Social Aggregation Without the Expected Utility Hypothesis

This paper investigates the possibilities for satisfaction of both the ex-ante and ex-post Pareto principles in a general model in which neither individual nor social preferences necessarily satisfy the Expected Utility Hypothesis. If probabilities are subjective and allowed to vary, three different impossibility results are presented. If probabilities are 'objective' (identical across individuals and the observer), necessary and sufficient conditions on individual and social value functions are found (Theorem 4). The resulting individual value functions are consistent not only with Subjective Expected Utility theory, but also with some versions of Prospect Theory, Subjectively Weighted Utility Theory, and Anticipated Utility Theory. Social Preferences are Weighted Generalized Utilitarian and, in the case in which individual preferences satisfy the Generalized Bernoulli Hypothesis, they are Weighted Utilitarian. The objective-probability results for social preferences cast a new light on Harsanyi's Social Aggregation Theorem, which assumes that both individual and social preferences satisfy the Expecte Utility Hypothesis.