Exploring the Impact of Government Subsidies on R&D Cost Behavior in the Chinese New Energy Vehicles Industry
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis study investigates the impact of government subsidies on R&D cost stickiness in China’s new energy vehicle (NEV) industry. Using financial data of A-share listed firms from 2007 to 2021, the authors demonstrate that subsidies significantly increase firms’ tendency to retain R&D resources during economic downturns. Through the ABJ and Banker models, the research validates the signaling effect and resource-supplementing role of subsidies, with heterogeneity across firm size and ownership. This study integrates cost stickiness theory with policy intervention, offering a novel perspective on the effectiveness of government support for corporate innovation. By employing dual models (ABJ and Banker) alongside Propensity Score Matching (PSM) methodology, the research strengthens the credibility of the findings. It’s recommended that this work demonstrates sufficient rigor and relevance to be suitable for publication in this journal with minor revisions.
- The research could further compare the ​relative importance of "signaling effects" vs. "resource effects" in driving R&D cost stickiness, or explore ​heterogeneous impacts of subsidy types on asymmetric cost behavior.
- The discussion should address potential unintended consequences (e.g., inefficient R&D scaling) by integrating TFP or patent citation data to evaluate innovation quality.
- The discussion could link findings to China’s “Dual Carbon” goals and NEV development plans, proposing actionable subsidy designs (e.g., phased subsidies, performance-based criteria).
Author Response
Dear Reviewer,
We would like to thank you for your critical and constructive comments for our manuscript entitled, “Government subsidies and R&D cost stickiness: the case of New Energy Vehicles in China”. We sincerely thank you for your constructive and insightful comments, which have greatly contributed to improving the quality and depth of our paper. We have revised the manuscript accordingly. Below, we summarize the major changes in response to your suggestions:
Comment 1: The research could further compare the ​relative importance of "signaling effects" vs. "resource effects" in driving R&D cost stickiness, or explore ​heterogeneous impacts of subsidy types on asymmetric cost behavior.
Response:
Regarding the distinction between signaling effects and resource effects, we have revised Hypotheses 2 and 3 to better reflect the integrated perspective (line 369-546). Using Banker et al.’s conditional cost model, we demonstrate how government subsidies act as signals of future market optimism, thereby raising managerial expectations and enhancing R&D cost stickiness. We hope this revision aligns with your suggestion and clarifies the theoretical mechanisms involved.
Comment 2: The discussion should address potential unintended consequences (e.g., inefficient R&D scaling) by integrating TFP or patent citation data to evaluate innovation quality.
Response:
Your comments prompted us to reflect more deeply on the substantive implications of our topic. In response, we have included a new analysis using green patents to measure innovation outcomes. This allows us to examine whether the positive association between subsidies and R&D cost stickiness also translates into meaningful innovation performance. We have incorporated this in our discussion of Hypothesis 3 and Table 6.
Comment 3: The discussion could link findings to China’s “Dual Carbon” goals and NEV development plans, proposing actionable subsidy designs (e.g., phased subsidies, performance-based criteria).
Response:
We expanded the discussion in the Introduction and Contributions sections to consider policy implications after the gradual phase-out of subsidies post-2022. Specifically, we highlight the potential risk of large firms monopolizing core technologies in the NEV sector and emphasize the importance of designing equitable and performance-based subsidy schemes to avoid such outcomes.
We again thank you for your valuable feedback, which has significantly improved the rigor and relevance of our research. Your suggestions have been truly inspiring and have greatly helped us improve the focus and clarity of our research. We have revised the manuscript accordingly and hope that our efforts meet your expectations.
Yours sincerely
All authors
Reviewer 2 Report
Comments and Suggestions for AuthorsBasic notes topic selection journal link:
The article presents the results of research on the topic of government subsidies and R&D cost rigidity: a case study of new energy vehicles in China - evaluation based on the selected statistical method. The topic concerns important areas of economics and sustainable development and statistical research for researchers, and partly in the subject of the journal and certainly deserves attention, in particular it concerns issues related to current problems related to the development expansion of electric vehicles.
