Can We Have a Sensible Definition of Collective Risk Aversion?

Abstract

This paper studies how society should be risk-averse accordingly to how individuals are risk-averse. We consider two axioms, one that if everybody prefers one risky prospect over another so should the society, the other that if everybody is more risk-averse so should the society. We show that the two axioms imply dictatorship, in the sense that the social ranking over risky prospects has to be always identical with one individual's risk preference in each equivalence class of risk preferences which yield the same profile of ordinal preferences over deterministic outcomes.