Understanding and measuring bivariate risk attitude: what can we learn from concordance?

In past years the study of the impact of risk attitude among risks has become a major topic, in particular in Decision Sciences. Subsequently the attention was devoted to the more general case of bivariate random variables. The first approach to multivariate risk aversion was proposed by de Finetti and Richard and it is related to the bivariate case. More recently, multivariate risk aversion has been studied by Scarsini. Nevertheless even if decision problems with consequences described by more than two attributes have become increasingly important, some questions appear not completely solved. This paper concerns with a definition of bivariate risk aversion which is related to a particular type of concordance: a bivariate risk averse Decision Maker is a Decision Maker who always prefers the independent version of a bivariate random variable to the random variable itself