Notes content of considerations:
However, both the objectives of the study (and hypotheses) are not well defined. In addition, the structure of the article is not fully transparent. The work is weakened by the lack of an innovative approach to the topic, i.e. the authors' contribution to the development of research - the authors made an unsuccessful attempt to assign a given method to the problem. The quality of the presented research is low and the research problem was not only incorrectly identified but also verified. Limited research sample - not allowing for conclusions and the research period considering that we have the year 2025 - therefore it would be appropriate to propose a forecast for the years 2025-2030. The analysis includes limited data in Shanghai and Shenzhen, - no information on how to obtain data for measurement. The selected research method is not innovative and the use of Anderson is limited by certain dependencies. Currently, completely different methods are used in the analysis aspect. I encourage the authors to familiarize themselves with the latest research topics covering this area. There is also a lack of description of the limitations of the adopted research scheme - which is currently a serious methodological and logical error. The authors forgot that for quantitative variables with a normal distribution, the Pearson correlation coefficient should be selected. If the data do not have a normal distribution or have ordered categories, then the Kendall tau-b or Spearman coefficients should be selected, which are a measure of relationships between ranks. .The work shows a lack of contribution from the authors, for example in the form of an original algorithm examining the adopted variables. There is no description of the scenarios and their limitations. The considerations presented below in Figure 17 are not applicable to the article. The clarity of the data in the tables is low. Again, as a form of expression, there are factual and logical errors in the content of the formulation of the Hypotheses. As in the content
All considerations are burdened with a basic logical and methodological error. First of all, they concern the past period, which raises questions about the purposefulness of the presented research. The whole is not based on the proposal of models and scenarios and requires significant corrections to meet the requirements for articles published in MDPI.
There are many areas that definitely require clarification, such as:
1. Title of the article - should be specified - because in its current form it is not precise and does not refer to the considerations presented in the article.
2. Abstract, should contain information about the context and background of the study, its purpose, procedures (selection of study participants, assumptions, measurements and methods, dates of research), key results and main conclusions and contribution to the current state of knowledge.
3. Keywords require re-editing.
4. Introduction - the introduction should clearly and concisely describe the background of the problem and the motives for undertaking the research. It should include a diagnosis of the current state in 2025. The aim of the work and possible research hypotheses for verification in the future should be verified and redefined correctly, the essence of the research problem and the purpose of the research should be indicated in the final part, the division into sections should be described.
5. The aim of the work and possible research hypotheses should be verified once again correctly.
6. The Literature Study chapter should be supplemented with the latest works on the issues raised in relation to both the cited research topic and the methods used.
7. The Research Method chapter should be divided into individual sections: method selection, tool description, measurements - limitations etc., justification etc. - (lack of justification for the tool used (what tools were used, what criteria were adopted, the description should be clear) for someone who is not an expert in this field of statistical research - especially since the currently used method, as already mentioned - does not allow for answering questions without limitations).
8. Assumptions for the selected method: Pearson correlation assumes linearity and normality, while Kendall and Spearman correlations assume fewer assumptions regarding the distribution of data. Data type: Pearson is best suited for continuous data, while Kendall and Spearman can handle ordinal (ranked) data - as is the case here. So there is a logical error in the selection of the method from the outset.
9. The correctness of the adopted formulas for calculations and their description should be checked again - for the new purpose of the work. In this case, they can be helpful in comparison with the forecast for the years 2025-2030.
10. The authors' contribution to the work may be the proposal and use of a forecasting method that allows for estimating the future based on historical data and comparing it with current results. Therefore, the research should be supplemented by presenting a forecast for the years 2025-2050.
11. The tables presented in the results section without a description and interpretation of the data contained have no scientific value. It is important to emphasize the limitations of the adopted methods at this stage. The summary does not refer to other researchers and the limitations of the presented research.
12. The literature must be supplemented with the latest items related to the cited research topic.
The work requires language correction for this academic.
Author Response
Dear Reviewer,
We would like to thank you for your critical and constructive comments for our manuscript entitled, “Government subsidies and R&D cost stickiness: the case of New Energy Vehicles in China”. We've carefully considered the comments and found your suggestions enlightening to our understanding of public policy and cost behavior. In revised manuscript, all the changes are highlighted in red for easy inspection. The revisions were addressed point by point below:
Comment 1: Title of the article - should be specified - because in its current form it is not precise and does not refer to the considerations presented in the article.
