Aggregate Demand, Real National Income, and the Compensation Principle

Abstract

government accounts, imperfectly competitive markets), a basic tenet of the new welfare economics has seemed to be that if all markets were perfetly competitive, and if we had only final good consumption to worry about, then an unambiguous and essentially value-free index number comparison of real national income would be possible. We shall demonstrate in this paper that such faith has been misplaced. More specifically, we shall show that an index number comparison implies an unambiguous Compensation Principle comparison if and only if individual preference orderings are identical and homogeneous. A problem very closely related to the problem of the evaluation of real national income, and of great scientific importance in economics, revolves around the question of what observable implications for market demand functions are implied by our theory of individual consumer choice. It has long been known that the assumption that the aggregate demand correspondence is actually a (singlevalued) function is overly strong, in the absence of restrictions on the distribution -of income (and we verify this fact in Section 3). However, one might suppose that a fairly wide class of individual preference orderings give rise to aggregate demand correspondences which satisfy the revealed preference conditions (more specifically, Richter's Congruence Axiom [26]) even in the absence of restrictions ion the admissible income distributions. In Section 4 we show, however, that the class of continuous preference orderings which are convex to the origin (Definition 2.8, below), and which yield aggregate demand correspondences satisfying Richter's Congruence Axiom exactly coincides with the class of m-tuples (where m is the number of consumers) of identical homogeneous preference orderings.