ABSTRACT
While formal demonstration of the general theorem of second best 1 is, in substance, unassailable since, like other impossibility theorems, 2 its negative corollaries rest securely on the posited absence of empirical limitations, it must be admitted that its forceful presentation several years ago disturbed us somewhat and carried just so much further the process of disillusion with conventional welfare economics. Not that we had any right to be disturbed, for it is clear enough now that in talking of optimum conditions we were, in any case, saying precious little; no more, in fact, than (assuming the appropriate degree of differentiability in our functions) that a constrained maximum entails necessary conditions. This much being conceded, the second-best theorem does no more than point out that, if additional constraints are imposed, the necessary conditions for a maximum are in general more complex. 3 The obvious corollary follows that, in order to identify a maximum position in these conditions, it is necessary to forsake the optimum conditions that are strictly relevant only to the simple case of a single and familiar constraint.