Response:
Thank you for your valuable comment. We agree that the original title was too broad and did not accurately reflect the core focus of the study. We have revised the title to more precisely capture the paper’s contribution:
New Title: Exploring the Impact of Government Subsidies on R&D Cost Behavior in the Chinese New Energy Vehicles Industry
Comment 2: Abstract, should contain information about the context and background of the study, its purpose, procedures (selection of study participants, assumptions, measurements and methods, dates of research), key results and main conclusions and contribution to the current state of knowledge.
Response:
We appreciate this suggestion. In response, we have thoroughly revised the abstract to clearly state the research problem, methodology, key findings, and theoretical as well as policy implications. The revised abstract now better conforms to academic norms and provides a more accurate summary of our work.
Comment 3: Keywords require re-editing.
Response:
Thank you for pointing this out. We have updated the keywords to more precisely reflect the main themes and concepts explored in the paper. The new set of keywords is:
Keywords:Government subsidies; R&D cost stickiness; New energy vehicle (NEV) industry; Asymmetric cost behavior; Green innovation; Managerial expectations; China
Comment 4: Introduction - the introduction should clearly and concisely describe the background of the problem and the motives for undertaking the research. It should include a diagnosis of the current state in 2025. The aim of the work and possible research hypotheses for verification in the future should be verified and redefined correctly, the essence of the research problem and the purpose of the research should be indicated in the final part, the division into sections should be described.
Response:
We fully agree with your insightful comment. Your suggestion encouraged us to reflect more deeply on the practical implications of our study. In the revised manuscript, particularly in Section 1.7 (lines 309–314), we have added a forward-looking discussion addressing the potential risk of technology monopolization by large firms in the post-subsidy era. We also highlight that our findings provide early signals for policy adjustments in light of the subsidy phase-out beginning in 2022. While we acknowledge that our predictive contribution is still limited, we hope to further enhance this in future research with more empirical evidence.
Comment 5: The aim of the work and possible research hypotheses should be verified once again correctly.
Response:
We have thoroughly revised the theoretical framework and all hypotheses (Sections 2.2–2.4) to enhance logical clarity and theoretical rigor.
Comment 6: The Literature Study chapter should be supplemented with the latest works on the issues raised in relation to both the cited research topic and the methods used.
Response:
We have updated the literature review to incorporate recent studies from the past five years, reflecting the importance of aligning research with the latest academic developments.
Comment 7: The Research Method chapter should be divided into individual sections: method selection, tool description, measurements - limitations etc., justification etc. - (lack of justification for the tool used (what tools were used, what criteria were adopted, the description should be clear) for someone who is not an expert in this field of statistical research - especially since the currently used method, as already mentioned - does not allow for answering questions without limitations).
Response:
In response to concerns about incomplete model descriptions, we have expanded Section 3.3 to provide clearer definitions of variables and added detailed explanations for each model and its corresponding results.
Comment 8: Assumptions for the selected method: Pearson correlation assumes linearity and normality, while Kendall and Spearman correlations assume fewer assumptions regarding the distribution of data. Data type: Pearson is best suited for continuous data, while Kendall and Spearman can handle ordinal (ranked) data - as is the case here. So there is a logical error in the selection of the method from the outset.
Response:
Thank you for your valuable critique regarding correlation analysis. We re-examined our variables and confirm that they are continuous and approximately normally distributed. Based on this, and consistent with prior studies such as Golden et al. (2020), we maintain that Pearson correlation is appropriate for our analysis. We have clarified this choice in the methodology section.
Comment 9: The correctness of the adopted formulas for calculations and their description should be checked again - for the new purpose of the work. In this case, they can be helpful in comparison with the forecast for the years 2025-2030.
Response:
Your comment on predictive analysis is highly appreciated. We agree that the lack of forward-looking insights may weaken the study’s contribution. To address this, we added a new hypothesis on the moderating role of green innovation to enhance the study’s relevance and foresight. We also acknowledged this limitation explicitly in the conclusion.
