Please use this identifier to cite or link to this item:
Full metadata record
DC FieldValueLanguage
dc.contributor.authorAzacis, Helmutsen_US
dc.contributor.authorVida, Péteren_US
dc.description.abstractWe study optimal bidder collusion at first-price auctions when the collusive mechanism only relies on signals about bidders' valuations. We build on Fang and Morris (2006) when two bidders have low or high private valuation of a single object and additionally each receives a private noisy signal from an incentiveless center about the opponent's valuation. We derive the unique symmetric equilibrium of the first price auction for any symmetric, possibly correlated, distribution of signals, when these can only take two values. Next, we find the distribution of 2-valued signals, which maximizes the joint payoffs of bidders. We prove that allowing signals to take more than two values will not increase bidders' payoffs if the signals are restricted to be public. We also investigate the case when the signals are chosen conditionally independently and identically out of n = 2 possible values. We demonstrate that bidders are strictly better off as signals can take on more and more possible values. Finally, we look at another special case of the correlated signals, namely, when these are independent of the bidders' valuations. We show that in any symmetric 2-valued strategy correlated equilibrium, the bidders bid as if there were no signals at all and, hence, are not able to collude.en_US
dc.publisher|aCardiff University, Cardiff Business School |cCardiffen_US
dc.relation.ispartofseries|aCardiff Economics Working Papers |xE2012/11en_US
dc.subject.keywordbidder-optimal signal structureen_US
dc.subject.keyword(Bayes) correlated equilibriumen_US
dc.subject.keywordFirst price auctionen_US
dc.subject.keywordPublic and private signalsen_US
dc.titleCollusive communication schemes in a first-price auctionen_US
dc.typeWorking Paperen_US

Files in This Item:
518.84 kB

Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.