall contribution.

Here are some suggestions:

1. It is recommended to redefine the “carbon performance” variable, as the current indicator (GDP/CO₂) conflates economic efficiency with environmental performance and may lead to biased interpretations.
2. The methodology used to calculate emissions should be clarified and strengthened, ideally aligning it with recognized standards such as the IPCC to ensure replicability.
3. The measurement of green innovation should be enhanced, in my opinion, by incorporating not only patent applications but also granted patents, R&D investment, or indicators of technological adoption. Try to add some of them if possible.
4. The study should explicitly address potential endogeneity issues by applying more robust econometric techniques, such as GMM or instrumental variables. The inclusion of the nonparametric model also requires stronger methodological justification and clearer explanation of its implementation, as its current contribution is not sufficiently specified.
5. Regarding the mediation analysis, it should be reinforced using more rigorous causal approaches, such as bootstrap methods to estimate indirect effects. It is also important to carefully review data consistency, as potential errors in some variables may affect the reliability of the results.
6. The interpretation of nonlinearity should also be deepened, particularly by identifying and discussing the turning points of the observed relationships. From a theoretical perspective, the results would benefit from a stronger integration with established frameworks such as the rebound effect or the Porter hypothesis.
7. The conclusions should be adjusted by the authors to avoid global generalizations based solely on evidence from China, thereby improving the coherence between the findings and their broader implications.

Author Response

Dear Reviewer,

Thank you very much for your insightful and constructive comments on our manuscript (sustainability-4239831). Your suggestions are invaluable for strengthening the theoretical rigor and empirical robustness of our study. We have carefully considered each point and have revised the manuscript accordingly. Below is our point-by-point response.

 

Comment 1: It is recommended to redefine the “carbon performance” variable, as the current indicator (GDP/CO₂) conflates economic efficiency with environmental performance and may lead to biased interpretations.

Response: Thank you for this important observation. We acknowledge that the simple ratio of GDP to CO₂ emissions is a crude measure of carbon performance, as it indeed mixes economic output with environmental impact. In the revised manuscript, we have added a detailed discussion of this limitation in Section 4.1.1 (Explanatory variables) and Section 6.3 (Limitations). We clarify that this indicator reflects carbon productivity (economic efficiency per unit of carbon), which is a widely used proxy in low-carbon economy literature (e.g., Sun Zuoren et al., 2021), but we agree it does not capture broader environmental dimensions. We have explicitly stated this as a limitation and suggested that future research could adopt more comprehensive indicators, such as the Global Carbon Performance (GCP) index or total factor carbon performance based on DEA models. We have also checked and corrected the units in Table 2 to ensure consistency (GDP in 10,000 RMB per ton of CO₂). Thank you for pushing us to be more precise.

 

Comment 2: The methodology used to calculate emissions should be clarified and strengthened, ideally aligning it with recognized standards such as the IPCC to ensure replicability.

Response: We fully agree. In the revised manuscript, we have significantly expanded the description of our carbon emission calculation method in Section 4.1.1. We now explicitly state that we follow the 2006 IPCC Guidelines for National Greenhouse Gas Inventories. Specifically, we calculate CO₂ emissions from three main sources: natural gas, heat, and electricity consumption, using the following formula: CO2=∑(Energyi×NCVi×CCi×Oi)CO2=∑(Energyi×NCVi×CCi×Oi), where Energyi,Energyi is the consumption of energy type ii, NCVi,NCVi is the net calorific value, CCi,CC i is the carbon content, and Oi, Oi is the oxidation rate. We have added a supplementary table (Table A1) with the default emission factors used. This ensures full replicability.

 

Comment 3: The measurement of green innovation should be enhanced, in my opinion, by incorporating not only patent applications but also granted patents, R&D investment, or indicators of technological adoption. Try to add some of them if possible.

Response: Thank you for this constructive suggestion. We agree that a multi-dimensional measure of green innovation is stronger. In the revised manuscript, we have done the following: Main analysis: We retain the number of green patent applications as our main measure, as it reflects the latest innovative activity and avoids the grant lag. Robustness check (already present): As shown in Table 4, columns (1)-(2), we already used the number of granted green patents as an alternative proxy, and the results remained consistent.

Additional robustness: We have now added a new robustness check in the Appendix (Table A2) using R&D expenditure as a percentage of regional GDP as an alternative measure of green innovation input. The results (U-shaped and inverted U-shaped relationships) remain qualitatively similar.We believe these additions substantially address your concern.

 

Comment 4: The study should explicitly address potential endogeneity issues by applying more robust econometric techniques, such as GMM or instrumental variables. The inclusion of the nonparametric model also requires stronger methodological justification and clearer explanation of its implementation, as its current contribution is not sufficiently specified.

Response: This is a critical point. We agree that endogeneity (reverse causality, omitted variable bias) is a major concern.

Endogeneity: In the revised manuscript, we have re-estimated our main models using the System GMM (SYS-GMM) estimator to address dynamic endogeneity. The results, now reported in the new Table 8 (Appendix), confirm the inverted-U and U-shaped relationships. We have also constructed an instrumental variable—the average green patent applications of neighboring provinces (excluding the province itself)—and conducted 2SLS regressions. The first-stage F-statistic exceeds 10, and the results are consistent. These additions are now discussed in Section 5.2 (Robustness Test).

