Estimating the Structure of Sunk Costs in the US Airline Industry

This paper studies economies of scale and sunk entry costs in the US airline industry. Our main interest is in estimating how costs of entry and operation in a city-pair market depend on a company's own network and on the number of competitors in that market (i.e., endogenous sunk costs). We use a panel dataset from the Airline Origin and Destination Survey with information on quantities, prices, and entry and exit decisions for every airline company, over more than two thousand city-pairs markets and several years. We propose a dynamic model of oligopoly competition where an airline's strategic variables are prices and the network of city-pairs where it operates. We first estimate variable profits using information on quantities and prices and the recent developed methods in the literature on estimation of differentiated demand systems. The structure of fixed operating costs and sunk entry costs is estimated using a dynamic game of entry and exit in local markets. Our econometric approach is based on the method proposed by Aguirregabiria and Mira (2005) for the estimation of dynamic games. Reduced form estimates for market entry and exit decisions show very large unobserved heterogeneity in market profitability even after controlling for variables profits. We exploit the panel structure of the data to apply a fixed-effect approach to control for this heterogeneity.