CMStatistics 2021: Start Registration
View Submission - CFE
Title: Stochastic dominance with uncertain preferences Authors:  Tommaso Lando - University of Bergamo (Italy) [presenting]
Abstract: The theory of stochastic dominance provides a model for predicting a decision maker's choice between pairs of uncertain prospects, without having a precise knowledge of her utility function. To improve such a modelization in terms of flexibility, some recent works establish continua of dominance relations, in which a decision maker's preferences are basically described by a risk aversion or a risk attraction parameter (or both). We study a general theory of stochastic dominance which introduces randomness into such models, by assuming that these parameters, representing preferences, are unknown random variables, to be inferred from data. Under some conditions, this new model makes it possible to estimate the probability of a particular choice between two uncertain prospects and to determine whether it could be expected that the decision-maker will take some decision, given her past behaviour.