This paper investigates the changing geographical pattern of manufacturing industries in Japan in the 1990s and explores factors of their geographical concentration. We start with an estimation of the geographical concentration of manufacturing industries using the coefficient of localization based on manufacturing employment and establishment data at the prefecture level. We then conduct a regression analysis to test some hypotheses of geographical concentration of manufacturing industries, which were derived from new theories of trade and economic geography that have been advanced by Fujita, Krugman, and Venables (1999). In the regression analysis, we consider the following three factors of geographical concentration: scale economies, transportation costs, and inter-industry linkages. We follow basically the approach used by Amiti (1998, 1999), which investigated the effects of scale economies and inter-industry linkages on the geographical concentration of manufacturing industries for EU countries using manufacturing employment and output data from EUROSTAT and UNIDO. As opposed to Amiti, we also consider transportation costs as a possible factor of geographical concentration. Furthermore, our analysis is based on regional data rather than country data. As a measure of plant-level scale economies, we use the ratio of total employment to the total number of establishments in each industry, while as a measure of the intensity of transportation costs, we use the ratio of intermediate transportation inputs to total inputs. To measure plant-level scale economies, we employ manufacturing data from the Statistics of Industry. On the other hand, to measure the intensity of transportation costs and inter-industry linkages, we use the national input-output tables. We expect that scale economies and inter-industry linkages have positive effects, while transportation costs have a negative effect, on the geographical concentration of manufacturing industries.
展开▼
