Justice Preferences and the Arrow Problem

Abstract

Arrow showed that there is no general way to aggregate non-interpersonally comparable preferences or welfare into either a sensible social choice or a social welfare measure. With majority rule the problem manifests itself as voting cycles. The standard response to this problem has been developing `spatial models' built on restricted preferences (or welfare). We develop an alternative family of solutions. By assuming a culturally accepted conception of justice within a utility function, we establish the possibility of sensible aggregate choice implementable via majority rule. Various assumptions regarding the form the utility function are discussed. Conditions for a Condorcet winner in a problem of pure redistribution are derived for a number of models. Some of the implications of this perspective for the theory of democracy are considered. Developing a normatively interesting social welfare function may require introducing normative concerns into the preferences of the individuals rather than just into the properties of the aggregation system.