 |
Pietro Dindo est professeur associé à l'Université Ca' Foscari de Venise où il enseigne la théorie économique
et l'économie financière.
Nous avons le plaisir de l'accueillir le 10 octobre 2019 (14 heures - salle 101 - Bâtiment Veil - DU Saint-Martin-d'Hères) pour la présentation d'un article co-écrit avec Filippo Massari : Learning New Models from Prices.
Abstract:
We consider dynamic financial economies where agents are Bayesian learners, have heterogeneous prior beliefs,
and trade on such differences. We assume that agents may assign a positive initial weight to the model stating
that an appropriate transformation of state prices reveals the truth and provide conditions under which such learning scheme leads to the truth. Although the result does not hold when all priors are misspecified, because no agent puts positive initial weight on the truth, we show that slowing down the learning rate, and having positive initial weight on the possibility that markets are accurate, increases both individual and market accuracy. We show how risk preferences play a role and clarify that results hold because the market selects among agents as a Bayesian who learns on a set of models derived from agents’ beliefs, as depending on their risk preferences and the aggregate endowment process.
Page web
|
