In this paper the authors examine the theoretical and empirical issues that arise for a public-sector decision maker who wishes to use the target population’s preferences as an input for designing new products and services. The market literature suggests using essentially the same approach as that for the private-sector problem, with one exception. Instead of maximizing profits, share or volume, the sum-of-benefits to society is maximized. However, the social welfare literature has suggested several other measures of welfare and shows that no single welfare function dominates all others. Therefore, to choose an appropriate welfare function to maximize the decision maker must first examine the theoretical and empirical similarities between the alternatives. For six prominent welfare functions, the authors examine the theoretical differences in terms of the notion of fairness promoted and the kind of preference data required. Empirically, they examine the similarities in social rankings and first choices induced by these welfare functions under various preference configurations. The authors also examine their sensitivity to measurement error and preference normalization.
