ChaisemartinDHaultfoeuille (DCDH) methodology-review-tracker promotion: In Progress -> Complete

[igerber]

Summary

• Flips the ChaisemartinDHaultfoeuille (DCDH) row in METHODOLOGY_REVIEW.md from In Progress to Complete (Last Review 2026-05-21).
• Rewrites the DCDH detail section to the Verified Components / Test Coverage / Corrections Made / Deviations / Outstanding Concerns template that ContinuousDiD (PR ContinuousDiD methodology-review-tracker promotion: In Progress -> Complete #476) and HAD (PR HeterogeneousAdoptionDiD methodology-review-tracker promotion: In Progress -> Complete #473) use.
• Adds a new ### Deviations from the paper / from R / library extensions block at the end of docs/methodology/REGISTRY.md ## ChaisemartinDHaultfoeuille, consolidating 7 documented deviations into a single AI-review-recognized labeled surface per CLAUDE.md "Documenting Deviations (AI Review Compatibility)" labels: (D1) equal-cell weighting (deviation from BOTH AER 2020 Equation 3 AND R DIDmultiplegtDYN); (D2) period-based vs cohort-based stable controls; (D3) balanced-baseline panel + interior-gap drops + terminal-missingness retention + cell-period-allocator targeted ValueError; (D4) SE normalization N_l vs R G (~4% smaller analytical SE); (D5) singleton-cohort degeneracy → NaN with UserWarning; (D6) <50% switcher warning at far horizons (library extension citing Favara-Imbs application, footnote 14 of NBER WP 29873); (D7) Phase 3 DID^X covariate first-stage equal-cell weights.
• Adds a [Unreleased] ### Added entry to CHANGELOG.md documenting the tracker promotion.
• Prunes DCDH from METHODOLOGY_REVIEW.md L27 In Progress example paragraph (re-pointed to WooldridgeDiD) and from L1289 priority-order queue item Add multi-period DiD support #6; items Prepare v0.2.0 release #7-Address code review feedback for data prep utilities #11 renumbered to Add multi-period DiD support #6-Add data preparation utilities for easier onboarding #10.
• Companion to PR PR-A: DCDH paper reviews (2020 AER + 2022 NBER WP) #478 (PR-A paper reviews); together they close the DCDH tracker promotion.

Methodology references (required if estimator / math changes)

• Method name(s): ChaisemartinDHaultfoeuille (DCDH) — DID_M, DID_+, DID_-, DID_M^pl, DID_l, DID_{g,l}, DID^{pl}_l, DID^n_l, delta cost-benefit, DID^X, DID^{fd}, state-set-specific trends, heterogeneity testing, Design-2, by_path, HonestDiD integration.
• Paper / source link(s):
  • de Chaisemartin, C. & D'Haultfœuille, X. (2020). Two-Way Fixed Effects Estimators with Heterogeneous Treatment Effects. American Economic Review, 110(9), 2964-2996. https://doi.org/10.1257/aer.20181169
  • de Chaisemartin, C. & D'Haultfœuille, X. (2022, revised July 2023). Difference-in-Differences Estimators of Intertemporal Treatment Effects. NBER Working Paper 29873. https://www.nber.org/papers/w29873
  • de Chaisemartin, C., Ciccia, D., D'Haultfœuille, X. & Knau, F. (2026). Difference-in-Differences Estimators When No Unit Remains Untreated. arXiv:2405.04465v6 (companion review for the universal-rollout extension; primary source for the HeterogeneousAdoptionDiD estimator).
• Any intentional deviations from the source (and why): No new deviations introduced. The new ### Deviations block consolidates the 7 already-existing scattered REGISTRY Notes into a single labeled surface for AI-review compatibility; the original inline **Note (deviation from R...):** entries remain in place at the point of maximum mechanical context.

Validation

• Tests added/updated: No test changes. Consolidation only against the existing 12 methodology tests (tests/test_methodology_chaisemartin_dhaultfoeuille.py), 26 R-parity tests (tests/test_chaisemartin_dhaultfoeuille_parity.py), 352 unit tests (tests/test_chaisemartin_dhaultfoeuille.py, 37 test classes), survey suites (tests/test_survey_dcdh.py, tests/test_survey_dcdh_replicate_psu.py, plus three dCDH cell-period coverage suites), and three primary-source paper reviews on disk (2020 AER + 2022/2023 NBER WP 29873 from PR PR-A: DCDH paper reviews (2020 AER + 2022 NBER WP) #478, 2026 Knau et al. universal-rollout companion).
• Backtest / simulation / notebook evidence (if applicable): N/A — documentation-only PR.

Security / privacy

• Confirm no secrets/PII in this PR: Yes

[github-actions]

Overall Assessment

✅ Looks good

No P0/P1 issues found. This PR is documentation-only, and the new DCDH deviation block/test inventory mostly matches the current repo state. The highest unmitigated finding is P2.

Executive Summary

• No estimator code, weighting, variance, or inference paths changed in this PR.
• The new DCDH REGISTRY deviations block is consistent with the existing documented DCDH notes.
• I verified the promoted inventory claims: tests/test_methodology_chaisemartin_dhaultfoeuille.py has 12 tests, tests/test_chaisemartin_dhaultfoeuille_parity.py has 26, and tests/test_chaisemartin_dhaultfoeuille.py has 352.
• One P2 documentation-scope issue remains: the promoted DCDH row treats the 2026 universal-rollout paper as a DCDH primary reference even though the DCDH REGISTRY entry and the 2026 review itself say that paper is HAD-specific / out of scope for DCDH.
• Residual risk is explicitly disclosed: the NBER review-of-record is the July 2023 revision.

