A Spatial Network Approach to Urban Configurations

This paper proposes a new approach to city formation by modifying conventional network games to suit particularly the study of spatial configurations and urban structures. While we postulate uncompensated knowledge spillovers as the main driving force of population agglomeration, we depart from the conventional wisdom by modeling explicitly knowledge transmission, aggregation and spillovers via links formed in our spatial networks. While it is beneficial to be connected to take advantage of knowledge spillovers from other locations, maintaining a link is costly. In contrast with previous studies, we allow the population in each location to be endogenously determined in equilibrium based on the locational choice of freely mobile players. Moreover, players at a location are also allowed to decide collectively to take the locality role as a "core" that serves as a knowledge aggregation and transmission node for other connected "peripheral" nodes. We establish conditions for the mostly commonly assumed monocentric configuration to arise as the unique and efficient equilibrium outcome. We also illustrate under what circumstances a spatial equilibrium may feature multicentric, urban-rural, or multiple (disconnected) urban areas and under what conditions each configuration is socially efficient. We further perform comparative statics with respect to changes in knowledge spillover, link maintenance, urban land rent, rent gradient, and urban unemployment parameters