Targeted Information Design

Consumers face partial uncertainty about their valuation. Through targeting information, platforms change demand and optimal prices giving them control over market and welfare outcomes. How should a platform that is not the only source provide information? How does total welfare change with the platforms objective?

Comparison of Experiments through Summary Statistics

Work in information design typically assumes that the designer has full control: She is the sole source of information and can fully reveal the truth. In such settings, when agents' decisions depend only on the expected state, the problem simplifies to optimizing over the distribution of expected states rather than the signaling structure itself. What happens when the designer lacks complete control?

Robustness Foundations B-IC

Classic results in mechanism design often rely on strong assumptions about what the designer knows regarding agents' beliefs. The robustness literature seeks to relax these assumptions. Ollár and Penta (2017) formalize this with the notion of belief restrictions, which specify exactly what is assumed about agents' beliefs. They study implementation under a generalized notion of ex post equilibrium. What changes when we instead work with Bayes–Nash equilibrium, a solution concept that allows the designer to elicit and make use of agents' beliefs directly?