Career: Reputation with Limited Information, Theory and Applications

该奖项资助经济理论方面的研究。首席研究员计划使用博弈论来理解经济主体（如公司和政治家）在长期关系中采取社会期望行动的动机。该研究试图回答以下具体问题。首先，当市场对企业过去行为的信息有限时，企业是否有动力提供高质量的产品？第二，网络平台应该向消费者提供什么样的信息，以激励卖家建立提供高质量产品的声誉？第三，研究人员能否对企业的行为做出可靠的预测，并用数据进行检验？首席研究员将通过使用重复游戏、声誉和学习理论中的新方法来回答这些问题。第一个项目分析了消费者观察其他消费者选择的能力（即观察学习）如何影响每个卖家为提供高质量而建立声誉的动机。消费者的信息被一个随机网络所总结，这个随机网络决定了消费者的行为是否能被他们的每一个继任者所观察到。研究结果确定了观察学习强化声誉激励的网络结构以及导致声誉激励失效的网络结构。第二个项目分析消费者没有关于公司历史的详细信息的情况。该分析揭示了长期社会记忆、市场信息的粗化、消费者信息的异质性对卖家维持声誉的动机和消费者福利的影响。第三个项目分析了长期存在的代理人（例如，卖家、政治家）可以操纵他们过去的记录的情况，例如，他们可以以一定的代价从他们的记录中删除信号。这是由于在线平台上的应用程序，卖家可以贿赂消费者，以换取撤下差评。主要结果表明，当卖家的寿命足够长时，他们建立声誉的回报就会被完全抹去，即使市场怀疑他们可能以很低的可能性抹去记录。研究还表明，玩家维持合作的动机并不一定会随着预期寿命的延长而增加；这一发现与文献中关于重复游戏的标准预测形成了对比。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 principal investigator plans to use game theory to understand the circumstances under which economic agents (such as firms and politicians) have incentives to take socially desirable actions in long-term relationships. The research seeks to answer the following specific questions. First, does a firm have incentives to supply a high quality product when the market has limited information about its past behavior? Second, what kind of information should online platforms provide to consumers in order to motivate sellers to build a reputation for supplying high quality product? Third, can researchers deliver robust predictions on firms’ behaviors that can be tested using data?The principal investigator will answer these questions by using novel methods in the theory of repeated games, reputation, and learning. The first project analyzes how consumers’ ability to observe other consumers’ choices (i.e., observational learning) affects each seller’s incentives to build a reputation for supplying high quality. Consumers’ information is summarized by a stochastic network, which determines whether a consumer’s action can be observed by each of their successors. The results identify network structures under which observational learning strengthens reputational incentives as well as those that cause reputational incentives to break down. The second project analyzes situations where consumers do not have detailed information about the firm’s history. The analysis sheds light on the effects of long social memories, the coarsening of market information, the heterogeneous quality of consumer information on the seller’s incentives to sustain a reputation, and on consumers’ welfare. The third project analyzes situations where long-lived agents (e.g., sellers, politicians) can manipulate their past records, for example, they can erase signals from their records at some cost. This is motivated by applications to online platforms in which sellers can bribe consumers in exchange for taking down the negative reviews. The main result shows that, when sellers are sufficiently long-lived, their returns from building reputations are entirely wiped out, even when the market suspects that they may erase records with a low probability. It also shows that players’ incentives to sustain cooperation do not necessarily increase with their expected lifespans; and this finding stands in contrast to the standard predictions found in the literature on repeated games.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.