Comment 10: The authors' contribution to the work may be the proposal and use of a forecasting method that allows for estimating the future based on historical data and comparing it with current results. Therefore, the research should be supplemented by presenting a forecast for the years 2025-2050.
Response:
We further examined heterogeneity and found that large and private firms benefit more from subsidies, which suggests important policy implications for subsidy phase-out strategies and the need to avoid biased support toward dominant firms.
Comment 11: The tables presented in the results section without a description and interpretation of the data contained have no scientific value. It is important to emphasize the limitations of the adopted methods at this stage. The summary does not refer to other researchers and the limitations of the presented research.
Response:
We revised our interpretation of key findings and included a clear discussion of the limitations regarding causal inference in the conclusion.
Comment 12: The literature must be supplemented with the latest items related to the cited research topic.
Response:
Finally, we have incorporated recent publications from 2023–2024 to ensure the literature foundation is current and robust.
We sincerely appreciate your valuable suggestions. Your comments prompted us to deeply reflect on the logical structure and limitations of our model, which in turn allowed us to significantly revise and refine the manuscript. This has been an extremely helpful process, and we are truly grateful for your guidance. We hope that the improvements we have made will meet your expectations and help enhance the quality of our work. We look forward to your further feedback. Thank you again for your comments and suggestions.
Yours sincerely
All authors
Reviewer 3 Report
Comments and Suggestions for Authors- Clearly position the study within a defined theoretical gap. Explain whether the study extends, tests, or challenges existing theories. Consider developing a conceptual framework linking subsidies, managerial behavior, and cost stickiness more rigorously.
-
Reframe the hypotheses around gaps or controversies in the literature (e.g., crowding out vs. crowding in effects; behavioral additionality). Include citations to show why these specific relationships require empirical testing.
- Acknowledge the limits of causal inference more explicitly. Consider an instrumental variable (IV) approach, a difference-in-differences design (if panel data permits), or at least robustness checks using lagged performance or fixed effects that isolate time-invariant bias.
- Incorporate a stronger theoretical foundation using Spence’s signaling model or relevant corporate finance literature. If empirical testing is limited, clarify that the signaling interpretation is theoretical and exploratory. Alternatively, use survey or text analysis data (e.g., earnings calls, MD&A) in future work to proxy for perceived signals.
- Revise wording throughout to reflect the correlational nature of findings, unless additional causal identification is provided. Use cautious language such as “associated with,” “related to,” or “suggests.”
- Substantially shorten and tighten the introduction and theoretical sections. Use subheadings and bullet points in long sections. Consider professional language editing.
- Draw on theories of firm capabilities, resource constraints, or absorptive capacity to explain these differences. Provide policy implications tailored to firm heterogeneity.
- Section 4.3 (p.15–17) lists robustness tests, but with minimal explanation of what concern each test addresses.
- Table 3 shows significant results for interaction terms, but the manuscript does not clarify how these coefficients should be interpreted in terms of managerial behavior.
- Firms receiving subsidies may systematically differ in innovation capability, lobbying effort, or connections none of which are directly controlled for in the models.
- The finding that large and non state firms respond more positively to subsidies is asserted, but no structural explanation is offered.
- Table 1 reports summary statistics without deeper insight into distributions or potential biases in the sample.
- Adjusted R² values are reported but not interpreted. For example, several models show R² of ~0.15–0.27, suggesting modest explanatory power.
- The paper mentions low VIF scores but does not report them. More importantly, no residual analysis is conducted to verify homoscedasticity, normality, or influential points.
- Clarify the handling of zero or near-zero subsidy values
- Report pre- and post-matching covariate balance (e.g., standardized mean differences, t-tests), include the matching caliper, and show a density plot of propensity scores by treatment status.
- The authors state that “subsidies increase stickiness” but do not ask should they? Is increased stickiness always beneficial, or can it signal wasteful persistence?
- Tables 6–8 present disjointed results without a framework.
Comments on the Quality of English Language
There are several instances of awkward sentence structures, improper use of academic terminology, and repetitive phrasing, particularly in the introduction and theoretical background sections.
