Axiomatic Foundations of Bayesian Persuasion

TL;DR

This paper develops axiomatic foundations for Bayesian persuasion, introducing models where the sender can influence the agent's bias after costly information acquisition and providing methods to infer costs from observable data.

Contribution

It presents novel axiomatic models of Bayesian persuasion with bias manipulation and offers an elicitation method based on observable menu-choice data.

Findings

Characterization of Bayesian persuasion models with bias steering

Method to infer information costs from observable choices

Framework for managing agent bias through costly information

Abstract

In this paper, we study axiomatic foundations of Bayesian persuasion, where a principal (i.e., sender) delegates the task of choice making after informing a biased agent (i.e., receiver) about the payoff relevant uncertain state (see, e.g., Kamenica and Gentzkow (2011)). Our characterizations involve novel models of Bayesian persuasion, where the principal can steer the agent's bias after acquiring costly information. Importantly, we provide an elicitation method using only observable menu-choice data of the principal, which shows how to construct the principal's subjective costs of acquiring information even when he anticipates managing the agent's bias.