学习与沟通以达成一致同意
Learning and Communication Towards Unanimous Consent
AI总结:
本文研究了在有限承诺下,委托人如何通过测试和沟通实现代理人一致同意的最优策略,发现阈值测试和区间测试在不同偏好对齐情况下各有优势。
摘要:
一个委托人和一个代理人可以基于一致同意启动一个项目。他们的项目收益取决于一个基础状态,而代理人私有地知道自己的偏好。委托人可以进行测试以了解状态,然后与代理人沟通,但具有有限的承诺能力,因为她可能误报她的发现。我们证明有限的承诺使二元测试最优。此外,当玩家的偏好正向一致时,最优的测试是阈值测试。当他们的偏好负向一致时,最优的测试是区间测试或尾部测试,这取决于代理人的相对风险态度。此外,委托人可以通过提供一系列测试来筛选代理人,这在无论代理人类型空间的复杂性如何时都具有简单的结构。
英文摘要:
A principal and an agent can launch a project under unanimous consent. Their individual payoffs from the project depend on an underlying state, and the agent privately knows his own preference. The principal can conduct a test to learn about the state and then communicate with the agent, but has limited commitment, as she may misreport her findings. We show that limited commitment makes binary tests optimal. Moreover, when players' preferences are positively aligned, the optimal test is a threshold test. When their preferences are negatively aligned, the optimal test is either an interval test or a tail test, depending on the agent's relative risk attitude. Additionally, the principal can benefit from screening the agent through a menu of tests, which admits a simple structure regardless of the complexity of the agent's type space.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。