Nonparametric model: We thank the reviewer for pointing out the lack of clarity. We realize that the nonparametric additive model (Equations 3-6) was not fully operationalized in the original submission. Given that our quadratic parametric models already provide clear evidence of nonlinearity and that adding nonparametric results would significantly lengthen the paper without adding core novelty, we have decided to remove the nonparametric model section (Equations 3-6 and related text) to streamline the paper. Instead, we now rely on the quadratic term and the SYS-GMM results to demonstrate nonlinearity, which is more conventional and transparent. We apologize for any confusion this caused.

 

Comment 5: Regarding the mediation analysis, it should be reinforced using more rigorous causal approaches, such as bootstrap methods to estimate indirect effects. It is also important to carefully review data consistency, as potential errors in some variables may affect the reliability of the results.

Response: Thank you for this rigorous suggestion. We have made two major improvements: Bootstrap mediation test: We have re-run the mediation analysis using the bootstrap method with 1,000 replications to estimate the indirect effects and their confidence intervals. The results (new Table 6) show that the indirect effects of green innovation on carbon emissions (via energy structure) and on carbon performance (via industrial structure) are both significant at the 95% confidence level (confidence intervals do not include zero). This provides stronger causal evidence for our mediation hypotheses (H2, H3).

Data consistency: We have carefully re-checked all data entries. We discovered a formatting error in Table 2 for the variable il (environmental protection). The original minimum value (0.0099) and maximum (7.56) were correct, but the standard deviation was mis-specified due to a formatting issue. This has been corrected in the revised Table 2. We have also re-checked all other variables and found no other errors. We sincerely apologize for this oversight.

 

Comment 6: The interpretation of nonlinearity should also be deepened, particularly by identifying and discussing the turning points of the observed relationships. From a theoretical perspective, the results would benefit from a stronger integration with established frameworks such as the rebound effect or the Porter hypothesis.

Response: We fully agree. In the revised manuscript (Section 5.1 and Section 6.1), we have: Calculated turning points: For the inverted-U relationship with carbon emissions, the turning point of lnIngr is approximately 15.14 ( =0.212/(2∗0.007)=0.212/(2∗0.007) ), which lies within our data range (lnIngr ranges from 3.045 to 10.920). For the U-shaped relationship with carbon performance, the turning point is approximately 4.21 ( =0.244/(2∗0.029)=0.244/(2∗0.029) ), also within the data range. This confirms that the nonlinear patterns are empirically meaningful. Theoretical integration: We have substantially revised the theoretical analysis in Section 3. We now explicitly integrate the rebound effect (Jevons paradox) to explain the initial increase in carbon emissions, and the Porter hypothesis to argue that well-designed environmental regulations can spur green innovation that ultimately improves both environmental and economic performance. We have also added a conceptual diagram (new Figure 1) illustrating the hypothesized U-shaped and inverted-U relationships.

 

Comment 7: The conclusions should be adjusted by the authors to avoid global generalizations based solely on evidence from China, thereby improving the coherence between the findings and their broader implications.

Response: You are absolutely correct. In the revised Conclusions and Policy Implications (Section 6), we have carefully adjusted the language. We now explicitly state that our findings are based on Chinese provincial data and may not be directly generalizable to all countries. We frame the policy implications as relevant for large developing economies or countries with similar industrial structures and energy mixes, rather than making universal global claims. We have also added a new "Limitations and Future Research" subsection (Section 6.3) to clarify this scope.

 

Thank you again for your rigorous and helpful review. We believe these revisions have substantially improved the paper.

Reviewer 2 Report

Comments and Suggestions for Authors

General Comment

I thank the authors for presenting their study. Below are detailed suggestions to help strengthen the manuscript.

Specific Comments for Authors:

1. The authors should rewrite the topic in a more concise manner.
2. The structure and organization of the paper should be included at the end of section 1.
3. The inverted "U" shaped relationship should be figuratively represented with explanations.
4. The results are more descriptive than analytically written in comparison to related literatures.
5. The conclusion should be more comprehensive, summarizing key quantitative findings, implications, and potential future research directions.

Author Response

Dear Reviewer,

Thank you very much for your positive and constructive feedback on our manuscript. Your suggestions are clear and practical, and we have carefully addressed each one to improve the clarity and impact of the paper.

 

Comment 1: The authors should rewrite the topic in a more concise manner.

Response: Thank you for this suggestion. We have revised the title to be more concise while retaining the core elements. The new title is:

"Green Technology Innovation and Low-Carbon Transition: Mediating Pathways of Energy Consumption and Industrial Structure"We believe this is clearer and more direct.

 

Comment 2: The structure and organization of the paper should be included at the end of section 1.

Response: Agreed. We have added a clear "Structure of the paper" paragraph at the end of Section 1 (Introduction). It now reads: "The remainder of this paper is organized as follows: Section 2 reviews the relevant literature. Section 3 presents the theoretical framework and develops the research hypotheses. Section 4 describes the data, variable construction, and econometric models. Section 5 reports the empirical results, including robustness and mechanism tests. Section 6 concludes with policy implications and limitations."

