Measuring Substitution Patterns and Firm Conduct in Differentiated Product Industries

National Bureau Of Economic Research Inc, Cambridge MA

Investigators

Abstract

In this project, the PIs propose a new set of tools useful to detect and measure the importance of market power in industries. This question is central to many economic problems. It is particularly relevant for a vast array of antitrust questions ranging from the approval of mergers to the detection of price fixing cases, as well as to help guide regulatory institutions in markets such as financial services, energy, and telecommunication. In most of these cases, measuring market power hinges on two key elements: (i) identification of the degree of substitution between products, and (ii) quantification of its impact on market prices. The key contribution of the proposed work to this literature is that it will provide a new set of empirical techniques to both measure degree of substitutability between products, and help better identify firms' pricing conduct in differentiated product industries. On the first point, the PIs build on an important literature in the field of Industrial Organization, devoted to the identification of substitution patterns in markets for differentiated products. The main contribution of this work is to show the existence of a source of variation in the data that drives the identification of the degree of substitutability between products in standard demand models. They exploit this variation to construct parametric and non-parametric estimators of models commonly used in the literature (e.g. Berry et al. (1995), Nevo (2001)). This addresses important reserves raised by practitioners regarding the robustness of the predictions offered by existing methods (e.g. Angrist and Pischke (2010)). On the second point, the PIs propose a novel approach to empirically estimate the supply relationship between prices and demand factors, without imposing strong and often untested assumptions. Early efforts to identify firms' conduct have relied on contrasting observed measures of pass-through of demand or cost shocks on prices and quantities, with fairly general theoretical predictions from models of competition or collusion (Bresnahan, 1989). This empirical strategy can bound the degree of market power in homogenous good industries, but cannot easily be extended to settings in which sellers supply differentiated goods and services. In contrast, researchers studying the extent of market power in differentiated product industries typically rely on more restrictive models of competition, which often eliminates entirely the possibility of testing for alternative conduct assumptions; such as perfect competition or collusion. To get around this problem the PIs generalize Bresnahan's (1987) strategy for testing for the presence of coordinated pricing to an environment with potentially rich product differentiation, allowing for a larger class of possible supply relations to determine firm conduct. This allows in turn to directly examine how firms exploit market power when setting prices, rather than assuming it in the form of stylized game theoretic models.