In this paper, we present an analysis of the determinants for overseas location choices of production sites for Japanese agriculture-based food industry multinational firms (MNF) in East Asia for 1985–2009 with a conditional logit model based on new economic geography (NEG). Especially, two periods 1985–1999 (weak agglomeration period) and 2000–2009 (strong agglomeration period) were focused on for the degree of agglomeration of Japanese food MNFs in East Asia, and overseas locations of all-goods and final goods production sites. Results illustrate the following four points. First, for all-goods production sites during the weak agglomeration period (1985–1999), we found that market access (MA) and domestic market access (DMA) in the target country, but not MA in neighboring countries (FMA) are determinant factors. In addition, supplier access (SA) and SA from surrounding countries (FSA), but not the domestic supplier access (DSA) in the target country and the SA from Japan (JFSA), are determinant factors. This is contrary to our expectations. Second, for all-goods production sites during the strong agglomeration period (2000–2009), we found MA and DMA are determinant factors, but FMA is not. DSA and JFSA, but not SA and FSA, are also determinant factors as expected. Third, due to rapid expansion of the domestic market in East Asia during 2000–2009, the estimation results for final goods production sites during the strong agglomeration period (2000–2009) revealed that the final goods production sites are located in the target country because of (1) MA and DMA but not FMA; (2) the DSA of materials and intermediate goods production sites and domestic agricultural output; and (3) inexpensive labor, infrastructure improvements, incentive policies, vertical agglomerations, and freeness of trade. Finally, the above two-period analysis demonstrates that it is necessary to take into account changes in the Japanese and global economic structures when setting the estimation period for Japanese food firms and choosing overseas locations.
Read full abstract