A Test for Collusion between a Bidder and an Auctioneer in Sealed-Bid Auctions

Abstract

This paper derives a regression-based test to detect bidder-auctioneer cheating in sealed bid auctions. I apply this regression test to data from the New York City School Construction Authority auctions, an approximate one billion dollar per year auction market in which an auctioneer engaged in bidder-auctioneer cheating. Using the regression analysis to compare lots where bid rigging occurred with certainty to all other auctions allows one to conclude that bidder-auctioneer cheating significantly distorted the bid distribution. Comparing specific auctioneer lots before news of the cheating scandal became public with those after the scandal, I find significant differences in bidding, at the 10 percent level of significance, for two auctioneers. Therefore, bidder-auctioneer cheating may not have been limited to the one auctioneer charged with rigging bids.

Keywords: First-price auctions, collusion, bid-rigging, bidder-auctioneer cheating, order statistics, spacing