Theoretical studies of new economic geography models in one-dimensional geography such as racetrack and line segment have revealed two regularities for spatial distribution on population. First, population agglomerations are evenly spaced, i.e.,
spatial periods emerge. Second, a monotonic decrease in transportation costs results in a gradual increase of the spatial period which at last leads to a unimodal distribution. In this paper, these theoretical findings are quantitatively assessed by a
spatial spectrum analysis of the Population Census data of Japan. An emergence of spatial periods in the population data at multiple time periods is detected by strong power spectra. By comparing the power spectra at different time periods, it is also shown that the spectrum for the unimodal pattern has monotonically grown over time. It is argued that these properties are consistent with theoretical prediction.
View full abstract