45th Congress of the European Regional Science Association: "Land Use and Water Management in a Sustainable Network Society", 23-27 August 2005, Amsterdam, The Netherlands
Agglomeration economies are usually divided into two categories: urbanization economies and localization economies. In 80fs a number of attempts have been devoted to estimate urbanization economies and/or localization economies. After the work by Glaeser et al. in 1992, however, historical effects on agglomeration called dynamic externalities in agglomeration are tried to estimate extensively. These externalities are named as MAR in a dynamic sense, and traditional agglomeration economies are evaluated in static sense. Besides urbanization and localization, more traditional sources of industrial concentration are found in industrial linkages, such as customer and supplier linkages or backward and forward linkages. These linkage effects come from the concentration of different kinds of industries while localization economies mean the benefit from the concentration of firms within the same industry. Also, linkage effects are often referred as pecuniary externalities. This paper tries to construct an estimable model of linkage effects among industries as well as agglomeration economies, and to estimate these effects separately within a framework of the Translog production function. In this model intermediate inputs play an important role as linkage effects. The empirical analysis is based on two-digit data for manufacturing industries in Japanese cities. Estimated results vary significantly among the two-digit industries. Furthermore, in order to capture dynamic effect in changes of agglomeration, a time variant production function model which is consistent to static production function model is constructed and estimated. From the time series evidence we find agglomeration economies are decreasing while linkage effects are still important.