Methodology

• P2 Impact: METHODOLOGY_REVIEW.md now lists the 2026 Knau et al. universal-rollout paper as a DCDH primary reference and counts its review toward DCDH primary-source coverage, but the DCDH REGISTRY primary sources are only the 2020 AER and 2022/2023 NBER papers, and the 2026 review itself says it covers the case ChaisemartinDHaultfoeuille does not handle and is the primary source for HAD. That makes the DCDH promotion narrative internally inconsistent and slightly overstates the methodological scope of the Complete label. Concrete fix: remove the 2026 paper from the DCDH Primary References and from the “Three primary-source paper reviews on file” checklist, or relabel it as companion / adjacent context rather than DCDH primary-source coverage. References: METHODOLOGY_REVIEW.md:L694-L716, METHODOLOGY_REVIEW.md:L744-L745, docs/methodology/REGISTRY.md:L486-L488, docs/methodology/papers/dechaisemartin-2026-review.md:L226-L228, CHANGELOG.md:L11-L11.

Code Quality

• No findings in the changed files.

Performance

• No findings in the changed files.

Maintainability

• No findings beyond the methodology-scope issue above.

Tech Debt

• No findings in the changed files.

Security

• No findings.

Documentation/Tests

• No findings. The promoted test-count and paper-review-file claims match the current repo state.

[github-actions]

🔁 AI review rerun (requested by @igerber)

Head SHA: 426e36c61e17464dc328782b1f20f5da3f3321a9

---

Overall Assessment

✅ Looks good

No unmitigated P0/P1 findings. This re-review resolves the prior documentation-scope issue: DCDH is now promoted against the 2020 AER and the July 2023 revision of WP 29873 only, while the 2026 universal-rollout paper is consistently treated as HAD-only adjacent context. The diff is documentation-only; no estimator, weighting, variance, inference, or default-behavior code changed. METHODOLOGY_REVIEW.md CHANGELOG.md REGISTRY.md dechaisemartin-2026-review.md

Executive Summary

• Previous review issue resolved: the 2026 paper is no longer counted as DCDH primary-source coverage in the tracker, changelog, or review narrative. METHODOLOGY_REVIEW.md CHANGELOG.md
• The new DCDH deviations block is a faithful consolidation of already-documented deviations/extensions, not a new undocumented methodology change. REGISTRY.md REGISTRY.md REGISTRY.md REGISTRY.md REGISTRY.md
• The promotion now rests on two on-disk DCDH primary-source reviews, which exist and are scoped correctly to the 2020 AER and July 2023 WP 29873. dechaisemartin-dhaultfoeuille-2020-review.md dechaisemartin-dhaultfoeuille-2022-review.md
• The promoted test inventory is internally consistent with the current source tree: 4 methodology classes / 12 test methods, 13 parity classes / 26 test methods, and 37 main-unit-test classes / 352 class test methods across the named files. METHODOLOGY_REVIEW.md tests/test_methodology_chaisemartin_dhaultfoeuille.py tests/test_chaisemartin_dhaultfoeuille_parity.py tests/test_chaisemartin_dhaultfoeuille.py
• Residual risk is disclosed rather than hidden: the review-of-record remains the “March 2022, revised July 2023” WP 29873 version, with a possible future 2024-revision re-review explicitly called out. METHODOLOGY_REVIEW.md docs/references.rst dechaisemartin-dhaultfoeuille-2022-review.md

Methodology

• P3-informational Impact: The new ### Deviations from the paper / from R / library extensions block is a documentation-surface change only. Each of its seven items maps back to pre-existing DCDH notes on weighting, control-set definition, ragged panels, SE normalization, singleton-cohort NaN behavior, far-horizon warnings, and DID^X weighting, so there is no new undocumented methodology deviation introduced by this PR. Concrete fix: None. REGISTRY.md REGISTRY.md REGISTRY.md REGISTRY.md REGISTRY.md REGISTRY.md

Code Quality

• No findings.

Performance

• No findings.

Maintainability

• No findings.

Tech Debt

• P3-informational Impact: The only remaining follow-up explicitly surfaced by the promoted entry is a possible paper-review pass against a later 2024 revision of WP 29873. It is disclosed as an outstanding concern rather than silently omitted, so it is not a blocker for this docs-only promotion. Concrete fix: Optional only; queue a follow-up review if that later revision materially changes the estimator. METHODOLOGY_REVIEW.md dechaisemartin-dhaultfoeuille-2022-review.md

Security

• No findings.

Documentation/Tests

• No findings. The referenced DCDH paper-review files and survey/cell-coverage test files exist, and the promoted test-count claims match the current file structure. I could not run pytest --collect-only in this environment because pytest is not installed, so that count check was static rather than runtime collection. METHODOLOGY_REVIEW.md dechaisemartin-dhaultfoeuille-2020-review.md dechaisemartin-dhaultfoeuille-2022-review.md tests/test_survey_dcdh.py tests/test_survey_dcdh_replicate_psu.py tests/test_dcdh_cell_period_coverage.py