题　目：Information Design of a Delegated Search

嘉　宾：Yangge Xiao, Ph.D. Candidate, National University of Singapore

主持人：花成 副教授 上海交通大学安泰经济与管理学院

时　间：2023年11月17日（周五）14:00-15:30pm

地　点：安泰经济与管理学院A503

内容简介：

A principal (e.g., homebuyer, pharmaceutical company) delegates a sequential search task (e.g., house hunting, drug development) over a finite horizon to an agent (e.g., real estate agent, biotech startup). While the principal has the right to stop the search, she cannot force the agent to continue it as the agent bears the search cost. Upon termination, the search payoff is split between the principal and agent according to a pre-specified proportion. Novel in our setting, only the principal but not the agent can evaluate each search outcome. Leveraging this informational advantage, the principal designs a policy to decide whether to continue the search and, if so, what information to provide to the agent in each period. We formulate the principal’s problem using a dynamic information design framework and obtain a complete analytical characterization of her optimal policy. The principal does not need to stop the search preemptively and her optimal information policy is an easy-to-implement threshold policy, prescribed by a sequence of deterministic, descending acceptance standards, one for each period. The agent is recommended and voluntarily willing to continue the search if and only if the current termination payoff fails to meet that period’s standard.

When the search results are recallable, the optimal policy features a regime change: the principal sets the acceptance standard at the highest possible level up to a cutoff period, effectively silencing all informative communication. The agent keeps searching up to that cutoff period, after which the principal gradually dial down the acceptance standards by equating the agent’s marginal cost and return from an additional search in that period, independently of other periods. In contrast, when the search results are not recallable, the acceptance standards are determined recursively across different periods as the optimal stopping thresholds that the principal would employ should she conduct each search by herself at a shadow cost. The shadow cost signifies how difficult it is for the principal to persuade the agent. All acceptance standards are informative, underscoring the importance of timely feedback to agent for nonretrievable search processes.

Our findings provide important managerial implications: for innovation-driven searches, like R&D or academic research, where search outcomes accumulate, search incentives should be front-loaded by adopting a hands-off approach early in the search, and only nudging the search later on; for searches like talent recruitment, where outcomes do not accumulate, acceptance standards should be loosened up as the search progresses so as to back-load the search incentives.

演讲人简介：

Yangge Xiao is a Ph.D. candidate at the Institute of Operations Research and Analytics, National University of Singapore. Prior to NUS, she obtained both the B.Eng. and M.Eng. degree in Management Science and Engineering from Huazhong University of Science and Technology.  Her research interests mainly focus on mechanism design and information design with applications to search models.

欢迎广大师生参加！