To Price Discriminate or Not: Product Choice and the Selection Bias Problem

In this paper, we investigate a gasoline station's incentive to price-discriminate by selling full-service gasoline as well as self-service gasoline. Unlike previous research, we explicitly model a firm's incentive to price discriminate by choosing to be either single-product or multi-product as a function of market and station characteristics. This allows us to make two contributions to research in the area: First, we highlight the importance of accounting for self-selectivity considerations that can arise in an empirical analysis of price discrimination that is based on market data. Second, we are able to show how the product and pricing choices of firms depend upon the market characteristics.Using cross-sectional survey data on prices, station and market characteristics for 198 gasoline stations in the Greater Saint Louis area, we estimate a switching regression model of station decisions. Specifically, we employ a binary probit framework that models a station's decision to price-discriminate through the choice of the station-type as a function of market and station characteristics. We then estimate conditional linear regressions with self-selectivity corrections for the station's choice of prices. We show that incorrect inferences about the incentive to price discriminate and about the differences in the prices charged between single-product and multi-product stations would result if the endogeneity in the choice of the station-type were ignored in the estimation. The empirical analysis shows that a larger income spread in the market implies a greater likelihood of the gasoline station being multi-product. In addition, we have support for the various within firm and across firm price differentials as predicted by the theory of price discrimination.