Author Response
Dear Reviewer,
We would like to thank you for your critical and constructive comments for our manuscript entitled, “Government subsidies and R&D cost stickiness: the case of New Energy Vehicles in China”. We sincerely thank you for your rigorous and constructive feedback, which has been highly instructive and valuable for improving both this paper and our future research. We have made substantial efforts to revise the manuscript based on your suggestions. While we acknowledge that the paper may still have room for improvement, we truly hope that our revisions demonstrate meaningful progress. In revised manuscript, all the changes are highlighted in red for easy inspection. The revisions were addressed point by point below:
Comment 1: Clearly position the study within a defined theoretical gap. Explain whether the study extends, tests, or challenges existing theories. Consider developing a conceptual framework linking subsidies, managerial behavior, and cost stickiness more rigorously.
Response:
Thank you for your valuable suggestion. In response, we have reconstructed the theoretical framing of the paper to highlight our specific marginal contribution. Accordingly, the Introduction and Theory sections have been substantially revised to clarify the research gap and position our study within the existing literature.
Comment 2: Reframe the hypotheses around gaps or controversies in the literature (e.g., crowding out vs. crowding in effects; behavioral additionality). Include citations to show why these specific relationships require empirical testing.
Response:
We have revised the literature review, Theory 1, and the corresponding hypothesis. Specifically, the hypothesis concerning the relationship between subsidies and cost stickiness has been rewritten to emphasize the mechanism of profit pressure induced by insufficient subsidy support.
Comment 3: Acknowledge the limits of causal inference more explicitly. Consider an instrumental variable (IV) approach, a difference-in-differences design (if panel data permits), or at least robustness checks using lagged performance or fixed effects that isolate time-invariant bias.
Response:
We appreciate your constructive suggestion regarding causal identification. We attempted a Difference-in-Differences (DID) approach. However, due to the inclusion of multiple interaction terms, the estimation results were not stable. We recognize the importance of robust identification strategies and will continue exploring more suitable causal inference methods in future research.
Comment 4: Incorporate a stronger theoretical foundation using Spence’s signaling model or relevant corporate finance literature. If empirical testing is limited, clarify that the signaling interpretation is theoretical and exploratory. Alternatively, use survey or text analysis data (e.g., earnings calls, MD&A) in future work to proxy for perceived signals.
Response:
In Hypothesis 2, we have restructured the theoretical discussion and hypothesis to better reflect the signaling perspective. The previous version did not clearly articulate the signaling effect. We have revised this section by drawing on Spence’s signaling theory to explain how subsidies may influence managerial expectations and result in cost stickiness.
Comment 5: Revise wording throughout to reflect the correlational nature of findings, unless additional causal identification is provided. Use cautious language such as “associated with,” “related to,” or “suggests.”
Response:
We carefully reviewed and revised relevant expressions and descriptions throughout the manuscript to improve clarity and academic tone.
Comment 6: Substantially shorten and tighten the introduction and theoretical sections. Use subheadings and bullet points in long sections. Consider professional language editing.
Response:
The overly lengthy Introduction section has been reorganized using subheadings. We rewrote this section to improve structure and readability.
Comment 7: Draw on theories of firm capabilities, resource constraints, or absorptive capacity to explain these differences. Provide policy implications tailored to firm heterogeneity.
Response:
Thank you for your insightful comment. To address this, we added Hypothesis 3, which examines the moderating role of green innovation capability in the relationship between subsidies and R&D cost stickiness. We believe this addition highlights the practical implications of our findings in the context of environmental and innovation policy.
Comment 8: Section 4.3 (p.15–17) lists robustness tests, but with minimal explanation of what concern each test addresses.
Response:
We appreciate your suggestion. We have revised Section 3 by expanding our discussion on robustness checks. This includes additional tests and explanations to ensure the reliability and consistency of our empirical results.
Comment 9: Table 3 shows significant results for interaction terms, but the manuscript does not clarify how these coefficients should be interpreted in terms of managerial behavior.
Response:
Thank you for your helpful comment. We have added a detailed explanation of the model associated with Table 3 (see lines 769–774) and provided a separate section for variable definitions and measurement (Section 3.3, lines 593–618) to enhance clarity and transparency.
