ABSTRACT
I first conducted two simulations focusing on how firms would choose locations to form agglomerations as a way of observing spatial evolution both across and within regions. I concentrated on how simple behavioral models of the firms interacting with each other give rise to complex, but orderly, spatial patterns. Based on the principle of increasing returns to choosing among competing locations, agglomeration of the firms is explained as a function of the initial payoffs of the regions as well as the opportunity and distance costs of moving. I further demonstrated the invariability and universality of the power law distribution through empirical studies in China. By comparing these with our computer simulations, I found that the power law distribution was a robust rule within and across cities and was independent of increasing returns and landscape variables. I also conducted extensive analyses in search of an allometric scaling relationship between various socioeconomic variables and urban populations in Taiwan. Together with the foregoing results, they can be summarized as follows: First, allometric scaling laws of cities exist for most socioeconomic variables. Second, we still do not know how the allometric relationships emerge for urban activities and socioeconomic quantities with respect to urban populations and how these cities behave along the scaling line. Based on the evidence introduced so far, I would argue that cities are complex spatial systems where actors interact with each other spatially, and spontaneous order thus emerges regardless of the patterns of such interaction.
