A NON-PARETIAN WELFARE ECONOMICS 285

following: economic systems are incapable of generating considerable amounts of predominantly useless discoveries over a long period of time. In other words, the only systems capable of sustaining a significant discovery process through time are also the systems that successfully utilize these discoveries. Without a ‘healthy’ discovery process, the capabilities of the system will tend to shrink on almost all fronts, thus curtailing discoveries. It may be possible to conceive of an intuitively undesirable system which can continually generate new discoveries in a few sectors. (The military establishment of the Soviet Union may be an example here.) Such systems, however, are incapable of generating significant discoveries across a wide range of sectors for a considerable period of time.

Empirical reasons might suggest that the discovery standard would not necessarily approve a misguided policy which subsidized discovery per se. Systems that do not pursue such policies will outdo (with respect to discovery, among other factors) systems that do. Of course, if this empirical claim were false, discovery would be a poor choice for a welfare standard. In this case, an alternative standard(s) should be chosen that does not suffer from this problem.

Using history to choose a good standard attempts to avoid the dilemma of either collapsing into Paretian theory or approving of our welfare standards per se. Our standards are chosen, not because they are always good, but because empirical analysis indicates that systems which are very successful with respect to the standard(s) are not likely to violate our intuitions concerning what a ‘good result’ would be.

Aggregation

Historically based non-Paretian approaches to welfare theory take a different view of the aggregation difficulties which have beset a number of other forms of welfare theory. Paretian welfare economics, for instance, insists that all propositions about societal welfare be derivable from the underlying individual preference functions. Aggregation can only proceed when such constructs as the representative consumer are applicable. Needless to say, this implies that rigorous aggregation is, in practice, almost always impossible.

A number of non-Paretian alternatives also insist upon foundational approaches to aggregation which imply that all statements concerning ‘societal welfare’ must be unambiguously derivable from a set of underlying postulates. Kirzner (1986, pp. 2 and 19), for instance, chooses the underlying postulates of subjectivism, methodological individualism and an emphasis on dynamic processes. These postulates impose stringent requirements upon the task of the welfare theorist. Kirzner notes that:

we shall refuse to recognize meaning in statements concerning the ‘welfare of society’ that cannot, in principle, be translated into statements concerning the individuals in society (in a manner which does not do violence to their individuality)…we shall not be satisfied with statements