Comment 10: Firms receiving subsidies may systematically differ in innovation capability, lobbying effort, or connections none of which are directly controlled for in the models.
Response:
Thank you for raising this point. We have expanded our discussion of Hypothesis 3 (lines 496–542) and provided empirical evidence that the positive effect of subsidies on R&D cost stickiness is more pronounced in firms with higher levels of green innovation. This finding reinforces the importance of considering firm-level innovation capability when evaluating policy impacts.
Comment 11: The finding that large and non state firms respond more positively to subsidies is asserted, but no structural explanation is offered.
Response:
We agree with your suggestion and have reorganized and clarified the discussion on heterogeneity. This includes examining firm characteristics such as ownership structure, size, and green innovation level, to better understand the differential effects of subsidies on R&D cost stickiness across subsamples.
Comment 12: Table 1 reports summary statistics without deeper insight into distributions or potential biases in the sample.
Response:
We appreciate the reviewer’s suggestion. While we acknowledge that further exploration of the distributional characteristics and potential sample biases could provide additional context, we believe that the current presentation of summary statistics in Table 1, together with the robustness checks and matching methods (e.g., PSM) employed later in the analysis, sufficiently address concerns regarding sample representativeness and bias. To maintain focus and avoid redundancy, we have opted not to expand this section further, but we are happy to revisit this point should the editor deem it necessary.
Comment 13: Adjusted R² values are reported but not interpreted. For example, several models show R² of ~0.15–0.27, suggesting modest explanatory power.
Response:
Thank you for pointing this out. We acknowledge that the R² value is relatively low. However, after reviewing a number of relevant studies, we found that this is common in R&D cost stickiness research. One reason is that R&D expenditures typically account for a much smaller proportion of sales revenue compared to operating costs. Another reason is that the inclusion of multiple interaction terms in our model can lead to a reduction in the R². For example, Chang et al. (2020) report an R² of 0.024 in their stickiness model, which aligns with our findings.
Comment 14: The paper mentions low VIF scores but does not report them. More importantly, no residual analysis is conducted to verify homoscedasticity, normality, or influential points.
Response:
Thank you for the suggestion. We conducted a multicollinearity test and added the Variance Inflation Factor (VIF) results as Table 2-2 (line 759). The results confirm that multicollinearity is not a concern in our model.
Comment 15: Clarify the handling of zero or near-zero subsidy values
Response:
Thank you for this helpful observation. We have revised Section 3.3 (line 595) to include a detailed explanation of how we handle missing values for subsidy variables. Importantly, we did not treat missing values as zeros. This clarification has been added to ensure transparency in variable construction.
Comment 16: Report pre- and post-matching covariate balance (e.g., standardized mean differences, t-tests), include the matching caliper, and show a density plot of propensity scores by treatment status.
Response:
Thank you for your comment. We have expanded our description of the Propensity Score Matching (PSM) procedure in Section 4.3.2 and added a new table (Table 12) to provide supporting details on the matching results.
Comment 17: The authors state that “subsidies increase stickiness” but do not ask should they? Is increased stickiness always beneficial, or can it signal wasteful persistence?
Response:
We appreciate this constructive suggestion. Based on your insight, we developed Hypothesis 3, which explores the moderating role of firm-level green innovation capability. We show that the complementary effect of subsidies on R&D cost stickiness is more pronounced in firms with high innovation levels. This addition strengthens the theoretical framework and empirical contribution of the paper.
Comment 18: Tables 6–8 present disjointed results without a framework.
Response:
Thank you for your valuable suggestion. We have reorganized the section on additional hypotheses to present a more coherent and logically structured narrative. We also apologize for the previously unclear presentation.
Once again, we greatly appreciate your time and thoughtful comments, and we hope that the revised version better aligns with your expectations.
Yours sincerely
All authors
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsAfter proofreading, the reviewer does not make any comments to the text.
I congratulate the authors on their idea and wish them creative continuation of research in this field.
English requires minor corrections.
Reviewer 3 Report
Comments and Suggestions for Authorsthe author already address all my concern