Uncovering Latent Stochastic Dominance Relations Using Prior Rankings

Abstract

The discriminatory power of Stochastic Dominance analysis is enhanced using side information about the ranking of external prospects. Restricting risk preferences to be consistent with known rankings strengthens the stochastic order by uncovering latent dominance relations and reducing the number of undominated alternatives. To operationalize the proposed approach, a tractable formulation is derived in terms of linear constraints on the Lower Partial Moments of the distribution functions. Representative applications to wellbeing analysis and portfolio optimization show substantial gains in ordering strength from using plausible side information about the ranking of simple prospects.

Keywords: Stochastic Dominance, Side information, Linear Programming, Wellbeing analysis, Portfolio optimization

JEL Classification: C61, D81, G11