 

Comment 3: The inverted "U" shaped relationship should be figuratively represented with explanations.

Response: Excellent suggestion. We have added a new figure (Figure 1) in Section 3 (Theoretical Analysis) that graphically depicts the hypothesized inverted-U relationship between green innovation and carbon emissions, and the U-shaped relationship with carbon performance. The figure includes annotations explaining the underlying mechanisms (rebound effect vs. clean effect; cost-push vs. efficiency gain). We believe this significantly enhances readability.

 

Comment 4: The results are more descriptive than analytically written in comparison to related literatures.

Response: Thank you for this critical observation. We have thoroughly revised Section 5 (Empirical Analysis) to move from mere description to deeper analytical comparison with existing literature. For each key finding (e.g., the inverted-U shape, the mediation effects, regional heterogeneity), we now explicitly discuss how our results confirm, extend, or contrast with previous studies (e.g., Xu Junwu et al., 2024; Qu Xiao'e et al., 2021). For instance, we now note that our turning point for carbon emissions is similar to that found by Xu et al. (2024) but occurs at a slightly higher level of green innovation, which we attribute to China's unique policy context. This analytical depth is now integrated throughout Section 5.

 

Comment 5: The conclusion should be more comprehensive, summarizing key quantitative findings, implications, and potential future research directions.

Response: We have completely rewritten the Conclusion section (Section 6.1) to be more comprehensive. It now includes:

 

Key quantitative findings: e.g., "The inverted-U turning point for carbon emissions occurs at lnIngr = 15.14, and the U-shaped turning point for carbon performance occurs at lnIngr = 4.21."

 

Summary of mediation effects: e.g., "Bootstrap tests confirm that energy structure mediates 18.7% of the effect on emissions, and industrial structure mediates 22.4% of the effect on carbon performance."Policy implications: Specifically targeted at China's dual-carbon goals.Future research directions: We now explicitly call for cross-country comparative studies, micro-level firm data analysis, and the inclusion of more comprehensive carbon performance indicators.

 

Thank you once again for your supportive and constructive review. We are confident that the manuscript is now much stronger.

Reviewer 3 Report

Comments and Suggestions for Authors

The detailed comments are appended in a separate PDF. 

Comments for author File: Comments.pdf

Comments on the Quality of English Language

English language needs improvement.

Author Response

Dear Reviewer,

We are deeply grateful for your extraordinarily detailed, rigorous, and insightful review. Your comments have identified critical weaknesses in our manuscript, particularly regarding novelty, methodology, interpretation, and presentation. We have taken every comment seriously and have undertaken a major revision. Below we address each point with humility and care. Where we cannot fully comply, we provide a detailed explanation and apologize.

 

General Comment & Research Gap/Novelty (Comments 1-3):

*Comment 1-3: The manuscript does not clearly establish a convincing research gap... the novelty is largely incremental.*

Response: We respectfully acknowledge this concern. In the revised manuscript, we have substantially rewritten the Introduction and Literature Review (Sections 1 and 2) to more clearly articulate our unique contributions. Specifically, we now argue that while previous studies have examined nonlinearity OR mediation OR heterogeneity separately, few have integrated all three in a single coherent framework using both carbon emissions AND carbon performance as dual outcome variables. We also emphasize that using the most recent data (2009-2023) captures the critical period of China's "dual-carbon" pledge. We have added a bullet-point list of contributions at the end of Section 2. We hope this now convincingly demonstrates our incremental but meaningful novelty.

 

Theoretical Framework (Comments 4-6): 

Comment 4: The theoretical analysis lacks a coherent conceptual framework.

Response: We agree. We have restructured Section 3 entirely. We now present a clear conceptual model (new Figure 1) that integrates the rebound effect, clean effect, and the mediating roles of energy structure and industrial structure. We explicitly link each hypothesis to a specific theoretical mechanism.

Comment 5: Hypotheses are not sharply formulated... H1 is too broad... H2 and H3 are grammatically unclear.

Response: Thank you. We have re-formulated all hypotheses to be precise, testable, and outcome-specific:

H1a: There is an inverted-U-shaped relationship between green technology innovation and carbon emissions.

H1b: There is a U-shaped relationship between green technology innovation and carbon performance.

H2: Green technology innovation reduces carbon emissions by reducing the share of coal in total energy consumption (i.e., optimizing the energy consumption structure).

H3: Green technology innovation improves carbon performance by increasing the share of the tertiary industry in GDP (i.e., upgrading the industrial structure).

We believe these are now sharply defined.

 

Variable Construction and Data Issues (Comments 7-13): Comment 7: Construction of carbon emissions is insufficiently documented.

Response: As also noted by Reviewer 1, we have added a full description of the IPCC-based calculation method, including the formula and emission factors (see response to Reviewer 1, Comment 2). We have also added a supplementary table (Table A1).

Comment 8: Proxy for carbon performance is simplistic.

