H:cobidder-concentration — The frequent-loser ranking concentrates CADE-adjacent cobidders¶
If the award-layer signal is informative for enforcement triage, then firms that always lose but bid alongside direct CADE defendants in adjudicated cartel environments should appear disproportionately in the upper-loser stratum identified by persistent zero-win participation. The hypothesis is about loser-side adjacency, not about cartel membership: the ranking should concentrate cobidders, not direct defendants.
Intuition (plain-language)
Among always-loser firms in Brazilian procurement, the ones that bid most often (frequent losers) concentrate disproportionately near cartels formally adjudicated by Brazil's antitrust authority. The hypothesis is the paper's headline empirical claim. Using only award records — no expensive bid-level data — the screen recovers 131 of 193 adjudicated cobidders. The economic logic: cartels need cover bidders to manufacture the appearance of competition, and those cover bidders leave a persistent footprint of zero-win participation.
Evidence strength: Partial (strongly supported). Eight converging lines of evidence on the BEC 2009–2019 panel: (i) headline AUC 0.924 [0.921, 0.926] for FL14, 0.939 [0.932, 0.946] for continuous log_tc, recovering 131/193 adjudicated cobidders (AN-004); (ii) cutoff sweep across 19 thresholds with a clean inverted-U peaking at FL13 (0.924) and FL10-FL15 plateau ≥ 0.90 (AN-025); (iii) subsample robustness: AUC 0.89–0.96 across full / data-rich / low-bid-count / high-bid-count subsamples (AN-026); (iv) continuous dominance: DeLong Z = −4.38, p = 1.2 × 10⁻⁵ (AN-011); (v) volume placebo: permutation AUC 0.713–0.783 ≪ 0.924 (AN-005); (vi) leakage audit: raw 0.995 → OOF 0.891 → temporal-holdout 0.864 (AN-014); (vii) strict ex ante (train 09-16 → test 17-19): firm AUC 0.767 (AN-006); (viii) continuous-only thesis without FL14 (gate D3 passes, AN-017). Promotion to 🟢 (Confirmed) requires independent replication on a non-BEC procurement panel — not satisfied by any of the above, all of which share the BEC × CADE data lake.
Theory¶
In cartels that allocate winning roles and rotate designated winners \citep{pesendorfer2000study,asker2010leniency,marshall2012economics}, firms used to manufacture the appearance of competition can plausibly be identified by their persistent zero-win exposure. The same logic appears in large-scale procurement evidence on coordinated bidding \citep{kawai2022detecting}. Persistent zero-win participation is a behavioral footprint of credible losing roles, not a label of cartel membership.
Prediction¶
The frequent-loser stratum (binary FL14 rule, top-loss-intensity quantile
of the continuous score) should contain a disproportionate share of CADE
cobidders relative to the baseline rate in the always-loser population.
Competing prediction¶
Volume-only explanation. Firms that bid in many tenders mechanically have more chances of co-bidding alongside any subset of firms, including CADE defendants. The cobidder concentration could reflect raw participation volume rather than a signal-carrying loser-side footprint. Adjudication disciplines this in H:exposure-discipline.
Case evidence¶
CADE adjudications in the 2009–2019 window cover bid-rigging conduct in São Paulo procurement environments. The cobidder set is constructed from these adjudicated cartels by identifying always-loser firms that bid in the same tenders as direct defendants in the adjudicated environments. See §2.3 of the manuscript.
Empirical test¶
- Outcome: cobidder indicator (1 if firm co-bid with a direct CADE defendant in an adjudicated environment).
- Variation: position in the FL stratum vs other always-loser firms.
- Specification: AUC of
FL14(binary) andlog(1+tenders_count)(continuous) over always-losers, with cobidders as the positive class. - Identification: scope of the loser-side ranking is restricted to zero-win firms; baseline benchmark before any audit discipline.
Data requirements and limitations¶
Requires BEC LANCES export (firm × item × OC) joined with CADE adjudication records linked at the CNPJ root level. Limitation: cobidder labels depend on adjudication coverage in the panel window, and the set is exposure-limited to CADE-adjudicated environments in São Paulo.
Evidence¶
| Analysis | Bearing | Status | Key takeaway |
|---|---|---|---|
| AN-001 (construction) | Setup | done | 16,843 always-losers; FL14 cutoff = 14 (median+1.5×IQR=13.5); 2,735 FL firms (16.2%) |
| AN-004 (baseline concentration) | Supports | done | FL14 AUC 0.924; continuous 0.939; 131/193 cobidder recovery |
| AN-005 (volume placebo) | Supports | done | Permutation AUC 0.713–0.783 ≪ 0.924 observed |
| AN-006 (timing discipline) | Supports | done | Strict ex ante firm AUC 0.767 [0.734, 0.800] |
| AN-011 (continuous vs binary) | Supports | done | Continuous dominates FL14, DeLong p = 2e-5 |
| AN-014 (leakage audit) | Supports | done | Raw 0.995 → OOF 0.891 → temporal 0.864; attenuation honest |
| AN-017 (continuous-only thesis) | Supports | done | Thesis holds without FL14 |
| AN-023 (operationalization audit) | Supports | done | FL14 not ontologically privileged; continuous is the primitive |
| AN-025 (cutoff sweep) | Supports | done | 19-threshold inverted-U; FL10-FL15 plateau ≥ 0.90; peak FL13 = 0.924 |
| AN-026 (subsample robustness) | Supports | done | AUC 0.89–0.96 across full / data-rich / low-bid / high-bid subsamples |
Open tests¶
- Sensitivity of concentration to alternative
FL14-like thresholds. - Sub-period replication (2009–2014 vs 2015–2019).
- Cross-modality (Convite vs Pregão) decomposition.