Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.4, No.1, December 2008
In recent economic geography, it is emphasized that the effect of cost decreasing in transportation on agglomeration is nonlinear. It is said that the influence of traffic infrastructure investment and the change in transportation cost on urban agglomeration does not appear until the cost is below a certain amount, and that once agglomeration arises that effect would be kept with higher probability. In theoretical models such as Krugman (1991) and Fujita, Krugman and Venables (1999), multiple equilibria and path dependence are emphasized, as well as non linearity. Those models are intuitive, but it is hard to have a statistical analysis because of the non linearity. About the macroeconomic effect of social overhead capital investment, starting from the analysis by Aschauer (1985, 1989), a lot of empirical research has been done on the productivity effect of social capital. For example, we have Asako et al. (1994), Mitsui and Ohta (1995). Moreover, Roback (1982) uses the Hedonic approach to find the effect of amenity-based social overhead capital (related to waste disposal plants, or sewage facilities), followed by Mitsui and Hayashi (2001) for a Japanese case. In these Japanese studies, they are only concerned about the topic about inefficiency of the social overhead capital distribution but not about theoretical progress in urban economics. If Krugmans model is true, however, there is a possibility that rural traffic infrastructure investment for the purpose of redistribution will experience both a decline in rural areas and agglomeration into urban areas. In the following, we will examine general theory about how we should observe the effect of traffic network provision in section II. We will estimate a market potential function and an index with which the geographical concentration degree is measured, and see how the agglomeration degree has changed historically. In section II we will conduct analysis through using prefecture data and municipal data, particularly in the Kyushu district 2 .