EfficientDiD methodology review (PR-A): Chen-Sant'Anna-Xie (2025) paper-review fidelity artifact

[igerber]

Summary

• Adds docs/methodology/papers/chen-santanna-xie-2025-review.md — a faithful, paper-sourced-only fidelity review of Chen, Sant'Anna & Xie (2025), Efficient Difference-in-Differences and Event Study Estimators (arXiv:2506.17729v1) — the Step-1 (PR-A) artifact of the EfficientDiD methodology-review validation.
• Transcribes (no code-deviation verdicts): assumptions S/O/NA/PT-Post/PT-All; the single-treatment-date EIF (Theorem 3.1, Eqs 3.2–3.6; Corollary 3.1) and the staggered analog (Theorem 3.2, Eqs 3.9–3.14; Corollary 3.2); the no-covariate closed form + the covariate sieve/kernel doubly-robust estimation path of §4 (Eqs 4.1–4.5); efficiency + SE results (Theorem 4.1; SE = sqrt(mean(EIF²)/n) or multiplier bootstrap); the Hausman PT-All-vs-PT-Post pretest (Appendix A, Theorem A.1); and the HRS Table 6 / Table 7 ARE cross-language anchor for the follow-up pass.
• Wiring: docs/references.rst upgraded from "Working Paper" to the arXiv record; REGISTRY.md ## EfficientDiD gains a Paper review on file: pointer (+ URL-linked citation, v1-pin note); the review is registered in docs/doc-deps.yaml; METHODOLOGY_REVIEW.md records the paper review under "Documentation in place" and corrects the priority queue.
• The tracker row stays In Progress — the source-validation pass (PR-B: tests/test_methodology_efficient_did.py, the HRS anchor, and the deviation log) flips it.
• Docs-only; no estimator, weighting, variance/SE, or default behavior changed. No paper PDF committed.

Methodology references

• Method name(s): EfficientDiD
• Paper / source link(s): Chen, X., Sant'Anna, P. H. C., & Xie, H. (2025). Efficient Difference-in-Differences and Event Study Estimators. arXiv:2506.17729v1 — https://arxiv.org/abs/2506.17729 ; Cowles Foundation Discussion Paper No. 2470.
• Any intentional deviations from the source (and why): None in this PR. PR-A records paper facts only. The implementation-vs-source items to verify (e.g. linear-OLS outcome regressions vs. the paper's general nonparametric m̂(X); fixed-weight vs. WIF-corrected bootstrap aggregation; the overall-ATT vs. ES_avg convention) are listed as neutral PR-B verification targets in the review's "Gaps and Uncertainties" / "Where the paper leaves a choice open" sections and will be documented as labeled deviations in PR-B.

Validation

• Tests added/updated: No test changes — PR-A is a docs-only paper-review artifact (the methodology test file is PR-B work).
• Backtest / simulation / notebook evidence (if applicable): N/A. Fidelity verified by direct cross-check of every cited theorem / equation / assumption against arXiv:2506.17729v1, plus 8 rounds of local Codex review (--backend codex, full standard mode) — converged clean (zero findings), having caught and fixed two genuine fidelity errors en route (an over-restricted Lemma 2.1 comparison set, and a reversed Hausman covariance direction).

Security / privacy

• Confirm no secrets/PII in this PR: Yes (docs-only; paper PDF kept outside the repo per the PDFs-never-committed convention)

---

Generated with Claude Code

[github-actions]

Overall Assessment

✅ Looks good

Executive Summary

• This is a docs-only PR: no estimator logic, weighting, variance/SE code, or defaults change in the diff.
• No unmitigated P0/P1 methodology findings. The new EfficientDiD paper-review artifact stays within a paper-transcription scope and does not newly claim code-vs-paper validation; that remains deferred to PR-B in the tracker. docs/methodology/papers/chen-santanna-xie-2025-review.md:12 METHODOLOGY_REVIEW.md:644 docs/methodology/REGISTRY.md:884. citeturn0search0turn0search1
• The remaining EfficientDiD validation work is explicitly tracked, so the lack of new methodology tests here is not a blocker. METHODOLOGY_REVIEW.md:645
• One minor issue: the new arXiv references are unversioned even though the surrounding prose says theorem/equation numbering is pinned to v1; that creates future drift risk if arXiv gets a later revision.
• No security, performance, or code-quality concerns surfaced in the changed files.

