Skip to content

Glossary · Game Theory

Bayesian Persuasion

also: information design · persuasion game · strategic disclosure · unraveling · disclosure game

Definition

Bayesian persuasion is the design of an information structure by a sender who commits to a signal before observing the state, in order to move a rational receiver's actions. Kamenica and Gentzkow (2011) characterized the optimum through concavification of the sender's value function. It formalizes when partial disclosure of reviews, ratings, or prices beats full transparency.

The model separates two ideas that everyday talk of persuasion runs together: the sender cannot lie, and the receiver is fully rational, yet the sender still gains by choosing what to reveal. The classic precursor is the unraveling result of Grossman (1981) and Milgrom (1981): when disclosure is costless and verifiable, silence is read as bad news, so every type above the worst discloses and the market reveals everything. Experiments by Jin, Luca, and Martin (2021) found unraveling fails in practice because receivers are insufficiently sceptical of silence, which is exactly the gap a persuader exploits. Review platforms, rating aggregators, drip pricing, and product-comparison pages are all information designs in this sense, and the theory gives a precise account of which of them a rational buyer should distrust.

Essays on this concept