Key words: Bayesian persuasion, swing voter’s curse, networks, voting, inducible distributions, informativeness

"Bayesian Persuasion: Elections and Informativeness of Beliefs"

This dissertation mainly focuses on situations in which a sender (e.g. an incumbent party) communicates with multiple receivers (e.g. voters) in order to persuade them to vote in favour of a proposal. We focus on the sender’s perspective and analyse how the sender can increase the probability of implementing the proposal. The first chapter focuses on such a situation, in which the proposal is either “good” or “bad” for society. While the sender only cares about passing the proposal, the receivers want it to pass only if it is good for society (e.g. a drug company trying to get an approval from the FDA). The study characterises the equilibrium in which sincere voting is optimal for each voter. The research addresses a common criticism to such communication models and assumes that receivers not only communicate with the sender, but among themselves as well. As the sender’s payoff in such models often depends on beliefs of the receivers, the last chapter refrains from considering elections, and focuses on the “feasible” beliefs of the receivers, as well as the informativeness of communication protocols inducing these beliefs.

Click here for the live stream.