Essays on Information Disclosure in Auctions and Contests

Abstract

In strategic decision making, the agents' decisions depend crucially on the information they have. Less information means more uncertainty about the other agents' possible behavior, making strategic decisions more complex. As information has a strategic value, it seems natural to expect that in competitive environments an agent has an advantage if his rivaling agents know as little as possible about him. In this thesis, we study competitive strategic situations where this is not necessarily the case: we look at first-price auctions and innovation contests in which the participating agents voluntarily reveal information about themselves.<br /> In the first chapter, we study information disclosure in innovation contests. In innovation contests, the progress of the competing firms in the innovation process is usually their private information. We analyze an innovation contest in which research firms have a stochastic technology to develop innovations at a fixed cost, but their progress is publicly announced. We make a comparison with the case of no information revelation: if the progress is disclosed, the expected profit of the firms is higher, but the expected profit of the sponsor is lower. Additionally, we show that firms may voluntarily reveal their information.<br /> The second chapter deals with a particular kind of auctions. In multi-attribute procurement auctions with multiple objects, the auctioneer may care about the interplay of quality attributes that do not belong to the same item -- like each item's delivery time, if all items are needed at once. This can influence the performance of the auction mechanism. We generalize the Ausubel-Milgrom ascending proxy auction to such an environment and show that the main properties still hold: equilibria in profit-target strategies exist, the final allocation maximizes the surplus and the payoff vector is in the core. <br /> Furthermore, the scoring rule used to evaluate the bids may contain valuable information about the auctioneer for his competitors, providing an incentive not to reveal it. It is possible to keep the scoring rule secret without changing the outcome of the auction and a universal equilibrium strategy exists for a particular class of scoring rules. Additionally, for additive scoring rules there is a close connection to the original proxy auction.<br /> The final chapter considers information revelation in first-price auctions. It is commonly assumed in private value auctions that bidders have no information about the realization of the other bidders' valuations. Nevertheless, an informative public signal about the realization may be released by a bidder while he learns his own valuation. Using a simple discrete asymmetric first-price auction setting, we show that a bidder may indeed benefit from the presence of an informative signal about his own valuation. We characterize the optimal signal and show that a signal is not beneficial if it is too precise. The latter result carries over to a general continuous asymmetric first-price auction model. Finally, we use a specific signaling structure with uniform distributions to show that signaling need not be beneficial for any precision of the signal.