2026-10-20
TROP and SDID time weights
One fixed representative example from each of 27 papers. TROP localization factors, SDID weights, estimates, tuning choices and exact-value downloads.
Open report →Panel reanalysis · updated 20 October 2026
Did or didn’t
A collection of replication reports, empirical time-weight comparisons and methodological checks for panel-data studies.
Original DiD replication collection
The original 22 papers, retained together with their results and Peter Hull’s established collection criteria. The main collection remains at its existing public address.
26 report entries
Weight comparisons
2026-10-19
All SDID time weights
All 754 returned examples across 27 papers. Defaults alone for 520 close matches; both numerical solutions for 234. Sixty unavailable attempts are listed separately.
Open report →2026-10-19
SDID defaults versus tighter optimization
Interactive comparison for the first 13 papers, holding samples and regularization fixed. Includes the refined optimized reference.
Open report →2026-10-19
SDID second tranche: 14 additional papers
360 returned examples from balanced subsets or shorter windows. Donor and treated-unit retention are shown; 58 zero-noise cases remain unavailable.
Open report →2026-10-19
Author-package SDID defaults: first tranche
394 returned fits from 13 papers using the authors’ default implementation, with time weights and iteration-limit diagnostics. The analysis.html file is an identical alternate entry point.
Open report →Inventory and provisional audits
2026-10-19
SDID readiness inventory
Study-level data readiness, pre-period lengths, donor counts and blockers across the 262-paper inventory. A readiness assessment, not a count of successful replications.
Open report →2026-10-13
Deep DiD audit follow-up
Specification-level reconciliation of the collaborator audit; 26 cases reopened for follow-up. Design evidence, data availability and implementation readiness remain separate.
Open report →2026-10-12
TWFE evidence in author replication code
File-and-line evidence for fixed-effect dimensions in 245 author deposits. TWFE syntax alone does not establish a binary-treatment DiD interpretation.
Open report →2026-10-12
Targeted PDF design audit
Page-cited review of 36 targeted papers and corrections to earlier whole-paper exclusions. Remaining unreviewed cases are explicitly identified.
Open report →2026-10-13
Community extension and study reports
The extension collection, calibration figures, inventories and provisional eligibility log. Individual study links retain their established public URLs.
Open report →2026-10-13
Provisional eligibility log
Entry point to the development collection’s eligibility log. Criteria and classifications remain subject to revision.
Open report →Numerical supplements
2026-10-20
TROP–SDID findings and profile sheets
Three-page plot PDF, selected parameters, estimates, comparison tables and verification results.
Open report →2026-10-20
TROP objective, tuning and solver checks
Pinned author objective, scalable equivalent optimization and direct author-code validation.
Open report →2026-10-19
SDID convergence findings
Reference-weight summaries, support and recent-mass changes, with links to the final precision refinements.
Open report →2026-10-19
SDID high-precision checks
Higher-precision solutions for nearly flat time-weight objectives, including the Italian opera examples.
Open report →2026-10-19
SDID convergence experiment
Author solver with tighter budgets, sparsification comparisons and initial QP diagnostics. Use the summary and precision supplement for the final reference weights.
Open report →2026-10-19
Second-tranche summary and profile sheets
Sample-retention summaries, numerical verification, final reference weights and the profile PDF.
Open report →2026-10-19
Second-tranche precision checks
High-precision checks for initializer-sensitive and low-donor examples.
Open report →2026-10-19
Second-tranche sample construction and analysis
Frozen complete-case and earlier-window rules, treatment-path checks and numerical analysis.
Open report →Earlier worked examples
2026-10
Credit Markets: counterfactual estimators
SDID, matrix completion, ridge and multisynth comparisons, event studies, tuning and bootstrap results. Historical report retained with its original method labels.
Open report →2026-10
Credit Markets: earlier MTGP/TROP experiment
Historical experiment predating the pinned author-TROP comparison above. Its TROP-labeled estimates use the earlier implementation and should be read separately.
Open report →2026-10
Credit Markets: extended replication template
Worked application of the extended regression and counterfactual-estimation template.
Open report →2026-10
Extended template: methods and source
Source and estimator-specific documentation accompanying the extended worked example.
Open report →2026-10-10
He & Wang: library pilot
Village officials and subsidies: the initial AEJ archive example, with replication results and estimator-specific limitations.
Open report →2026-10
Credit Markets: initial event-study example
Earlier event-study and deviations-test worked example. Later counterfactual and collection reports are linked separately.
Open report →Related thread notes
2026-10
Unique Games and semidefinite relaxations
A separate methodological note shared in the discussion, outside the empirical DiD collection.
Open report →No reports match this search.