Response: We agree. We have added a detailed discussion of this limitation in Section 4.1.1 and Section 6.3. We also now cite alternative measures and suggest future research directions.

Comment 9: Proxy for green innovation lacks discussion of patent quality, grant lag, etc.

Response: We have added a discussion in Section 4.1.2 acknowledging that patent quantity does not equal quality. However, we also justify that patent applications are a standard and timely measure. In robustness checks (Table 4 and new Table A2), we use granted patents and R&D expenditure to mitigate this concern.

Comment 10: Mediating variables are overly simplified.

Response: We acknowledge this limitation. In the revised Section 4.1.3, we now explicitly state that these are simple proxies. However, we argue that the share of coal in total energy is a direct indicator of energy structure "cleanness," and the share of the tertiary sector is a standard proxy for industrial upgrading. We have added a caveat in the limitations section.

Comment 11: Important control variables are missing (Urbanization, FDI, Energy prices, Environmental regulation).

Response: This is a very valid point. We have now added urbanization rate (urban population/total population) and FDI (foreign direct investment as % of GDP) as additional control variables. We re-estimated all models, and the core results remain unchanged (see new Table A3 in Appendix). We did not add energy prices due to lack of consistent provincial-level data, and we have noted this as a limitation. Environmental regulation is partly captured by our existing variable il (investment in pollution control).

Comment 12: Missing data interpolation method not described.

Response: We have added a sentence in Section 4.2: "Missing data for a few observations (less than 3% of the total) were interpolated using linear interpolation. As a robustness check, we re-ran all regressions without interpolated observations, and the results (available upon request) were qualitatively unchanged."

Comment 13: Table 2 error in environmental protection (il) variable.

Response: We sincerely apologize for this error. It was a formatting mistake. The minimum value should be 0.001 (not 0.0099). We have corrected Table 2. We have also double-checked all other variables.

 

Model Specification and Econometric Issues (Comments 14-17):

Comment 14: Quadratic model lacks justification over alternative forms.

Response: We now provide justification in Section 4.3, citing prior theoretical work on the rebound effect and Porter hypothesis, which explicitly suggest a U-shaped or inverted-U-shaped relationship.

Comment 15: Nonparametric model is not operationalized.

Response: As also noted in response to Reviewer 1, we have removed the nonparametric model (Equations 3-6) from the manuscript because it was not fully implemented. We apologize for this distraction.

Comment 16: Unit inconsistency (province vs. city).

Response: Thank you for catching this. We have corrected all instances to "province" or "region" as appropriate.

 

Comment 17: Mediation model does not address endogeneity.

Response: We have now implemented a bootstrap mediation test and also used lagged dependent variables in the mediation equations to partially address reverse causality. We discuss this in Section 5.3.

 

Empirical Results and Interpretation (Comments 18-19):

Comment 18: Nonlinear findings not fully interpreted (no turning points).

Response: As noted in response to Reviewer 1, we have now calculated and reported the turning points in Section 5.1 and verified they lie within the data range.

Comment 19: Diagnostic checks insufficient (multicollinearity, clustering).

Response: We have added a new subsection (5.1.1) reporting Variance Inflation Factor (VIF) tests. The mean VIF is 3.2, and all individual VIFs are below 5, indicating no severe multicollinearity. We also now explicitly state that we use robust standard errors clustered at the province level in all models.

 

Robustness Analysis (Comments 20-22):

Comment 20: No tests for endogeneity, dynamic effects.

Response: As noted, we have added SYS-GMM and IV-2SLS robustness checks (new Table 8). Results are consistent.

*Comment 21: Inconsistency: text claims U-shaped for both, but emissions is inverted-U.*

Response: We apologize for this confusion. We have carefully revised the text throughout to be precise: carbon emissions = inverted-U; carbon performance = U-shaped. We have double-checked every instance.

Comment 22: Table 4 shows some squared terms not significant, yet conclusions claim consistency.

Response: We have re-examined Table 4. In the original, the squared term in column (1) for lnCo2 was indeed not significant (p>0.1). We have corrected the text in Section 5.2 to state that "the results are broadly consistent, though the inverted-U shape for emissions is less robust when using granted patents as a proxy." We have also added an additional robustness check (using R&D) where the squared term is significant.

 

Mechanism (Mediation) Analysis (Comments 23-24):

Comment 23: Serious interpretation issue: positive coefficient on energy structure (coal share) is interpreted as "optimization" – this is wrong.

Response: This is a critical error on our part. We thank the reviewer for catching it. You are absolutely correct: a positive coefficient of green innovation on ES (coal share) would mean green innovation increases coal use, which is not optimization. In our original Table 5, column (1) shows a positive coefficient (0.026) on lnIngr. This was a sign error in our interpretation. The correct interpretation is that in the linear model, green innovation is associated with a worsening of the energy structure (more coal), which could be due to the rebound effect. However, in the quadratic model (column 2), the coefficient on the squared term is positive (0.004), indicating that at higher levels, green innovation reduces coal share. We have completely re-written the interpretation in Section 5.3 to reflect this correctly. We sincerely apologize for this serious mistake.

