Monopolistic competition, the Dixit–Stiglitz model, and economic analysis

Joseph E. Stiglitz – December 2017

Research in Economics

I welcome this opportunity to provide a comment for this Special Issue of Research in Economics honoring forty years since the publication of the Dixit-Stiglitz 1977 paper. I have been pleased both by the way that our simple parameterization has provided a tool that others in a variety of sub-disciplines—growth theory, macro-economics, and international trade—have found so useful. In many ways, I am not surprised—indeed, I was aware of some of these possible applications at the time that we wrote the paper. As I noted in the essay I wrote on the occasion of the 25th anniversary of our paper (Stiglitz, 2004), the standard competitive model—where every firm faces demand curves of infinite elasticity—leads to numerous conundrums. For instance, small open economy could quickly restore itself to full employment by a slight adjustment of the exchange rate—it would easily remedy any deficiency of aggregate demand. In almost all sectors of the economy, firms face downward sloping demand curves, and competition is “imperfect.”

Of course, there are many different forms of imperfect competition. Some forty years before our work, Chamberlin (1933) had posited a simple model where there were enough firms that no firm believed it had any effect on the behavior of others but the products they produced were sufficiently differentiated that each faced a downward sloping demand curve. Moreover, there was free entry, ensuring that the (marginal) firm earned zero profits. His graphical analysis assumed U shaped average cost curves; there were important fixed costs. But fixed costs were explicitly excluded from the standard Arrow-Debreu paradigm: it presented intractable problems both in the proof of the existence of competitive equilibrium and its optimality. Strikingly, until our work and that of Spence (1976), there had been almost no theoretical development, and the common wisdom was motivated by the observation that in monopolistic competition, firms operated at output levels that were smaller than that generating the minimum average costs; average costs could accordingly be decreased if there were fewer firms, each producing more. Hence, it was argued that with monopolistic competition there were too many firms, each producing too little. But this analysis simply ignored the reason for downward sloping demand curves; there was a social value to product variety.

Our work was in part motivated to answer the normative question of the efficiency of markets with monopolistic competition, in a formal model, recognizing that the reason that there are multiple firms is that society values variety— different individuals value different things and/or individuals value a range of products. Our benchmark model, which has proved so useful, established the constrained efficiency of the market. Yet, in many respects, it was the rest of the paper that was the most important. Any benchmark model is just that—a point of departure. The Arrow-Debreu model is a benchmark model through which we understand the stringent conditions required for markets to be efficient, i.e. we glean insights into market failures. As we noted, the efficiency of the benchmark model depended crucially on assumptions of symmetry and constancy of the elasticity of demand. That’s why some of the extensions and generalizations of our basic model over the last forty years, including the contributions to the 2017 Special Issue of Research in Economics, are so important.

The formalisms, though, sometimes hide the simple intuitions that we tried to provide graphically towards the end of the paper. A profit-maximizing market does not maximize societal well-being; it does not care about consumer surplus, and there is no fixed relationship between profits and consumer surplus. We explored an asymmetric case, where the entry of good with a high own-elasticity of demand (a mass produced good) with low consumer surplus “knocks” out a good with a low own elasticity of demand, with much higher consumer surplus. More generally, and more intuitively, the marginal entrant “steals” customers from other firms. If it has a high elasticity (and a low consumer surplus) and the firms from which it steals have a low elasticity (and a high consumer surplus), then its entry may be profitable yet societal welfare decreased.

There is an important warning here: one has to be very careful about making welfare statements in trade, macroeconomic, and growth models using the Dixit–Stiglitz framework. If one obtains a result that the market is inefficient, the analysis can be useful, in isolating another market failure. If one obtains a result that the market is in some sense constrained Pareto efficient, take it with a grain of salt.

Click here access read full article. Subscription may be required.