研讨论文

Learning It Sooner Or Later

主讲人

姜渭（北京大学光华管理公司2021级博士研究生）

论文简介

This paper examines an information design problem for a sophisticated present-biased agent facing a task with an uncertain consequence if uncompleted by a deadline. The agent can perform the task at any period before the deadline. A principal, aiming to maximize the agent's long-run welfare, commits to a dynamic information disclosure rule regarding the consequence. We find that the optimal rule generally features "learning it sooner or never': information is disclosed exclusively at the first period.

Furthermore, as the number of available periods for task completion increases, more information is optimally disclosed. This optimal rule can be readily implemented if the long-run self can commit to the information structure at period 1. Both self-control concerns and decision-making considerations are pivotal to these findings. We also identify two scenarios where the principal might optimally delay information disclosure: first, if the agent is naive and cannot anticipate future procrastination, thereby lacking self-control concerns; second, if the principal's objective is solely to maximize task completion probability, rather than the agent's long-run welfare, thus eliminating decision-making concerns.

点评老师

张志祥、尹训东

时间

10月20日 (14：30-16：00）

线下地点

学术会堂712

活动对象

太阳成集团tyc9728员工

人数规模

30人

主办单位

太阳成集团tyc9728

撰稿：姜渭

审稿：尹训东

来源：太阳成集团tyc9728

审核：王颍