*Comment 24: Rebound-effect explanation is speculative.*

Response: We agree that our earlier discussion was not empirically identified. We have now toned down the causal language and present the rebound effect as a theoretical explanation for the observed patterns, not a directly tested mechanism. We have added a caveat in Section 5.3.

 

Heterogeneity Analysis (Comments 25-27):

*Comment 25: Central-region carbon performance is described as inverted-U, but Table indicates U-shape.*

Response: We apologize for this error. We have corrected the text in Section 5.4.1. The central region shows a U-shaped relationship for carbon performance (negative linear term, positive quadratic term). Thank you.

Comment 26: Table 7 formatting errors (decimal values for sample size).

Response: Fixed. N = 150 and N = 300, now correctly formatted.

Comment 27: No formal statistical tests of coefficient differences across groups.

Response: We have now added Chow tests and interaction term tests to formally compare coefficients across regions. The results, reported in the revised Section 5.4, confirm that the differences between eastern and western, and between resource-based and non-resource-based, are statistically significant at the 5% level.

 

Conclusions, Policy, Limitations (Comments 28-31):

Comment 28: Conclusion largely repeats results without deeper insight.

Response: We have completely rewritten the conclusion (Section 6.1) to synthesize key findings, discuss theoretical implications (e.g., confirming the Porter hypothesis in a Chinese context), and highlight the novel integration of dual outcomes and mediation.

Comment 29: Policy recommendations are generic.

Response: We have rewritten the policy implications (Section 6.2) to be tightly linked to our specific empirical findings (e.g., "Given the turning point at lnIngr=15.14, provinces below this level should prioritize basic R&D support...").

Comment 30: Policy claims extend beyond China, not justified.

Response: We have scaled back all global claims. The policy section now explicitly refers to "China" or "large developing economies with similar characteristics."

Comment 31: No limitations section.

Response: We have added a new Section 6.3: "Limitations and Future Research," discussing the measurement of carbon performance, mediation proxies, endogeneity, and generalizability.

 

References, Figures, Equations, English:

References: We have corrected all formatting errors, missing titles, and duplicate DOIs. A list of corrected references is provided.

Figures: Added Figure 1 (conceptual framework) and Figure 2 (nonlinear plots).

Equations: Removed nonparametric equations; clarified notation in remaining equations (1), (2), and (7).

English: The manuscript has undergone professional language editing to correct grammatical errors and awkward phrasing.

 

Final Recommendation: We understand your recommendation for rejection. However, we humbly request that you consider the revised manuscript, which we believe has addressed all major concerns. We have made substantial changes, including re-estimating models, adding robustness checks, correcting critical interpretation errors, and restructuring the theoretical framework. We are confident the paper is now significantly improved.

 

Thank you again for your rigorous and fair review. We have learned a great deal from your comments.

Reviewer 4 Report

Comments and Suggestions for Authors

This study analyzes how green technology innovation affects China's low-carbon economic transition through its non-linear relationship with green technology innovation based on provincial data from 2009 to 2023. The research utilizes econometric models which include quadratic components together with nonparametric additive models to investigate non-linear effects while studying how energy consumption patterns and industrial composition function as mediators. The research topic shows current relevance and the research methods used in the study are appropriate yet major methodological problems together with presentation issues and theoretical framework gaps need complete revision work before the manuscript can achieve publication standards required by Sustainability.

• The introduction and literature review provide a descriptive overview but they do not explain the research gap which the study addresses together with its unique contribution. The literature review cites several studies which already demonstrate the existence of nonlinear relationships together with their mediating mechanisms (Xu et al. 2024; Qu et al. 2021). The authors should explicitly state what this study adds beyond existing work. The study needs to establish its main research focus through three different research elements which include nonparametric methods and carbon performance and emissions and regional heterogeneity. The manuscript requires clarification of this point from the beginning.
• Table 1 defines it as GDP/CO₂ emissions. However, in the descriptive statistics (Table 2), the variable ranges from 0.048 to 2.920. Given that GDP is typically measured in billions or trillions, this ratio seems implausibly low. The authors should verify the units and calculations.
• Equations (3) through (6) are presented without sufficient explanation of how the nonparametric functions fk​are estimated. Which smoothing method was used? How were bandwidths selected? The inclusion of both linear and nonparametric terms for the same variables raises concerns about identification and interpretation.
• The authors state they follow Jiang (2022) but present only one equation (7) for the mediator. A proper mediation test typically requires multiple steps: (a) effect of independent variable on mediator, (b) effect of independent variable on outcome, and (c) effect of mediator on outcome controlling for the independent variable. The current approach is incomplete.
• The models do not handle possible endogeneity which includes reverse causality and missing variable effects. Instrumental variable methods or robustness checks using lagged variables would strengthen the analysis.
• The column headings and row labels in Table 3 show improper alignment because The table uses two different variables that require different notation which needs to be corrected. The notation "lnIngr" should be used because "InIngr" was used to define the variable. The document requires uniformity.
• The table 3 shows that The quadratic term (InIngr)² is missing from columns (1) and (3) but the discussion references it. The presentation should clearly show which models include the quadratic term.
• The structure of Table 5 creates confusion for readers. The authors present four columns but appear to report results for two different dependent variables (energy consumption structure and industrial structure) across two model specifications. The content needs to be reorganized to achieve better clarity.
• No figures are included in the provided file. Given the nonlinear relationships discussed, graphical representation of the inverted U-shaped and U-shaped relationships would significantly improve readability and impact.
• The analysis of heterogeneity divides regions into three geographic parts which include eastern and central and western regions together with their different resource availability through two categories which are resource-based and non-resource-based. The authors present a common method yet they fail to explain through theoretical methods how these particular distinctions affect the connection between green innovation and carbon outcomes. The subgroup analyses require a more powerful theoretical framework to achieve better credibility.
• The manuscript contains numerous grammatical errors, awkward phrasing, and unclear sentences that detract from the scholarly quality. 

