时间：9月25日（周四）10:30 a.m.-12:00

地点：北京大学经济学院302

主持老师：吴泽南，石凡奇（经院）；胡岠（国发院）

参与老师：胡涛，吴泽南，石凡奇（经院）；汪浩，胡岠，邢亦青（国发院）；翁翕，刘烁（光华）

Talk 1

主讲人：姜渭（北京大学光华管理学院）

题目: Learning it Sooner or Later

摘要：Suppose a present-biased agent can perform a task at any period before the deadline but she is uncertain about the consequence if the task is not completed. We study the information design problem where a principal, who wants to maximize the long-run welfare of the agent, can commit a rule to dynamically disclose information about the consequence. We show that the optimal rule always features ``learning it sooner or never'' --- information is disclosed only at the first period. When the agent has more periods to perform the task, more information will be disclosed. Such optimal rule can be implemented by a sequential persuasion game where at each period, the agent can only commit the information structure for the next period. Our results are robust to the payoff structure with immediate rewards and delayed costs.

主讲人介绍：Wei Jiang is a Ph.D. candidate of economics at Guanghua School of Management, Peking University. His research interests include learning theory, social network and search theory.

Talk 2

主讲人：王禛阳（北京大学光华管理学院）

题目: Disclosure and Manipulation in Reputation System

摘要：Online reputation systems integrate soft information (ratings) and hard information (reviews) to mitigate information asymmetry in e-commerce markets, yet remain vulnerable to strategic manipulation by sellers. While the extant literature primarily examines these feedback mechanisms and their manipulation in isolation, the welfare implications of manipulation in joint systems that combine both channels remain theoretically underexplored. We develop a game-theoretic model to analyze three feedback system designs: rating-only, review-only, and joint feedback systems, both with and without seller manipulation. Our analysis yields a counterintuitive result: in a joint system, permitting simultaneous manipulation of both ratings and reviews can generate higher consumer welfare than prohibiting manipulation entirely or restricting it to only one channel. These findings challenge the prevailing view that manipulation uniformly reduces welfare and provide novel insights for platform governance, suggesting that optimal reputation system policies must consider the endogenous interactions between different feedback mechanisms.

主讲人介绍：Zhenyang Wang is a Ph.D. candidate of economics at Guanghua School of Management, Peking University.