ON THE AGGREGATION OF MONEY MEASURES OF WELL-BEING IN APPLIED WELFARE ECONOMICS

Abstract

This article investigates the properties, good and bad, of social evaluations based on four money of well-being or changes in well-being: compensating variations, money metrics, extended money metrics, and welfare ratios. Consistency of social rankings (transitivity, asymmetry of preference), the possibility of incorporating inequality aversion, independence of the choice of reference prices, and the ethics implicit in the evaluations are considered. In addition, these procedures are contrasted with utility aggregation using equivalence scales. Cost-benefit analysis and other applications of welfare economics employ money of well-being or changes in well-being and evaluate economic states of affairs with aggregates of these measures-usually simple sums. The most favored measure is the compensating variation or willingness to pay. It is an index of welfare change for a household or individual. The equivalent variation (another Hicksian consumer's surplus) measures welfare change as well. The Marshallian consumer's surplus, with all its difficulties, is still used, usually as an approximation to the compensating variation (Willig). Other money of the level of well-being are money metric utilities -in both standard and extended forms-and welfare ratios. Compensating and equivalent variations and ordinary money metrics are determined by preferences alone, while extended money metrics and welfare ratios require interpersonal comparisons of utilities. This article surveys the literature on the properties, good and bad, of these measures, using as criteria various aspects of their performance in performing social evaluations. In addition, their performance is contrasted with the use of aggregate of utilities-direct indexes of well-being-implemented with an equivalence-scale methodology. Performance in social evaluation is viewed from several standpoints. First, rationality properties (transitivity, asymmetry of preference) of social binary relations based on the various are discussed. Second, given that states of affairs are ordered consistently, the possibility of incorporating inequality aversion is examined. Because concavity or nonconcavity of the utility functions that money represent has consequences for inequality aversion, concavity is the third area of investigation. Fourth, when indexes depend on a reference price or quantity vector or a reference household type-as money metrics and extended money metrics do-conditions for independence of the social ordering of this choice are investigated. Fifth, the ethics implicit in the evaluations produced with these aggregates are examined. The possibility of incorporating inequality aversion is included because procedures that are restricted to zero inequality aversion are of limited usefulness in societies whose prevailing ethical judgments include a concern (even a small one) for distributive justice. Further, ethically flexible evaluation procedures can themselves contribute to a sensible discussion of the idea of distributional equity and the incentive tradeoff. The setting for these exercises is a general equilibrium one, with marketed private goods and unpriced public or semipublic goods. Two domains for prices and unpriced goods are employed: the first allows prices and the consumption of unpriced goods to differ among households, and the second requires them to be the same for all.