Spatial Competition in Retail Markets for Gasoline

This dissertation is an empirical study of the entry decisions and location choices of gasoline stations along interstate highways, as well as their pricing behavior. First, I test for incentives for spatial differentiation. The variation in the location of interstate exits allows me to test how distances between exits can explain entry decisions at an exit. I find that when an interstate exit becomes more geographically isolated, the probability of entry and the number of stations at an exit increases. In addition, I find that when entering a market, stations choose to locate farther from rivals in markets with lower concentration, which supports the first finding that there are incentives for spatial differentiation.;Then, I present descriptive regressions of how spatial differentiation affects gasoline prices and price dispersion along interstate highways. The results demonstrate that price decreases as the number of rival gasoline stations increases. The effect is stronger for closer by stations but becomes insignificant for farther away stations. However, the competitive effects for diesel prices are all insignificant. Since the consumers of diesel are mostly trucks, and since truck drivers are more informed than long-distance travelers about fuel prices, this result suggests that the relation between geographic locations and prices is different when consumer's information about prices is different. In addition, I show that there is a higher market average price and more price dispersion in markets where gasoline retailers are more dispersed. The former suggests that differentiation is a mechanism for retailers to gain market power, and the latter is consistent with the theory that more differentiated firms can better price discriminate across locations.;Finally, I develop a model of demand and supply of gasoline, in which the utility maximization problem of heterogeneous consumers is integrated into the firm's optimization problem. I allow a number of consumers who have no information about prices but do observe stations' locations. I explain how to estimate the model's structural parameters and run counterfactual experiments that will allow me to test how consumers' information affects prices.