Price competition, mergers and structural estimation in oligopoly

• Main Author: Salvo, Alberto
• Published: London School of Economics and Political Science (University of London) 2005
• Subjects: 338.82
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.417448

This thesis examines the exercise of market power by oligopolistic firms. The first part deals with a phenomenon that has important implications for market power: horizontal mergers. I seek to uncover why the pattern of equilibria in sequential merger games of a certain type is similar across a fairly wide class of models studied in the literature. By developing general conditions characterising each element of the set of possible equilibria, I show that the solution to models that satisfy a certain sufficient condition will be restricted to the same subset of equilibria. This result is of empirical relevance in that the pattern of equilibria obtained for this class of models is associated with mergers happening, not in isolation, but rather in bunches. I extend the results to the analysis of cross-border mergers, studying two standard models that satisfy my sufficient condition: Sutton's (1991) vertically-differentiated oligopoly and Perry and Porter's (1985) fixed-supply-of-capital model. The second part is concerned with the structural inference of market power, a central theme in empirical Industrial Organisation. I demonstrate that when an industry faces potential entry and this threat of entry constrains pre-entry prices, cost and conduct cannot be identified from the comparative statics of equilibrium. In such a setting, the identifying assumption behind the well-established technique of relying on exogenous demand perturbations to distinguish empirically between alternative hypotheses of conduct is shown to fail. The Brazilian cement industry, where the threat of imports restrains market outcomes, provides an empirical illustration. In particular, price-cost margins estimated using this established technique are biased heavily downwards, underestimating the degree of market power. I propose a test of conduct, adapted to this constrained setting, which suggests that outcomes in the industry are collusive and characterised by market division. Robustness of this result is verified along several dimensions: by considering simple dynamic multimarket games which in equilibrium give rise to market division; by reviewing the spatial competition literature; and by resorting to a gravity model to statistically analyse shipments.