北大经院工作坊第918场
Persuading While Learning
微观理论经济学工作坊
主讲人：施贤文（多伦多大学教授）
主持老师：
（北大经院）吴泽南、石凡奇
（北大国发院）胡岠
参与老师：
（北大经院）胡涛
（北大国发院）汪浩、邢亦青
（北大光华）翁翕、刘烁
时间：2024年6月26日（周三）10:30-12:00
地点：必赢线路检测3003no1302会议室
主讲人简介：
Xianwen Shi is a Professor of Economics at the University of Toronto. He is an expert in mechanism design and information design, known for his work on mechanism design without transfers, with limited commitment, with endogenous information acquisition, or with endogenous information disclosure, as well as for his work on contest design. He has published papers on leading economic journals such as American Economic Review, Econometrica, Journal of Political Economy, and Review of Economic Studies. His current research focuses on information disclosure and manipulation in strategic dynamic contexts.
摘要：
We introduce a dynamic persuasion model with a long-lived sender and a sequence of short-lived receivers. All players share a common prior belief about a persistent underlying state which is either good or bad. The sender receives information over time and can commit to a dynamic revelation policy at the beginning. In each period, the sender privately receives a noisy signal about the state and reveals a potentially garbled signal to the short-lived receiver who arrives in this period. The short-lived receiver observes all information revealed by the seller in the past and update their belief about the state being good. If their belief is above a fixed cutoff they act; otherwise they do not act. The sender would like to induce as many receivers to act as possible and values early actions more than later actions. To characterize the optimal policy, we introduce a new class of Markov martingales, so-called “Blackwell order preserving martingales,” which have the property that the Blackwell information order is preserved by the associated probability kernel. All belief martingales generated by conditionally independent signals are Blackwell martingales. We prove that, if the seller’s sequence of private information can be summarized by a Blackwell order preserving martingale, then the optimal policy must persuade an interval of beliefs surrounding the cutoff for action. We apply this characterization to show that, if the sender is sufficiently impatient and receives information according to a random walk, then full transparency is approximately optimal.
供稿：科研与博士后办公室
美编：初夏
责编：度量、雨禾、雨田
