Robust predictions in games with incomplete information (Q2864814)

This paper is rather a monograph. It is interesting as it proposes procedures that work adequately in any case, as they do not rely on the game parameters, which are usually unknown. The independence of the information on the game provides confidence on the robustness of the proposals for leading with the analysis of games. The authors cope with the problems assuming symmetry, normality and a quadratic loss function. Hence, the study of Bayes equilibria and the associated properties and behavior depends heavily on these assumptions. The kernel of the paper is a study of the Bayes correlated equilibrium and the determination of how it works within the constructed framework. A comparison with Nash equilibria establishes relationships that sustain a certitude that using the proposals is to be useful dealing with in practical situations. That is the case for searching for improving welfare. 14 propositions support the claims on the robustness of the proposed methods. The role of the models seems to be useful for identification but not that successful for prediction.