The Welfare Economics of Ex Post Optimality

The analysis of welfare economics under uncertainty has typically focused on the attainment of optimal equilibria from an ex ante viewpoint. Normative judgments as to the desirability of a particular resource allocation are made with respect to the Pareto criterion based on the ex ante preferences of individuals. It is well known that, under a suitable set of assumptions, these preferences can be represented by a set of utility functions, one for each state of nature, which represent an individual's tastes for alternative consumption bundles and his subjective probability distribution over alternative states of nature. Consumer sovereignty under uncertainty implies that alternative resource allocations should by judged in a manner that takes account of both the tastes and the probability beliefs of the individuals in the economy. Some economists, for example Diamond (1967), Dreze (1970), Feiger (1976), Guesnerie and de Montbrial (1974), Hammond (1976), Harris (1976b, 1978), Mirrlees (1974) and Starr (1973), have suggested, however, that the subjective probability beliefs of individuals should not "count" in making social decisions about resource allocation. This raises the question of what should count? Suppose we take the view that what should count in making social decisions are the tastes of individuals in those states of nature that are realized. This leads to the notion of ex post efficiency first explored by Starr (1973). Under a set of particular assumptions about probability beliefs, tastes and technology, this ex post view can be reconciled with an ex ante view. In most circumstances however, the ex post criterion is not sufficient to provide judgments about intertemporal resource allocation decisions under uncertainty. In these cases, one is explicitly forced to adopt a social welfare function in which individuals' tastes "count" but their probability beliefs do not. Furthermore, the specification of the social welfare function implies the existence of a set of social probability beliefs. We will refer to such social welfare functions as "ex post social welfare functions". We explore the implications of the divergence between ex ante and ex post views of social welfare for certain issues in the welfare economics of uncertainty in this paper. The format of the paper is as follows. In Section I we discuss the conditions under which a divergence between ex ante and ex post optimality occurs and why these conditions are plausible. The conflict between the two types of optimality is illustrated for an exchange economy, and we note that there exist market-like mechanisms that may be applied ex ante to achieve expost objectives. Section II then examines the emergence of certain institutions that may be interpreted as attempts by society or government to achieve ex post optimality for some specific cases of resource allocation under uncertainty.