Author Response

Dear Reviewer,

Thank you very much for your detailed and constructive review. Your comments on methodological clarity, presentation, and theoretical framing are very helpful. We have carefully addressed each point below.

 

Comment 1: The introduction and literature review do not clearly state the research gap and unique contribution.

Response: As also noted by Reviewer 3, we have substantially rewritten the end of Section 1 and Section 2 to clearly articulate our research gap. We now explicitly state that while prior studies have examined nonlinearity, mediation, or heterogeneity in isolation, our contribution is to integrate all three within a unified framework using both carbon emissions and carbon performance as dual outcome variables, based on the most recent data (2009-2023) that captures China's post-dual-carbon pledge period.

 

Comment 2: *Table 1 defines carbon performance as GDP/CO₂, but Table 2 shows implausibly low values (0.048 to 2.920).*

Response: Thank you for catching this. The issue was a unit inconsistency. GDP was originally in RMB, but CO₂ was in tons. The ratio thus represents 10,000 RMB per ton of CO₂. The range (0.048 to 2.920) is actually plausible given that some western provinces have very low carbon productivity. However, to avoid confusion, we have now re-scaled the variable to GDP (in 10,000 RMB) per ton of CO₂ and revised the descriptive statistics accordingly. The range is now 0.48 to 29.20, which is clearer. We have also added a footnote in Table 1 explaining the unit.

 

Comment 3: *Equations (3)-(6) lack sufficient explanation of nonparametric estimation.*

Response: As noted in responses to Reviewers 1 and 3, we have removed the nonparametric additive model from the manuscript entirely. It was not fully operationalized, and its inclusion caused confusion. We now rely solely on the quadratic parametric model, which is standard and transparent. We apologize for this.

 

Comment 4: Mediation test is incomplete (only one equation).

Response: We have now implemented a full mediation test following Baron & Kenny's three-step approach, supplemented by a bootstrap test (1,000 replications). The revised Section 5.3 now presents:

Step 1: Effect of lnIngr on outcome (lnCO2, GCO2) – Table 3.

Step 2: Effect of lnIngr on mediators (ES, SERV) – Table 5, columns 1 & 3.

Step 3: Effect of both lnIngr and mediator on outcome – new Table 5 (revised columns 2 & 4). The bootstrap indirect effects are reported in the text. This is now complete.

 

Comment 5: Models do not handle possible endogeneity.

Response: As noted, we have added SYS-GMM and IV-2SLS robustness checks (new Table 8). We discuss these in Section 5.2.

Comment 6: Table 3 has alignment issues and inconsistent notation ("InIngr" vs "lnIngr").

Response: We have corrected all notation to "lnIngr" (lowercase L, not uppercase I). We have also reformatted Table 3 for proper alignment.

 

Comment 7: Table 3 shows quadratic term missing from columns (1) and (3) but discussion references it.

Response: We have clarified in the table notes and in the text that columns (1) and (3) are linear models (without quadratic term), while columns (2) and (4) include the quadratic term. This is now clearly stated.

 

Comment 8: Table 5 structure is confusing (four columns for two DVs).

Response: We have reorganized Table 5. It now has two panels: Panel A for Energy Structure (ES) and Panel B for Industrial Structure (SERV). Each panel has two columns: linear model and quadratic model. This is much clearer.

 

Comment 9: No figures are included.

Response: We have now added two figures: Figure 1 (conceptual framework) and Figure 2 (nonlinear plots based on our estimated coefficients).

 

Comment 10: Heterogeneity analysis lacks theoretical justification for the divisions.

Response: In the revised Section 5.4, we have added explicit theoretical justifications. For geographic heterogeneity, we cite the "dual economy" literature in China. For resource-based vs. non-resource-based, we cite the "resource curse" theory and argue that resource-dependent regions face higher sunk costs in traditional energy, making green innovation less effective.

 

Comment 11: Numerous grammatical errors and unclear sentences.

Response: The entire manuscript has undergone professional English language editing. We have corrected grammar, improved sentence structure, and standardized terminology (e.g., "cleaning effect" is now "clean effect"). We believe the language is now publication-ready.

 

Thank you again for your thorough and constructive review. Your comments have forced us to significantly improve the clarity, rigor, and honesty of our manuscript. We are very grateful.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have satisfactorily addressed all the observations provided. The manuscript has been carefully revised, and all concerns have been appropriately resolved. Therefore, the publication is now ready for acceptance.

