Risk apportionment via bivariate stochastic dominance

This paper extends to bivariate utility functions, Eeckhoudt et al.’s (2009) result for the combination of ‘bad’ and ‘good’. The decision-maker prefers to get some of the ‘good’ and some of the ‘bad’ to taking a chance on all the ‘good’ or all the ‘bad’ where ‘bad’ is defined via (N,M)-increasing concave order. We generalize the concept of bivariate risk aversion introduced by Richard (1975) to higher orders. Importantly, in the bivariate framework, preference for the lottery [(X̃,T̃);(Ỹ,Z̃)] to the lottery [(X̃,Z̃);(Ỹ,T̃)] when (X̃,Z̃) dominates (Ỹ,T̃) via (N,M)-increasing concave order allows us to assert bivariate risk apportionment of order (N,M) and to extend the concept of risk apportionment defined by Eeckhoudt and Schlesinger (2006).