On the Predictive Content of Models of Monopolistic Competition

Perhaps the strongest criticism ever directed at the theory of monopolistic competition is the assertion that the theory yields no testable hypotheses, and that therefore it is not a suitable medium through which one can study those market phenomena which cannot be explained by the theories of pure competition and pure monopoly. This position was taken by Stigler [6] twenty years ago. More recently, this theme has been taken up by G. C. Archibald [1, 2] who has amplified it, and given it a more concrete form of expression. Specifically, Archibald has shown that if a shift parameter, such as a tax parameter, is introduced into the model of the monopolistically competitive firm, then those comparative static properties which are typical of monopolistic competition turn out to be indeterminate. In other words, one cannot, in general, predict the effects of a change in the tax on the optimal levels of such decision variables as advertising and price. While Archibald’s formal results are correct, there seems to be no compelling reason to subscribe to the view that ‘… the difficulties of the “ qualitative calculus” applied to maximising models of the firm are too great to be overcome...’.2 It is the purpose of this paper to suggest one way of surmounting these difficulties — at least partially