Manufacturing Agglomeration and Corporate Environmental, Social, and Governance Performance

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. The data sources are unclear. Clear data sources ensure the scientific validity of the research. Please provide the data sources.
2. The regression results in Tables 5, 6, and 7 exhibit sample size losses. The impact of these lost samples on the regression results must be considered. Please explain this issue.
3. The discussion section is essential. It enhances the research value of the manuscript.
4. The final section requires the addition of research limitations and future prospects.

Author Response

Point to point response

Comment 1: The data sources are unclear. Clear data sources ensure the scientific validity of the research. Please provide the data sources.
Response: We thank the reviewer for this important suggestion. To clarify the data sources and variable construction, we have added a new table (Table 1 on Page 10) detailing all variables used in the baseline and subsequent regressions. This table includes descriptions of the variables, their measurement/processing methods, and their specific data sources.

Comment 2: The regression results in Tables 5, 6, and 7 exhibit sample size losses. The impact of these lost samples on the regression results must be considered. Please explain this issue.
Response: We appreciate the reviewer for noting this. The sample size differences across tables are due to the specific requirements of each analysis:

• For Table 5 (now Table 6), columns (2), (4), and (6), the sample size is reduced because these regressions use a sample of manufacturing listed firms that did not publicly disclose ESG information for comparative analysis.
• For Table 6 (now Table 7), the baseline regression sample is split into two groups based on whether the firm belongs to a high-pollution industry. The total sample size remains largely unchanged when considering the groups sequentially.
• For Table 7 (now Table 8), columns (1), (2), and (5), the sample size decreases because these regressions incorporate compliance cost data (proxied by D&O insurance expenditure), which is separately disclosed by firms. Matching this data with other financial data resulted in the loss of some unmatched observations.
  We have added specific explanations for any sample size changes in the notes accompanying the respective empirical regression tables.

Comment 3: The discussion section is essential. It enhances the research value of the manuscript.
Response: We thank the reviewer for this valuable suggestion. Following your advice, we have replaced the original "Conclusion and Implications" section with a comprehensive "Discussion" section (Page 21). This new section includes a summary and discussion of the research hypotheses and empirical results, as well as more targeted policy implications derived from the findings.

Comment 4: The final section requires the addition of research limitations and future prospects.
Response: Thank you for this suggestion. As mentioned in the response to Comment 3, we have integrated the discussion of research limitations and future research directions into the new "Discussion" section (specifically, subsection 6.3 on Page 23).

Reviewer 2 Report

Comments and Suggestions for Authors

The paper examines how the agglomeration phenomenon influences environmental, social, and governance disclosures in listed Chinese companies for the period 2010-2023.

Overall, it is a well-conceived and well-written article with interesting contributions and a bibliography that, while not extensive (37 references), only 16% of which does not date from the last five years.

In my opinion, the paper is relevant, and I have no special suggestions for improvement beyond the following:

1) This is a case study of Chinese industries, which implies limited international generalization. My recommendation is to seek results from other geographic areas to issue comparative assessments and, thus, expand the bibliographic references. If the request cannot be fulfilled, I understand that this should be recognized as a limitation of the study.

2) Would it be possible to propose a more detailed public policy based on the findings?

3) In the conclusions and implications section, provide a summary of each of the three hypotheses proposed and their results (specifically mentioning each H), i.e., H1: agglomeration drives ESG disclosure. And the result is H2: the average productivity of agglomeration reduces the opportunity cost of disclosure.

4) I am not familiar with the formal requirements for the references section, but if it is APA, the journals should be in italics. I recommend reviewing this aspect.

I congratulate the authors on their work.

Author Response

Point to point response

Comment 1: This is a case study of Chinese industries, which implies limited international generalization. My recommendation is to seek results from other geographic areas to issue comparative assessments and, thus, expand the bibliographic references. If the request cannot be fulfilled, I understand that this should be recognized as a limitation of the study.
Response: We thank the reviewer for this pertinent observation. Indeed, due to significant differences in manufacturing scale and focus, studies using a manufacturing agglomeration lens to analyze its impact on corporate non-financial indicators (like ESG) are less common outside China. Following your recommendation, we have acknowledged this as a limitation of our study. This discussion is now included in the new "Discussion" section (subsection 6.3 on Page 23).

Comment 2: Would it be possible to propose a more detailed public policy based on the findings?
Response: We appreciate this suggestion. As part of the revision initiated in response to other reviewers' comment, we have thoroughly revised the policy implications section. It is now located within the new "Discussion" section (subsection 6.2 on Page 22). The policy implications have been refined to be more specific and directly aligned with the empirical results, enhancing their practical relevance.

