Aggregation of Misspecified Experts

We investigate how a cautious decision maker might aggregate various opinions of experts who face model misspecification. We first consider a more general case in which both the decision maker’s and experts’ preference relations are variational introduced by Maccheroni et al. (2006). We show that, under a Pareto-type axiom and a cautious axiom, the decision maker is the least ambiguity averse among those who are more ambiguity averse than each expert. We then use Cerreia-Vioglio et al. (2022)’s criterion, a special case of variational preferences, to capture model misspecification. Under our axioms, the decision maker is also concerned with model misspecification: the decision maker is very cautious by considering all the experts’ structured models as his best guesses of the correct model. Moreover, we formally provide a distance function, which is based on experts’ opinions, the decision maker might use