Stability and Collective Rationality

Abstract

THE PRINCIPLE of pairwise comparisons is a well established one in social choice theory. The idea of collective rationality-that social decisions be made consistently with the maximization of some ordering of the alternatives-is the most familiar application of this principle, and indeed, is a defining property of the social welfare functions of Bergson (1938) and Samuelson (1947), and of Arrow (1951). As is well known, the idea of collective rationality involves comparing every social state to every other social state in terms of the welfare assigned by those states to a given group of individuals, and declaring a social state to be fair or good if it is good when considering all such pairwise comparisons with other social states. Alternatively, one could imagine a social decision process based on comparing every individual to every other individual in terms of the welfare assigned to those individuals by a given social state, and declaring a social state to be good or fair if it is fair when considering all such pairwise comparisons of individuals. A social decision process in this spirit has been suggested by Harsanyi (1959),