要旨：Typical analyses of industrial agglomeration start with some aggregate measure of “agglomeration degree” for each industry, and attempt to explain differences in these values across industries by regressing them on sets of industrial attributes. But this aggregation makes it difficult to capture the spatial aspects of individual agglomerations. In the present paper, we develop a more explicit spatial approach to identifying agglomeration determinants by means of a two-stage analysis. First, we detect individual spatial clusters of each industry on a map. We then attempt to explain differences in these cluster patterns between industries by employing an appropriate regression framework. Here, cluster employment sizes are regressed on selected regional attributes for each industry-cluster pair, and significant differences between industries are captured in terms of industry-level interactions with these attributes. This modeling approach is then applied to the three-digit manufacturing industries in Japan.

【大阪大学経済学研究会と共催】