Explicit characterization of private randomized bidding
Propositions 1–2 derive the optimal winning-bid distribution and identify when it strictly improves on common-knowledge bid rotation.
↳ Section 2.2, Propositions 1–2
Abstract
Abstract We study how incentive constraints can be relaxed by randomization in a repeated-game setting. Our study is motivated by the workings of a detected bidding cartel that adopted a protocol of keeping the winning bid secret from the designated losers when defection was a concern. Keeping the winning bid secret makes accurately undercutting the winning bid more difficult and makes defection less attractive as potential defectors risk not winning the auction even if they deviate. We formalize these ideas in the context of a repeated-game setting and show that a cartel can attain higher payoffs by having the pre-selected winner randomize its bid and keep it secret from other members. Calibration of the model to the bid data of the cartel suggests that randomization may increase firms’ profits by about 56%.
Key Claims
Primary findingWell-supported
a cartel can attain higher payoffs by having the pre-selected winner randomize its bid and keep it secret from other members.
explicit incentive constraints and equilibrium construction establish the conditional gain within the stated repeated-auction model
Abstract; Propositions 1–2, Section 2.2
SecondaryTentative
Calibration of the model to the bid data of the cartel suggests that randomization may increase firms’ profits by about 56%.
single-cartel calibration with assumed cost and discount parameters and unresolved reserve-price tail sensitivity
Abstract; Section 5
SecondarySupported
an auctioneer facing a cartel may strictly benefit from restricting the information that she makes public after each auction.
the disclosure comparison supports this implication within the specified bid-rotation model, not as a generally validated procurement intervention
Introduction; Section 2.2.4, Proposition 3
SecondaryTentative
calibrated model can replicate cartel bidding patterns such as the winning margin and how it varies with the reserve price.
only one auction discriminates the model from common-knowledge bidding
Introduction; Section 5, Figure 4
Derived from the full evaluation — not a separate score.
Strengths · Limitations
Strengths
Verifiable incentive constraints and boundary analysis
The supplied Appendix checks winner and loser incentives, while the extensions examine disclosure restrictions, unrestricted equilibria, and a protocol without mediation.
↳ Appendix, Propositions 1–2; Sections 2.2.4–2.3
Direct engagement with competing collusion theories
The literature discussion distinguishes bid rotation from symmetric stationary equilibria and explicitly acknowledges the baseline restriction's costs.
↳ Section 1.1; Section 2.3.1
Limitations
Empirical discrimination rests on one auction
Only the Obara auction exceeds the calibrated threshold where common-knowledge and private randomized bidding make different predictions. The remaining observations therefore provide little discrimination between the mechanisms.
↳ Section 4.2; Section 5, Figure 4
Quantified gains depend on calibration assumptions
The 56% gain depends on the cost margin, discount factor, and fitted reserve-price tail. The Pareto shape of 0.575 implies an infinite mean despite the model's finite-expectation assumption, although bounded equilibrium-profit calculations preserve the reported arithmetic.
↳ Section 2.1; Section 5, Footnote 36
Empirical fit language exceeds discriminating evidence
The Introduction says predictions are 'borne out' and Section 5 says the model explains the data quite well. These interpretations do not sufficiently distinguish compatibility with observed bids from validation against competing explanations.
↳ Introduction; Section 5, Figure 4
Nabu Commentary
Propositions 1–2 provide the paper's strongest foundation through explicit incentive constraints and an equilibrium construction for private randomized bidding. The disclosure and mediation extensions clarify where that mechanism operates. Figure 4 supplies a useful illustration, but only the Obara auction distinguishes the competing bidding predictions, so the 56% figure is best read as a model-implied calibration rather than an estimated general effect. The fitted Pareto's infinite mean and conflicting percentage, discount-factor, and auction-count reports warrant attention without establishing failure of the core theoretical result.
Check another paper
Author Note / Field Feedback
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Dimension breakdown
Study Quality
Strong4.0
Confidence mediumContribution
4.1
The score reflects a delineated theoretical advance linking private randomized recommendations to relaxed deviation constraints and higher cartel payoffs. The calibration adds a quantitative illustration, but only one auction distinguishes the competing bidding predictions.
Section 1.1; Section 2.2, Propositions 1–2; Section 5, Figure 4
Methodological Rigour
4.1
The central supplied derivations, explicit primitives, and boundary analyses support scoring near the reviewer mean. Calibration sensitivity, the finite-expectation mismatch, and the single discriminating observation keep the assessment below the higher excellence range.
Section 2.1; Sections 2.2.4–2.3; Appendix, Propositions 1–2; Section 5 and Footnote 36
Reporting
4.0
Definitions, propositions, and figures make the information structure and incentive mechanism traceable. Localized overstatement of empirical fit and inconsistent numerical reporting temper the otherwise coherent exposition.
Introduction; Section 2.2; Figures 1 and 4; Section 6
Positioning
3.9
The paper directly distinguishes its equilibrium class from close alternatives and acknowledges data selection and baseline bid-rotation suboptimality. The implications of the single discriminating auction are not fully carried through into the broader interpretation of empirical fit.
Section 1.1; Section 2.3.1; Section 4.1; Section 5, Figure 4
Reliability
Caveats4 of 4 checks
Results checkNoted
- Calibration of the model to the bid data of the cartel suggests that randomization may increase firms’ profits by about 56%: "finding that cartel surplus can increase by as much as 50% from randomized bids" (Section 6 (Discussion), first paragraph) vs "the gain is about 56% (V ck = 525 million versus V = 820 million)" (Section 5, calibration results). (820 − 525)/525 = 0.5619, supporting approximately 56% rather than the Discussion's 50%. → Conclusion holds
Reporting and transparencyClean
A named Zenodo DOI is provided for data and code. No specific reporting or research-conduct concern is established; repository contents were not independently inspected.
Data availability: declared · Code availability: declared
Reference IntegrityClean
60 of 61 checkable references verified
67 references in manuscript 6 are books, websites or datasets — counted, but not index-checkable
Post-Publication RecordClean
No retraction or expression of concern found in Crossref (includes Retraction Watch).
Last checked: 20 Oct 2026
Sources: Crossref (incl. Retraction Watch) ✓
Readiness
Where this paper’s evidence sits on the path from initial observation to real-world use.
Demonstrated
The formal mechanism and calibrated case study establish an early application basis rather than a validated intervention. Implementation guidance and testing across procurement systems remain absent.
Section 2.2.4, Proposition 3; Sections 5–6
AI-generated, human-governed. Something look off? Contact us to request a review.