Spatial competition based on customers' choice histories: A study of trade flows

In a world of differentiated firms, customers with the same observable characteristics reveal different choice behaviors. This research contributes to the modeling of competition in a market characterized by spatially differentiated firms and significant consumers' taste heterogeneity. With recognition of the fact that the degree of competition depends on the extent to which a representative customer switches among firms, this research proposes an emprical framework for quantifying competitive interactions. The framework is built on the Latent Class Logit and customers' choice histories that allow estimation of heterogenous preferences. Furthermore, relying on unique data of geographical distribution of demand, this framework particularly emphasizes the spatial dimension of competition and provides a disaggregated measure of regional contestability, which is often ignored in the previous studies.;The applicability of the proposed framework is tested on shipper-level data from a business trade dataset to model choices for seaports, as critical nodes in logistics networks that support global trade. We estimate how distance, crossing a national border and port characteristics influence port choice across shippers of different size. The results suggest significant heterogeneity in preferences that gives rise to the intensity of competition. Segments of customers that are willing to cross borders and travel broad geographical distances toward efficient ports generate international competitive forces and encourage ports to improve efficiency, while customers that avoid borders and put large negative value on distance support the local monopoly of the closest ports with no incentive for efficiency improvement.