The dynamics of retail oligopoly

This paper examines competition between supermarket chains using a dynamic model of strategic investment. Employing a unique eleven year panel dataset of store level observations that includes every supermarket operating in the United States, we propose and estimate a fully dynamic model of chain level competition. Using a structural model of dynamic oligopoly where firms compete each period in a static stage game, we estimate the dynamic parameters of the model using the methods proposed in Bajari, Benkard, and Levin (2006). The estimation takes place in two stages. In the first stage, the static parameters governing the outcome of product market competition are estimated using a differentiated products discrete choice demand system. We then employ a second, two-step procedure in which policy functions are first estimated from each firm’s observed actions and outcomes are then matched to a (Markov perfect) equilibrium condition using forward simulation. The parameters of the structural model will then be used to evaluate the competitive impact of eliminating Superscenters using the stochastic algorithm developed in Pakes and McGuire (2001).