Working paper

Optimal Public Information Disclosure by Mechanism Designer

Takuro Yamashita


We consider a mechanism design environment where a principal can partially control agents' information before they play a mechanism (e.g., a seller disclosing quality information). Assuming that the principal can ex ante commit to his disclosure policy, this is a Bayesian persuasion problem with an endogenous payoff function as a consequence of optimal mechanism design. Wefirst show that, if the principal's and agents' information are independent or affiliated, and if the implementable set of (non-monetary) allocation rules is invariant to disclosure policies, then it is optimal for the principal to disclose all the relevant information. In case of negative correlation or in case the set of implementable allocation rules varies with disclosure policies, then full disclosure is not necessarily optimal. We then characterize the optimal (non-full) disclosure policy under additional assumptions, which, viewed as a Bayesian persuasion problem, provides a solution to a novel class of environments.


Takuro Yamashita, Optimal Public Information Disclosure by Mechanism Designer, TSE Working Paper, n. 18-936, July 2018.

See also

Published in

TSE Working Paper, n. 18-936, July 2018