Comment 3: In the conclusions and implications section, provide a summary of each of the three hypotheses proposed and their results (specifically mentioning each H), i.e., H1: agglomeration drives ESG disclosure. And the result is H2: the average productivity of agglomeration reduces the opportunity cost of disclosure.
Response: Thank you for this suggestion. The new "Discussion" section (subsection 6.1 on Page 21) now includes a structured summary of the three research hypotheses and their corresponding empirical results, explicitly referencing H1, H2, and H3.

Comment 4: I am not familiar with the formal requirements for the references section, but if it is APA, the journals should be in italics. I recommend reviewing this aspect. I congratulate the authors on their work.
Response: We thank the reviewer for the compliment and for pointing this out. We have reviewed and updated the reference list format according to the journal's requirements, ensuring journal names are italicized where appropriate.

Reviewer 3 Report

Comments and Suggestions for Authors

See the attached revision

Comments for author File: Comments.pdf

Author Response

Point to point response

Comment 1: The model specification and variable definitions need to be made explicit and internally consistent. Please introduce formal notation directly under the baseline equation... Ensure the same symbols appear in tables and text. Provide the exact formula for the location quotient... fully document construction of the "Environmental Regulation" index... Clarify the scale and transformation of "Local Government Debt"... specify units/transformations for all firm-level controls... Define the CSR measure... and the E/S/G sub-indices... introduce "RE" (firm profit) in the text before it appears in tables, and explain the dynamic term "L.ESG"... For the productivity mechanism, state the Olley-Pakes specification you estimate... so the TFP measure is reproducible.
Response: We are grateful to the reviewer for these detailed and crucial suggestions regarding methodological clarity and reproducibility. In response:

• We have created a comprehensive variable description table (Table 1 on Page 10) that lists the core variables used in the baseline and extended analyses. This table includes formal definitions, measurement methods, transformations (e.g., logarithms), and data sources.
• The construction details for the "Environmental Regulation" index (keyword dictionary, text processing) and the formula for the Location Quotient (LQ) are explicitly stated in this table.
• The variable "RE" (Firm Profit) is now defined in the text (Section 3.1) before its appearance in tables.
• The Olley-Pakes (OP) method specification for TFP calculation, including inputs, proxy variable, and deflators, is detailed in a footnote (Footnote 3).
• For variables used in specific extended analyses beyond the baseline, their measurement and sources are clarified in the footnotes of the corresponding tables.
  We believe these additions significantly enhance the transparency, internal consistency, and reproducibility of our empirical analysis.

Comment 2: The identification and estimation strategy requires targeted but feasible upgrades. Replace the contemporaneous peer variable with a leave-one-out, one-period-lagged peer ESG to mitigate simultaneity... In your system-GMM, collapse instruments and limit lag depth; report instrument count, Hansen/Sargan, and AR(1)/AR(2) diagnostics... For the IV results... report Kleibergen-Paap rk Wald F, Stock-Yogo critical values, and first-stage partial R²; include at least one placebo outcome... Recompute standard errors with two-way clustering (firm × year), and probe basic nonlinearity by adding a quadratic term...
Response: We thank the reviewer for these essential methodological recommendations. We have implemented the following upgrades:

• The peer effect variable has been recalculated using the leave-one-out, one-period-lagged peer ESG measure.
• For the System GMM estimation, we have collapsed instruments, limited lag depth, and reported the instrument count, Hansen test (p-value), and AR(1)/AR(2) test statistics in the corresponding table (Table 4).
• For the IV estimations, we now report the Kleibergen-Paap rk Wald F statistic, Stock-Yogo critical values (10% maximal IV size), and first-stage partial R² in Table 4.
• Standard errors throughout the baseline analysis (including Table 2) have been recomputed using two-way clustering at the firm and year levels.
• A quadratic term (Agglomeration^2) was added to the baseline regression (Table 2, Column 2) to test for basic nonlinearity. (The result was not significant, but the test was conducted as suggested).

Comment 3: Align terminology with measurement throughout. If the dependent variable is an ESG rating... avoid labeling it "ESG disclosure" and instead use "ESG rating" or "ESG performance," with a brief note acknowledging that ratings partially reflect disclosure but are not equivalent to it.
Response: We appreciate the reviewer for highlighting this important distinction. We have revised the terminology throughout the manuscript accordingly. We now consistently use "ESG performance" when referring to the rating/score (the dependent variable). We have also added a note in the model derivation section (Section 2, page 7) clarifying that ESG ratings reflect disclosure quality but encompass more than just disclosure itself.

