Estimating a Model of Strategic Store-Network Choice

Competition among multi-store chains is common in retail industries. This paper proposes a method for estimating a model of strategic store-network choices by two chains. In contrast to previous studies, I allow chains to not only choose which markets to enter but also how many stores to open in each of those markets. I use lattice-theoretical results to deal with the huge number of possible network choices. I show that a chain's net trade-off between costs and benefits from clustering their stores in a market can be either positive or negative while still ensuring the existence of an equilibrium. By doing so, the model provides a way to freely estimate this within-market effect from the data. Incorporating revenue data allows us to interpret parameters in monetary units and to decompose the within-market effect into cost savings from clustering stores (economies of density) and lost revenues from competition with one's own stores (own-chain business-stealing effect). I apply the technique to a new data set from the convenience-store industry in Okinawa, Japan. Parameter estimates confirm that own chain business-stealing is an important consideration for a chain. I then use the estimated structural model to perform two counterfactual analyses. First, I consider a hypothetical merger of two chains and find that the merger would decrease the number of stores and total sales, and raise the acquirer's profits, thereby reallocating surplus from consumers to the acquirer. Second, I examine how eliminating the zoning regulation introduced in Japan in 1968, which has been at the forefront of urban policy debates, affects store-network choices.