Embedding Consumer Taste for Location Convenience into a Structural Model of Equilibrium

We apply Berry, Levinsohn and Pakes (1995) to jointly estimate demand and cost parameters for all 178 brands (products) in Retail Carbonated Soft Drinks over 28 bi-monthly periods. Our innovation is to randomize over store coverage (weighted by store size) to create a distribution of consumers’ taste for location convenience to each brand. Such distributions and their interaction with price enter utility and store coverage enters cost. Monte Carlo simulation techniques indicate that allowing for product differentiation to come heterogeneity in a product specific consumer attribute, in addition to that coming from product characteristics, improves greatly the predictive power of the structural model.