Models of monopolistic competition are useful in many areas of economics when there is a need for a description of industry equilibrium in the presence of product differentiation and increasing returns to scale. There are two main models of monopolistic competition: the symmetric model based on Chamberlin's analysis of monopolistic competition, and recently formalised by Spence-Dixit-Stiglitz, and the spatial (circle) model of Salop. The conceptual foundations of these models are very different, and both impose specific restrictions. For example, the Spence-Dixit-Stiglitz (Chamberlain-type) model starts with aggregate demand without explaining how this follows from aggregation over individuals, and it imposes the property that demands are completely symmetric in the sense that a price cut by one firm draws customers from all other firms equally, a property we term fully global competition. On the other hand, Salop starts with a description of individual behaviour and then aggregates over consumers. In this framework, each firm competes directly with only its two neighbours: competition is purely local. More realistically, demand systems should lie somewhere between the extremes of fully global and purely local competition.

In Discussion Paper No. 1328, Simon Anderson and André de Palma introduce a framework that has known models of oligopolistic competition with differentiated products as limit cases. This integrative approach incorporates both localized and global competition, as well as price-sensitive individual demands. It is used to consider the qualitative impact of major changes over the last two centuries: reduction in transport costs, increased taste for variety, population growth, and use of technologies with greater returns to scale. The paper works out the properties of an extended Chamberlain model that should be useful both in industrial organisation and urban economics.