Assessing risky social situations

This paper reexamines the welfare economics of risk. It singles out a class of criteria, the “expected equally distributed equivalent,” as the unique class that avoids serious drawbacks of existing approaches. Such criteria behave like ex post criteria when the final statistical distribution of well-being is known ex ante and like ex ante criteria when risk generates no inequality. The paper also provides a new result on the tension between inequality aversion and respect of individual ex ante preferences, in the vein of Harsanyi’s aggregation theorem