Methodology

• No findings. The changed docs are internally consistent with the existing EfficientDiD registry/tracker framing and do not introduce a new undocumented methodology deviation in this PR’s scope. docs/methodology/papers/chen-santanna-xie-2025-review.md:12 METHODOLOGY_REVIEW.md:644 docs/methodology/REGISTRY.md:884

Code Quality

• No findings in the changed lines.

Performance

• No findings in the changed lines.

Maintainability

• Severity: P3. Impact: the new references use unversioned arXiv URLs while the prose explicitly says numbering is pinned to arXiv v1; if a v2+ appears later, these links will silently drift away from the reviewed source version. Affected lines: docs/methodology/REGISTRY.md:884, docs/references.rst:219, docs/methodology/papers/chen-santanna-xie-2025-review.md:5, docs/methodology/papers/chen-santanna-xie-2025-review.md:16. Concrete fix: switch those URLs to .../2506.17729v1, or add an explicit version-pinned link alongside the unversioned citation.

Tech Debt

• Severity: P3. Impact: the substantive source-validation work is still outstanding, but it is explicitly tracked in the methodology review and therefore is not a blocker for this docs-only PR. METHODOLOGY_REVIEW.md:644 METHODOLOGY_REVIEW.md:647. Concrete fix: no action required in this PR; land the already-scoped PR-B items before promoting EfficientDiD to Complete.

Security

• No findings. The diff is documentation/registry wiring only, and I did not see secrets, tokens, or PII in the changed files.

Documentation/Tests

• Severity: P3. Impact: no tests were added here, but the PR accurately represents itself as PR-A documentation work and keeps the estimator review status In Progress; that is acceptable because the missing validation artifacts are already tracked. METHODOLOGY_REVIEW.md:642 METHODOLOGY_REVIEW.md:645. Concrete fix: none required for this PR; keep PR-B scoped to the numbered methodology test file, empirical anchor, and deviation log.

[github-actions]

🔁 AI review rerun (requested by @igerber)

Head SHA: 15aedf84937de3129f88e082e30559ee89d93bfc

---

Overall Assessment

✅ Looks good

Executive Summary

• This remains a docs-only PR. It does not change estimator code, weighting, variance/SE implementation, identification checks, or defaults; the new entry explicitly says so at CHANGELOG.md:L11-L11.
• The prior re-review issue is resolved: the changed references now pin the source to arXiv:2506.17729v1 in docs/methodology/REGISTRY.md:L884-L884, docs/references.rst:L219-L221, and docs/methodology/papers/chen-santanna-xie-2025-review.md:L4-L5.
• The PR does not overclaim validation status. EfficientDiD stays In Progress, and the remaining source-validation work is still explicitly deferred to PR-B at METHODOLOGY_REVIEW.md:L642-L647 and METHODOLOGY_REVIEW.md:L1413-L1417.
• I spot-checked the new paper-review artifact against the cited paper’s core claims that matter most here: the abstract framing, Theorems 3.1/3.2, Theorem 4.1’s SE statement, Appendix A’s Hausman covariance direction, and the HRS Table 6/Table 7 anchor. I did not find a transcription mismatch in the changed text. (arxiv.org)
• The documentation wiring is complete for this scope: the new review is linked from the registry/tracker and registered in docs/doc-deps.yaml:L295-L303.

Methodology
No findings. The changed files keep the paper-review artifact clearly scoped as a paper-sourced baseline and continue to defer source-vs-code validation, deviation logging, and the empirical anchor to PR-B (docs/methodology/papers/chen-santanna-xie-2025-review.md:L12-L12, METHODOLOGY_REVIEW.md:L644-L647). My source spot-check did not uncover an undocumented methodology mismatch in the new review text. (arxiv.org)

Code Quality
No findings. The diff is documentation/registry wiring only.

Performance
No findings. No runtime code path changed.

Maintainability
No findings. The one prior maintainability concern from the earlier AI pass, unversioned arXiv links, is fixed in the changed references at docs/methodology/REGISTRY.md:L884-L884, docs/references.rst:L219-L221, and docs/methodology/papers/chen-santanna-xie-2025-review.md:L4-L5.

Tech Debt
No findings. The remaining EfficientDiD validation work is explicitly tracked rather than silently deferred, so it is non-blocking for this docs-only PR (METHODOLOGY_REVIEW.md:L644-L647, METHODOLOGY_REVIEW.md:L1415-L1417).

Security
No findings. I did not see secrets, tokens, or PII in the changed files.

Documentation/Tests
No findings. For a docs-only PR, no test changes are acceptable here because the tracker still marks the method In Progress, and the follow-up validation work is spelled out concretely. The new review is also correctly added to the doc dependency map at docs/doc-deps.yaml:L295-L303.