     

Author Response

Dear Reviewer,

We would like to express our sincere thanks to you for your time, effort and valuable comments in reviewing our manuscript. We highly appreciate your positive evaluation and confirmation that all the raised concerns have been fully and appropriately resolved. We have carefully revised the manuscript according to your suggestions, and we are very pleased that the revised version meets the journal’s requirements for acceptance. We sincerely thank you again for your kind support for our publication.

Thank you again for your time and expertise.

Sincerely,

The Authors

Reviewer 2 Report

Comments and Suggestions for Authors

After reviewing the new version of the submitted manuscript, I see that all the suggested improvements have been successfully incorporated, so I consider this version suitable for publication.

Author Response

Dear Reviewer,

We greatly appreciate your careful review and constructive suggestions on our work. Thank you very much for confirming that all the recommended improvements have been successfully incorporated into the revised manuscript and for your approval of the current version for publication. We have carefully polished and improved the manuscript accordingly. We are truly grateful for your recognition and support, and we hope the manuscript can be accepted for official publication.

Thank you again for your time and expertise.

Sincerely,

The Authors

Reviewer 3 Report

Comments and Suggestions for Authors

Detailed comments are listed in a separate file. 

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

We sincerely thank you for your detailed and constructive comments on our manuscript (sustainability-4239831). These comments have greatly helped us improve the quality of our paper. Below we respond to each point item by item, and we have revised the manuscript accordingly.

 

Comment 1: The paper still includes a nonparametric model, but no corresponding results are presented or discussed. This part remains disconnected from the empirical analysis and should either be removed or properly implemented.

Response: We fully agree with the reviewer. The nonparametric model was originally included but we did not run the corresponding estimations. To avoid any confusion and maintain the coherence of the manuscript, we have removed equations (3)–(6) and all related text from Section 4.3. The mediation analysis now relies solely on the parametric quadratic model and the approach of Jiang (2022). We have revised Section 4.3 accordingly (see pages 7–8 of the revised manuscript).

 

Comment 2: The manuscript still does not adequately address endogeneity. Although the text mentions possible reverse causality and even refers to IV/GMM attempts in the limitations section, such results are not actually shown in the reported analysis.

Response: This is a very important point, and we apologize for not being clearer. We acknowledge that our current analysis cannot fully resolve endogeneity. We attempted to use instrumental variables (lagged GTI and neighboring regions’ GTI) and system GMM, but the results were not sufficiently robust (e.g., weak instruments or failed over-identification tests). Given the scope of this revision, we cannot rerun the analysis with a new identification strategy. Therefore, we have strengthened our limitations section (Section 6.3) to state clearly that our findings are associational, not strictly causal. We also suggest specific quasi-natural experiments (e.g., low-carbon city pilot policies) for future causal identification. We hope the reviewer accepts this honest acknowledgment.

 

Comment 3: The interpretation of the nonlinear relationships is still incomplete. The paper does not report the turning points or show whether they lie within the observed range of the data.

Response: Thank you for this crucial suggestion. We have now calculated and reported the turning points for both the inverted-U (carbon emissions) and U-shaped (carbon performance) relationships in Section 5.1. Specifically:

For carbon emissions: turning point at lnIngr ≈ 15.14, which lies above the sample maximum (10.92), indicating that emissions are still rising within the observed range.

For carbon performance: turning point at lnIngr ≈ 4.21, which lies within the sample range (3.045–10.92), confirming both decreasing and increasing segments are observed.

These calculations have been added right after Table 3.

 

Comment 4: There is still a serious problem in the mechanism analysis. Energy structure is measured as the share of coal in total energy consumption. Therefore, a positive coefficient means higher coal dependence, not optimization. However, the manuscript still interprets this result as improvement in energy structure.

Response: The reviewer is absolutely correct, and we apologize for this serious interpretive error. We have rewritten the relevant paragraph in Section 5.3 to correctly interpret the coefficients. Specifically, we now explain that the positive linear coefficient (0.026) indicates that, on average, GTI is associated with increased coal dependence—a counterintuitive but possible result due to rebound effects. However, the significant quadratic term reveals a U-shaped pattern: at low GTI levels, coal share rises; beyond a threshold, it falls. We have also revised the hypothesis testing statements accordingly. We thank the reviewer for catching this mistake.

 

Comment 5: The heterogeneity analysis still contains an interpretation error. For the central region, the reported coefficients for carbon performance indicate a U-shaped relationship, but the text describes it as inverted-U.

Response: This is another clear error on our part. We have corrected the text in Section 5.4.1 (Geographic heterogeneity). The correct description is now: “In the central region, the impact of green technology innovation on carbon performance shows a significant U-shaped relationship (coefficient of lnIngr = -0.749, p<0.01; coefficient of (lnIngr)² = 0.066, p<0.01).” We apologize for this oversight.

 

Comment 6: *Some consistency and presentation issues also remain. The data are provincial, but the equations still say that "i represents the city." Missing-data interpolation is mentioned, but its extent and method are not explained. Table 7 still presents sample sizes in decimal form.*

Response: We have addressed all these issues:

Changed “i represents the city” to “i represents the province” in equations (1)–(3) and the accompanying text.

