Risk, Impersonality, and the Social Welfare Function

In the recent literature on welfare economics it has been held by some writers (for example, Arrow, 1964, pp. 9-10, and Friedman and Savage, 1948, p. 283, n. 11) that individual preferences among risky prospects have little or no relevance for constructing a pattern of social preferences for non-risky social states. In contrast, other writers have expressed the opinion that not only are individual preferences among risky prospects relevant, but also they have a certain logical connection with the intuitive concept of social welfare. The purpose of this note is to examine some of these formulations, to clarify the basic concepts involved, and to assess their bearing on the Arrowian problem of constructing a social ordering by aggregating the individual preferences. The formulations examined are those of Harsanyi (1953, 1955)1 and Vickrey (1960). Section I gives the main propositions advanced by these writers, section II analyzes their ethical bases, and section III suggests how some of Harsanyi's concepts can be used to deal with Arrow's problem. Finally in section IV we indicate some of the difficulties likely to be met in our attempt to use Harsanyi's concepts for deriving a social welfare function of the Arrow type.