Auction Market Design

该项目的研究目标是为FCC的激励拍卖（价值数十亿美元的电视转播权将在广播公司之间重新分配）和互联网广告拍卖（许多现代互联网公司的主要收入来源）等拍卖创造新的知识、方法和分析工具。在第一种应用中，有成千上万的无线电干扰限制，限制了可能授予广播公司的权利组合。这些约束使得这成为有史以来使用拍卖管理的最具计算挑战性的资源分配问题。在第二个应用中，困难的挑战是逆向选择，当不同的竞标者获得非常不同的信息时，就会出现逆向选择。例如，在互联网广告中，直接从点击链接中获益的广告商比通过提高消费者在实体店购买决策的意识而获益的品牌广告商更了解情况。在这个项目中，PI的团队将分析新的拍卖机制来克服计算和逆向选择的挑战，以及新的博弈论方法来分析替代拍卖设计的可能结果。本项目首先要执行的三项任务是：(1)发展并证明FCC激励拍卖机制的定理，该机制涉及解决一个np完全资源分配问题，并描述计算挑战如何影响分配效率，以及拍卖机制的激励、成本和简单性；(2)描述一种新的、反向选择的互联网广告拍卖自由拍卖机制，并评估其在各种环境中的表现；(3)发展一种新的博弈论均衡精化，这种精化既建立在基本原则之上，又能有力地消除一类拍卖博弈中不可信的纳什均衡。

The research objective of this project is to create new knowledge, methods and analytical tools for such auctions as the FCC's incentive auction, in which television broadcast rights worth billions of dollars will be reassigned among broadcasters, and Internet advertising auctions, which are the main revenue source for many modern Internet-based companies. In the first of these applications, there are hundreds of thousands of radio-interference constraints that limit the combinations of rights that may be granted to broadcasters. These constraints make this the most computationally challenging resource allocation problem ever managed using an auction. In the second application, the hard challenge is one of adverse selection, which arises when different bidders have access to very different information. For example, in Internet advertising, advertisers who benefit directly from clicks on their links are much better informed than brand advertisers who benefit by increasing consumer awareness for purchase decisions in brick-and-mortar stores. In this project, the PI's team will analyze new auction mechanisms to overcome the challenges of computation and adverse selection, and new game theoretic methods to analyze the likely outcome of alternative auction designs. The first three tasks to be undertaken in this project are these: (1) to develop and prove theorems about the FCC incentive auction mechanism, which involves solving an NP-complete resource allocation problem, and to characterize how the computational challenges affect the efficiency of the allocation, as well as the incentives, cost, and simplicity of the auction mechanism, (2) to characterize a new, adverse-selection free auction mechanism for Internet advertising auctions and evaluate its performance in a variety of environments, and (3) to develop a new game-theoretic equilibrium refinement that is both well-founded in fundamental principles and also powerful for eliminating implausible Nash equilibria of a class of auction games.