Added a clear explanation of missing data handling in Section 4.2: “For a small number of missing values (less than 2% of the full sample), linear interpolation was applied separately for each province.”

Corrected the sample sizes in Table 7 from “150.000” and “300.000” to “150” and “300.”

 

Comment 7: The policy implications are still broader than the evidence supports. Since the study uses Chinese provincial data, the discussion should focus mainly on China rather than repeatedly generalizing to governments worldwide.

Response: We accept this comment and have substantially revised Section 6.2 (Policy Enlightenment). We now focus the recommendations on China’s low-carbon transition, while noting that other developing economies may find them informative. The phrases “governments around the world” have been replaced with “Chinese governments at all levels” or similar. We have also kept five policy recommendations but now ground them explicitly in China’s policy context (e.g., “dual-carbon goals,” “low-carbon city pilot policies”).

 

overall: The revised manuscript is better than the original version, but several important methodological and interpretation problems remain unresolved... my recommendation is major revision.

Response: We deeply appreciate the reviewer’s continued engagement and the recognition of our improvements. We have now addressed every specific point raised, including removing the nonparametric model, adding turning points, correcting two major interpretation errors (energy structure and central region’s U-shape), improving consistency, and narrowing policy implications. For the remaining issue of endogeneity, we have been transparent about our limitations and suggested clear directions for future causal identification. We respectfully believe the manuscript has been substantially strengthened and hope it now meets the journal’s standards for publication.

Thank you again for your time and expertise.

Sincerely,

The Authors

Reviewer 4 Report

Comments and Suggestions for Authors

The manuscript is scientifically sound, well-positioned in the literature, and methodologically rigorous after revision. It is acceptable for publication.

Author Response

Dear Reviewer,

We sincerely thank you for your rigorous and professional review of our manuscript. We highly appreciate your positive comments that the revised manuscript is scientifically sound, well-positioned in the literature, and methodologically rigorous. We are very grateful for your confirmation that our manuscript is acceptable for publication. Thank you again for your valuable time and professional evaluation. We sincerely hope that the manuscript can be formally accepted and published in your journal.

Thank you again for your time and expertise.

Sincerely,

The Authors

Round 3

Reviewer 3 Report

Comments and Suggestions for Authors

The revised manuscript has improved substantially and most major concerns have been addressed. I would recommend minor revision only, with attention to the following small edits before acceptance:

1. Language polishing is still needed in a few places, particularly for grammar, article use, and awkward phrasing (e.g., Sections 5.3, 5.4, and 6.3 contain several long sentences that could be streamlined).

2. Consistency in notation and formatting should be checked throughout the manuscript, especially for variables (e.g., lnCO₂/GCO₂ capitalization, italics, subscripts, and consistency between text and tables).

3. Tables 6 and 7 should be carefully proofread, as a few coefficient patterns appear repeated or potentially duplicated (for example Urban and FDI entries), which may be typographical carryovers.

4. Some interpretations in the heterogeneity discussion remain slightly repetitive and occasionally contradictory (e.g., description of central-region effects around lines 320–325) and should be clarified.

5. References need final formatting cleanup, including punctuation inconsistencies, spacing, and some journal citation style irregularities.

Overall, the revision satisfactorily addresses previous concerns, and after these minor editorial corrections, the manuscript would be suitable for publication.

Author Response

Dear Reviewer,

Thank you very much for your professional, careful, and constructive comments on our revised manuscript. We highly appreciate your time and efforts to help us improve the quality of this paper. We have carefully revised the manuscript according to all your suggestions, and the detailed responses are as follows:

 

Comment 1: Language polishing

Response: We have thoroughly polished the English expression, especially in Sections 5.3, 5.4, and 6.3. We streamlined long and complex sentences, corrected grammatical errors, article misuse, and awkward phrasing to improve readability and academic standard.

 

Comment 2: Consistency of symbols and formatting

Response: We have checked and standardized all variable symbols in the text and tables (including lnCo2, Gco2, italics, subscripts, uppercase letters) to ensure complete consistency between the text and the tables.

 

Comment 3 : Proofreading of Tables 6 and 7

Response: We carefully proofread Tables 6 and 7, removed duplicated coefficient entries (e.g., Urban and FDI), and corrected typographical errors caused by copy-pasting. All coefficients and standard errors remain original empirical results.

 

Comment 4 : Clarification of heterogeneity interpretation

Response: We revised the repetitive and slightly contradictory descriptions in the heterogeneity analysis, especially the interpretation of the central-region effects. We streamlined the logic and made the conclusions clearer and more consistent.

 

Comment 5: Reference formatting

Response: We standardized all references, corrected punctuation, spacing, and citation style irregularities to meet the journal’s requirements.

 

We believe the manuscript has been significantly improved and is now suitable for publication. We sincerely hope the revised version meets your approval. If you have any further questions, please feel free to contact us.

Thank you again for your valuable guidance.

Sincerely,

The Authors

Author Response File: Author Response.pdf