Skip to content

Changelog

v4 --- March 2026 (Current)

Manuscript:

  • Rewrote contributions paragraph with explicit H1/H2 structure and scope limitation
  • Softened over-claims throughout (9 instances: "establish" to "show", "confirming" to "indicating")
  • Expanded IV section: preferred spec declaration, lambda interpretation, LATE discussion
  • Bajari--Ye restructured around 3 sequential claims with magnitude hedge
  • DiD shrunk to 2 paragraphs (complementary, underpowered)
  • Mechanisms: added connective sentences and bridge paragraph
  • Heterogeneity: relabeled "high/low-suspicion" to "concentrated/competitive-market FL"
  • Added robustness roadmap paragraph (6 groups)
  • Added implementation blueprint (Flag, Triage, Investigate, Monitor)
  • Added enforcement integration and legal boundary sentences
  • Abstract trimmed to 136 words with policy contribution sentence
  • Broader implications paragraph added to conclusion
  • Float numbering fixed: appendix counters reset (A.1, B.1, etc.)
  • Added missing cross-references for 3 main-body floats

Tables and figures:

  • 35 tables (12 main body, 23 appendix)
  • 14 figures (6 main body, 8 appendix)
  • 59 pages, 0 errors, 0 undefined references

Pipeline:

  • 13 R scripts, full pipeline in ~8 minutes
  • New tables: IV balance, IV Panel C, Bajari--Ye tender FE, network interactions, homogeneous CV, welfare bounds (updated)

v3 --- February 2026

  • CADE validation section with CNPJ enrichment (3 FL firms matched)
  • Network-split IV regressions
  • Bajari--Ye tender fixed effects specification
  • Homogeneous sub-samples via price-CV approach
  • Welfare analysis with cross-fit bounds
  • Staggered DiD with MDE and Rambachan--Roth sensitivity

v2 --- January 2026

  • Bajari--Ye exchangeability and conditional independence tests
  • Network-based FL classification (winner HHI, repeat partners)
  • Regime test: complementary vs. coordinated cover bidding
  • Sensitivity analysis (Cinelli & Hazlett, 2020)
  • Matching estimators (CEM and IPW)
  • Year-by-year coefficient stability

v1 --- December 2025

  • Initial manuscript with OLS and IV regressions
  • FL definition (always-losers + IQR threshold)
  • Leave-one-out instrumental variable
  • 18 tables, 6 figures
  • Threshold sensitivity and cross-fitting robustness