Monotone Methods in Reputations: Behavioral Predictions and Reputation Sustainability

该奖项资助经济理论方面的研究。PI希望推进博弈论的应用，以了解战略决策者如何建立和维护声誉。这项研究的重点是了解一个长寿的个人、公司或组织在面对一系列短命的对手时将如何做出决策，这些对手对相关事实的了解较少。该项目将侧重于开发可以用数据验证的具体预测。这项研究试图回答具体的问题。首先，决策者何时能维持声誉？其次，在声誉建设成本高昂、监控嘈杂的情况下，决策者将如何表现？第三，在长期互动数据有限的情况下，我们如何才能最好地测试声誉模型的预测？首席研究员将用单调的方法回答这些问题。第一个项目将相互依赖的价值观引入到声誉研究中。这将声誉结果的适用性扩展到更丰富的信息环境集。除了玩家的收益外，该项目还将提供关于玩家行为的可靠预测。第二个项目将研究相互依存的价值观是否以及如何有助于声誉的可持续性。目标是将长期结果（如声誉可持续性）与基本要素联系起来，包括阶段博弈收益的超模块性或亚模块性，以及玩家对收益相关状态的信念。第三个项目开发了另一种研究声誉的框架。这个框架可以帮助我们更好地理解玩家的声誉建立行为。它还可以准确预测长期玩家的行动频率，不仅适用于所有均衡策略，也适用于所有均衡最佳对策。这意味着该预测可以通过观察平衡作用的实现路径来检验。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This award funds research in economic theory. The PI wants to advance the use of game theory to understand how strategic decision makers build and maintain reputations. The research focuses on understanding how a long lived individual, firm, or organization will make decisions when it faces a series of short lived opponents who have less information about relevant facts. The project will focus on developing specific predictions that can be tested with data. The research seeks to answer specific questions. First, when can decision makers sustain reputations? Second, how will decision makers behave when reputation building is costly and monitoring is noisy? Third, how can we best test the predictions of reputation models given limited availability of data on long-term interactions? The principal investigator will answer these questions by using monotone methods. The first project introduces interdependent values to the study of reputations. This expands the applicability of reputation results to a richer set of informational environments. The project will give robust predictions on player’s behaviors in addition to their payoffs. The second project will examine whether and how interdependent values can contribute to the sustainability of reputations. The goal is to relate long run outcomes (such as reputation sustainability) to primitives, including the supermodularity or submodularity of stage-game payoffs, and players’ beliefs about the payoff relevant state. The third project develops an alternative framework to study reputations. This framework may help us better understand players’ reputation building behaviors. It also can yield sharp predictions on a long-run player’s action frequencies that apply not only to all of his equilibrium strategies, but also to all equilibrium best replies. This means that the prediction can be tested by observing a realized path of equilibrium play.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.