Vassili Vergopoulos (Université Paris 2) – Information and beliefs: a weakly Bayesian approach

We study the relationship between the information that an economic agent acquires and the beliefs he forms in consequence. The standard model based on Bayesian updating and conditional probability may remain too restrictive for economic applications involving bounded rationality. Indeed, an agent sometimes forgets, neglects or distrusts certain features of the information they acquire. Sometimes, they also let subjective attitudes, or moods, affect their information processing. We propose a more general and flexible approach that accommodates such patterns. In this approach, Bayesian updating continues to play a role. But, in a sense to be clarified, it only holds on average. We also explore the implications of this alternative approach in three different directions: (a) value of information, (b) trade under public information and (c) agreement theorem.