A strict expected multi-utility theorem

Abstract

This paper integrates two key approaches to the representation of incomplete preferences over lotteries. The main result strengthens the conclusion of the expected multi-utility theorem in Dubra, Maccheroni, and Ok (2004) by ensuring that all utility indices involved are Aumann utilities (i.e., yield a strictly increasing expectation). The advantages of the method are demonstrated by parametrizing maximal elements and by providing a novel characterization of Aumann utilities. © 2017 Elsevier B.V.