Competition and collusion in auctions and other input markets

This dissertation examines firm behavior in auctions and other markets for inputs. Chapter 1 studies the transmission of private information via the outcomes of repeated auctions. This phenomenon is not present in the existing auction literature, because such information transmission is irrelevant both in series of independent auctions and in sequential auctions in which participants desire only a single item. However, when bidders desire multiple items, and when the values of those items to a bidder are correlated, the incentive to learn about opponents' values and to obscure one's own drive equilibria to depart systematically from those in standard models. I examine structural information transmission, created through various policies for conducting auctions, and its effect on strategic information transmission, which arises as an optimal response to given structural policies. Initially, I model information acquisition in situations where bidders do not see rival bids and learn only the identity of the winner. Even this simple structure has behavior vastly different from a series of independent auctions. In particular, it has softer bidding in early rounds, because winning with a low bid conveys more information about opponents' values than does winning with a high bid. This dominates the competing effect that losing with a high bid conveys more information than does losing with a low bid, because the value of information following a loss is inherently lower than the value of information following a win, due to the likelihood of repeated winning. I then use this model to show how the desire to conceal information about oneself affects behavior by examining repeated auctions with publicly announced bids. Finally, I show that a vertical merger between a buyer and a seller can be procompetitive due to both structural and strategic information transmission.Chapter 3 provides a technical discussion on the monotonicity of bidding functions in the standard sequential auction model. Such auctions differ from those discussed in Chapter 1 in that bidders in sequential auctions desire only a single item. Therefore, the incentive to alter one's behavior to acquire information about one's opponents, which was the centerpiece of Chapter 1, is absent. This proof validates the assumption made in the literature, and it shows that earlier work which characterized monotone equilibria actually characterized all equilibria. A technical appendix proves that best response correspondences must also be strictly increasing in a player's type, thus ruling out the possibility of mixed strategy equilibria except at a countable number of bidder valuations.Chapter 2 examines how the structure of input markets affects overall competition in both input and output markets. For example, suppose two firms meet repeatedly in an output market. Contrary to conventional wisdom, the firms may prefer to face one another in their input markets, rather than have monopsony power. This possibility suggests that the structure of relevant input markets be examined when a merger is proposed by firms which possess market power in their input market. The converse story, that two duopsonists may prefer duopoly over monopoly, has implications for the welfare effect of models of cross-hauling, and introduces an explanation for potential welfare enhancing gains from exclusive territories imposed by distributors.