An Industrial Agglomeration Approach to Central Place and City Size Regularities

An empirical regularity designated as the Number-Average Size (NAS) Rule was first identified for the case of Japan by Mori, Nishikimi and Smith [13], and has since been extended to the US by Hsu [6]. This rule asserts a negative log-linear relation between the number and average population size of cities where a given industry is present, i.e., of industry-choice cities. Hence one of its key features is to focus on the presence or absence of industries in each city, rather than the percentage distribution of industries across cities. But despite the strong empirical regularity of this rule, there still remains the statistical question of whether such location patterns could simply have occurred by chance. In this paper an alternative approach to industry-choice cities is proposed. This approach utilizes the statistical procedure developed in Mori and Smith [15] to identify spatially explicit patterns of agglomeration for each industry. In this context, the desired industry-choice cities are taken to be those (economic) cities that constitute at least part of a significant spatial agglomeration for the industry. These cluster-based choice cities are then used to reformulate both the NAS Rule and the closely related Hierarchy Principle of Christaller [2]. The key empirical result of the paper is to show that the NAS Rule not only continues to hold under this new definition, but in some respects is even stronger. The Hierarchy Principle is also shown to hold under this new definition. Finally, the present notion of cluster-based choice cities is also used to develop tests of both the locational diversity of industries and the industrial diversity of cities in Japan.