Comment 4: Finally, reflect these changes in table notes and the methods section, and provide minimal replication materials (variable dictionary and code snippets for the peer variable, GMM settings, and IV diagnostics).
Response: Thank you for this suggestion. The changes mentioned above (variable definitions, methodological upgrades) are indeed reflected in the relevant table notes and the methods section. Furthermore, as part of our commitment to reproducibility, we have uploaded supplementary replication materials. These include the dataset used for the main analyses and basic Stata code snippets covering variable construction, GMM settings, and IV diagnostics.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Please write the explicit formula for the location quotient in the data section and use a single term consistently throughout (either “Location Quotient (LQ)” or “location entropy”). One line with the numerator/denominator and the reference population is sufficient; this will fully close the reproducibility gap around the agglomeration measure.

Please define the dynamic term “L.ESG” in the text or in the GMM table note as the one-period lag of ESG and, for completeness, state the sampling frequency relevant to that lag. This removes any ambiguity in the dynamic specification.

Please make explicit in the methods or in the GMM table note that instruments are collapsed and that lag depth is limited; specify the maximum and minimum lags used. Your current diagnostic values are acceptable, but the implementation details should be stated.

Please add one falsification/placebo outcome for the IV design—i.e., a firm outcome that should not be affected by the agglomeration instrument(s)—and report the null effect alongside your IV diagnostics. A succinct appendix table is adequate and will strengthen the exclusion restriction claim.

Please finish the terminology cleanup in the hypotheses block (and title/keywords if applicable) so that the dependent variable is consistently described as an ESG rating/performance. If you discuss “disclosure,” treat it explicitly as a related mechanism or construct distinct from the rating.

Author Response

Point-to-point Response

We thank the reviewer for these thoughtful and constructive comments. We have revised the manuscript accordingly, and our point-by-point responses are detailed below.

 

Reviewer Comment 1: Please write the explicit formula for the location quotient in the data section and use a single term consistently throughout (either "Location Quotient (LQ)" or "location entropy"). One line with the numerator/denominator and the reference population is sufficient; this will fully close the reproducibility gap around the agglomeration measure.

Response: We thank the reviewer for this suggestion. We have now explicitly defined the Location Quotient (LQ) formula in the Data section (Page 8). The term "Location Quotient (LQ)" is now used consistently throughout the manuscript, and the data sources have been specified in greater detail. The formula is presented as follows:
                                 

Where  denotes the manufacturing location quotient of region i,  represents the manufacturing employment in region i,  is the total employment in region i, and  is the ratio of national manufacturing employment to total national employment. The relevant employment statistics are sourced from the China City Statistical Yearbook and the China Statistical Yearbook.

 

Reviewer Comment 2: Please define the dynamic term "L.ESG" in the text or in the GMM table note as the one-period lag of ESG and, for completeness, state the sampling frequency relevant to that lag. This removes any ambiguity in the dynamic specification.

Response: We agree that this clarification is important. The term L.ESG is now explicitly defined in Section 4.3 (Page 12) as the one-period lag of the ESG performance variable. We have also specified that it is measured as the mean value at the end of the quarter of the following year, clarifying the sampling frequency.

 

Reviewer Comment 3: Please make explicit in the methods or in the GMM table note that instruments are collapsed and that lag depth is limited; specify the maximum and minimum lags used. Your current diagnostic values are acceptable, but the implementation details should be stated.

Response: We have revised Section 4.3 to explicitly state the details of the System GMM implementation（Page 12）. The text now clearly indicates that the instrument matrix was collapsed to avoid overfitting and that the lag depth was restricted, specifying that we used lags from t-2 to t-4 as instruments.

 

Reviewer Comment 4: Please add one falsification/placebo outcome for the IV design—i.e., a firm outcome that should not be affected by the agglomeration instrument(s)—and report the null effect alongside your IV diagnostics. A succinct appendix table is adequate and will strengthen the exclusion restriction claim.

Response: Following the reviewer's excellent suggestion, we have conducted a falsification test using a firm's registration location (a historical decision) as the placebo outcome. The results, which show a statistically insignificant effect of the instrumented manufacturing agglomeration on this outcome, are now reported and discussed in Section 4.3（Page 12-13）. The full results are presented in a succinct Appendix Table A1, strengthening our claim regarding the exclusion restriction.

Reviewer Comment 5: Please finish the terminology cleanup in the hypotheses block (and title/keywords if applicable) so that the dependent variable is consistently described as an ESG rating/performance. If you discuss “disclosure,” treat it explicitly as a related mechanism or construct distinct from the rating.

Response: We thank the reviewer for this important suggestion. We have refined the focus of the study to center consistently on ESG performance. The title has been revised to “Manufacturing Agglomeration and Corporate Environmental, Social, and Governance Performance.” Correspondingly, the Keyword, Hypothesis 1 and Hypothesis 2 have also been appropriately adjusted. Terminology of ESG “disclosure” are now strictly limited to the context of the first stage of the heterogeneous firm model, where it explicitly represents the binary decision of whether or not to disclose an ESG report. This ensures terminological consistency throughout the paper.