<p>In Bayesian persuasion, Receiver simply plays an action after Sender’s public signaling. However, in some applications, it seems natural that Receiver could elicit more information from Sender by offering a screening contract. We study its economic implications, mainly in a stylized binary, quasilinear environment. In the first model where Sender acquires full information as his private information, the public signal is less informative than that in Bayesian persuasion, suggesting that the public and private signals are “substitutes”. Meanwhile, both Sender and Receiver are better off than in Bayesian persuasion: Sender prefers “being further screened”. In the second model where Sender jointly designs both public and private signals, he sometimes finds it more profitable to have less precise private information. The outcome is most efficient (both in terms of information and welfare) with the first model, less so with the second model, and least in Bayesian persuasion. This suggests that economic predictions based on the standard Bayesian persuasion model might be biased toward overly inefficient outcomes.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Bayesian persuasion followed by receiver’s mechanism design

  • Takuro Yamashita,
  • Shuguang Zhu

摘要

In Bayesian persuasion, Receiver simply plays an action after Sender’s public signaling. However, in some applications, it seems natural that Receiver could elicit more information from Sender by offering a screening contract. We study its economic implications, mainly in a stylized binary, quasilinear environment. In the first model where Sender acquires full information as his private information, the public signal is less informative than that in Bayesian persuasion, suggesting that the public and private signals are “substitutes”. Meanwhile, both Sender and Receiver are better off than in Bayesian persuasion: Sender prefers “being further screened”. In the second model where Sender jointly designs both public and private signals, he sometimes finds it more profitable to have less precise private information. The outcome is most efficient (both in terms of information and welfare) with the first model, less so with the second model, and least in Bayesian persuasion. This suggests that economic predictions based on the standard Bayesian persuasion model might be biased toward overly inefficient outcomes.