Information and Incentives in Dynamic Games

数理经济学为研究广泛的重要和基本的经济问题提供了严格的模型和强大的分析工具。这个项目强调了数理经济学中一些最活跃的研究领域，即关于信息和激励在经济主体之间动态讨价还价中所起作用的工作。该项目由三个部分组成。第一部分探讨了游戏中的均衡可能是玩家从之前的游戏中学习的结果。这一新方法没有将学习过程建模为均衡的一部分，而是假设参与者使用不同类型的“合理”行为规则。一个特别受到关注的问题是，对这些行为规则的限制与各种均衡精炼之间的关系。第二部分是对单人声誉效应研究的综述。之前的研究发现，当只有一个玩家有维持声誉的动机时，声誉效应特别强大。单人玩家能够通过维护声誉来选择他最喜欢的均衡。第三部分研究了风险中立的委托人和厌恶风险的代理人之间的最优契约的性质，当代理人的行为面临道德风险，并且在代理人履行了自己的职责但代理人的行为的所有影响尚未显现之前，双方可以自由地重新谈判合同。本项目这一部分的目标是了解对防止重新谈判的合同的约束如何限制代理人可以被迫承担的风险量，以及可以诱使代理人提供的工作量。这个项目为我们理解经济博弈动态中的信息和激励做出了方法论和实质性的贡献。在方法上，该项目开发了一种新的动态游戏方法，允许分析师推导出游戏参与者的信念，而不仅仅是假设。传统的方法通过对玩家的信念做出“看似合理的”假设来获得动态博弈的解决方案。这些假设被称为均衡精化。不幸的是，博弈论模型通常无法测试，因为关于玩家信念有许多不同的合理假设，每个假设都有不同的结果；因此，这种新的方法使我们更接近可测试的动态博弈论的目标。实质上，该项目为经济主体获得和维护声誉的动机提供了新的见解。契约理论被扩展到允许研究代理人采取行动的时间与其影响的实现时间之间有很长差距的情况下的最优合同，以及在此期间有可能重新谈判的情况。

Mathematical economics provides rigorous models and powerful analytical tools for studying a wide range of important and fundamental economic issues. This project emphasizes some of most active areas of research in mathematical economics, namely work on the roles played by information and incentives in dynamic bargaining among economic agents. The project consists of three parts. The first part explores the idea that equilibrium in games might arise as the result of the players learning from thir previous play. This new approach does not model the learning process as part of an equilibrium, but rather assumes that the players use different sorts of "reasonable" behavioal rules. One question given special attention is the relationship between restrictions on these behavioral rules and various sorts of euilibrium refinements. The second part generalizes single player studies of reputation-effects. Previous research found the reputation effects particularly powerful when only a single player has an incentive to maintain a reputation. The single player is able to pick out the equilibrium he most prefers through maintaining a reputation. The third part studies the nature of the optimal contract between a risk-neutral principal and a risk-averse agent when the agent's action is subject to moral hazard and the parties are free to renegotiate the contract after the agent has performed his duties but before all the effects of the agent's actions have been revealed. The objective of this part of the project is to learn how the constraint to renegotiation-proof contracts limits the amount of risk that the agent can be forced to bear, and the amount of effort that the agent can be induced to provide. This project makes methodological and substantive contributions to our understanding of information and incentives in the dynamics of economic games. Methodologically, the project develops a new approach to dynamic games that allows the analyst to derive and not just assume the beliefs of the participants in the games. The traditional approach obtains solutions to dynamic games by making "plausible" assumptions about the beliefs of players. These assumptions are called equilibrium refinements. Unfortunately, game-theoretic models usually can not be tested because there are many different plausible assumptions about player beliefs, each with a different outcome; hence, this new approach moves us closer to the goal of a testable dynamic game theory. Substantively, the project provides new insights into the incentives of economic agents to obtain and maintain reputations. Contract theory is extended to permit the study of the optimal contract for settings in which there is a long gap between the time the agent's actions are undertaken and the time their effects are realized and where in the interim there is the possibility of renegotiation.