Summary of Changes — Second Revision
Manuscript sustainability-4566524
"Corporate Strategic Orientation and ESG Performance: Financial Distress and Boundary Conditions in China's Evolving Green-Policy Environment"

All edits below are applied to both the clean and highlighted copies; new text is marked in red in the highlighted copies. New numbers are taken from the Stata logs archived in the replication package.

1. Table 5 labeling system (with linked count sentences)

- H2: "Measure-dependent association (O-Score only)"
- H3a: "Measure-dependent indirect association (3.9% share)"
- H4b: "Nominally supported; exploratory after FDR correction; not robust to inference level"
- H5b: "Nominally supported (linear spec); association concentrated in the amount margin"
- H7a: "Nominally supported (10% level)"; H8b: "Nominally supported (1% level)"
- The table note now defines "nominally supported" and "measure-dependent."
- The Section 4.7 and 4.8 count sentences were rewritten to match: two hypotheses supported outright (H1a, H1b), three measure-dependent (H2, H3a, H3b), four nominally supported (H4b, H5b, H7a, H8b), four certified null (H6a, H6b, H7b, H8a), two indeterminate (H4a, H5a). Corresponding statements in Sections 2.4, 4.2–4.5, 4.8, 5.1, and 5.2, and the abstract, Discussion, and Conclusion were updated in the same pass.

2. Table 4: two-part subsidy estimates

The subsidy column now reports the two-part estimates alongside the linear ones (receipt −0.450, p = 0.038; amount +3.674, p = 0.005; linear 2.404, p = 0.046), with a table note explaining the two margins.

3. Table S6 and Table S1 reporting

- Table S6 carries a new "BH q" column with the exact Benjamini–Hochberg q-values for the fifteen-test family; the two group-specific policy-effect tests are reported in the S6.5 text (q = 0.833 and 0.873). The sentence referring readers to the replication-package output log now points to the table.
- Table S1 is re-estimated on the intersection of the three sub-dimension coverage sets (N = 12,960 in all three models); the notes to S1, S5.5, and S5.6 describe the intersection.
- All "p = 0.000" entries in the supplementary tables now read "p < 0.001."

4. New analyses (Section S6.16 of the supplementary material)

(i) Non-overlapping distress index. An index of leverage and liquidity only (z of leverage minus z of the current and quick ratios; no earnings component). Its correlation with ROA is −0.37, against −0.72 for the O-Score. Under this index the orientation-to-distress link is no longer detectable (a-path: Profit −0.0019, p = 0.89; Growth 0.0154, p = 0.44), and both bootstrap intervals for the indirect effects contain zero. The distress channel is therefore described throughout as a measure-dependent association; in Figure 1 and in the description of this channel, the text uses "indirect association" rather than "mediation." In Text S5.3, the discussion of the non-overlapping index now references this analysis directly, so that the VIF diagnostics in S5.2 (which reflect only linear correlation among the regressors) are no longer the only line of response to the construct-overlap concern.
(ii) Pre-policy subgroup ATTs. Cohort-robust (not-yet-treated) subgroup ATTs for the low-carbon pilot: pre-policy profit-oriented −1.876 (SE 0.827, p = 0.023); pre-policy growth-oriented +3.911 (SE 1.720, p = 0.023); remainder +0.881 (SE 0.465, p = 0.058); pre-trend tests by group p = 0.273 and 0.446. These estimate group-average effects and are not used to support H4b.
(iii) Missing-pilot sensitivity. Excluding the 983 firm-years with missing pilot status leaves the interactions essentially unchanged (DID × Profit = 0.455, p = 0.043; N = 12,015); reported in the Table 3 note.
(iv) Broader sample. Re-including the 249 ST firm-years excluded at baseline that carry Bloomberg ESG scores reproduces the main effects on 13,247 firm-years (Profit 1.792, Growth −1.092, both p < 0.001). The recovered firm-years average 28.6 on the ESG score (main sample 30.9) and are growth-oriented in 22.1% of cases (main sample 7.9%) and profit-oriented in 10.8% (main sample 38.4%).

5. Wording calibration

- "Policy resistance" (13 occurrences) is positioned as an interpretive pattern and candidate proposition throughout; the introduction now applies it "as an interpretive lens" and "propose[s] a candidate micro-foundation."
- Policy language: the conclusion now states that "looser financing shows no detectable ESG effect on either path in our data"; the green credit and green bond inference is removed; prescriptive "must" constructions are replaced by "warrant testing" and "should be matched."
- Evidentiary language: "Revealing its true asymmetric form" becomes "to document how this symmetric framework breaks in practice"; "sharply different results" becomes "a contrasting pattern"; "depends jointly on" becomes "is consistent with a joint dependence on"; Section 4.5 intensifiers are removed. The abstract now presents the financial-distress channel with its measure-dependent qualifier inline ("appears under an Ohlson-type measure ... but does not generalize to an Altman Z-score or temporally ordered specifications").
- Text S3 is retitled "The Ordering of the Mediation Analysis and Its Rationale," with the causal-direction framing rewritten; the matching sentence in Section 2.3, the supplementary list entry, and the corresponding sentence in Text S4.5 were updated.

6. R4 correspondence items

(1) Section 4.2 closes with the statement that a sequential-ignorability sensitivity analysis (Imai et al. [45]) was considered and not pursued, and why. (2) Table 4 carries the two-part estimates (Section 2 above). (3) Section 5.3 states that R&D intensity is not available in our data. (4) Section 4.6 states the rater-dependent dimensional pattern (dimensional regressions in Text S6.8). (5) Section 4.6 reports the firm-size interaction under two-way clustering (Growth × Size = −0.346, SE 0.147, p = 0.035; also in S6.10).

7. Figure 1 and construct labels

Figure 1 uses neutral labels: the construct boxes read "relative profit-growth position," the indirect paths read "indirect association (via O-Score)," and the caption states that the orientation constructs denote relative profit-growth positions and that the temporal reading (Section 2.1) is an interpretive lens. Figures S1–S4 were already neutral and are unchanged.

8. Reproducibility

The manuscript's Supplementary Materials paragraph now names the replication package (do-files, codebook, city–pilot crosswalk, and output logs) and the minimal dataset; because the submission system accepts one file per slot, both are included as separate, clearly named files at the root of the uploaded supplementary archive.
