Detecting Bidders Groups in Collusive Auctions †
American Economic Journal: Microeconomics 2016, 8(2): 138 http://dx.doi.org/10.1257/mic.20130254 By Timothy G. Conley and Francesco Decarolis* We study entry and bidding in procurement auctions where contracts are awarded to the bid closest to a trimmed average bid. These auctions, common in public procurement, create incentives to coordinate bids to manipulate the bid distribution. We present statistical tests to detect coordinated entry and bidding choices. The tests perform well in a validation dataset where a court case makes coordination observable. We use the tests to detect coordination in a larger dataset where it is suspected, but not known. The results are used to interpret a major market shakeout following a switch to first price auctions. (JEL D44, D47, H57, R42) ... At the first meeting they said: `Why should we kill ourselves and let those coming from the outside laugh at us?' Here [in Turin] firms from the South were coming and getting the jobs, setting the averages, they used to come with 20, 30 or 40 bids, they used to get the jobs and then what was left for us?... --Confession of Bruno Bresciani, found guilty of having rigged 94 average bid auctions and other related crimes; sentenced to seven years of jail in 2008 n recent years, economists have helped design new auction markets for activities ranging from electricity supply contracts to the sale of spectrum licenses to mobile operators. The extent to which these auctions can deliver the intended results depends crucially on how bidders respond to strategic incentives. We study the case of a market in which average bid auctions (ABAs) are used for the procurement of public works and show the sophisticated response of bidders to the incentive to use multiple bids to pilot the contract awarding. We introduce two statistical tests that work well to detect this type of behavior and use them to study the drop in firm entry following a switch to first price auctions (FPAs). * Conley:...