Collusion and Equilibrium Selection in Auctions

Abstract

We study bidder collusion and test the power of payoff dominance as an equilibrium selection principle in experimental multi-object ascending auctions. In these institutions low-price collusive equilibria exist along with competitive payoff-inferior equilibria. Achieving payoff-superior collusive outcomes requires complex strategies that, depending on the environment, may involve signalling, market splitting, and bid rotation. We provide the first systematic evidence of successful bidder collusion in such complex environments without communication. The results demonstrate that in repeated settings bidders are often able to coordinate on payoff-superior outcomes, with the choice of collusive strategies varying systematically with the environment. Auctions for timber, cars, oil drilling rights and spectrum bandwidth are just a few examples of markets where multiple heterogeneous lots are offered for sale simultaneously. In multi-object ascending auctions, the applied issue of collusion and the theoretical issue of equilibrium selection are closely linked. In these auctions, bidders can profit by splitting the markets. By acting as local monopsonists for a disjoint subset of the objects, the bidders lower the price they must pay. As opposed to the sealed bid auction, the ascending auction format provides bidders with the opportunity to coordinate tacitly on dividing markets and to punish non-collusive bidding. Thus, low-price collusive equilibria co-exist along with high-price competitive equilibria even in a non-repeated setting (Brusco and Lopomo, 2002). If there are complementarities in bidder values across objects, and auctions are repeated, bidders may further benefit by splitting the markets over time and taking turns in buying packages at low prices. These opportunities make the multi-object ascending auction an ideal institution for the study of equilibrium selection. Assuming that communication between the bidders is limited to the bidding, any splitting of the markets must be tacitly coordinated in the early stages of the auction. Thus achieving payoff-superior collusive outcomes requires complex strategies that, depending on the environment, may involve signalling, market splitting and bid rotation. While such tacit coordination is possible in theory, it is an empirical question of whether it may be achieved in practice.