Simulation Analyses on Production Patterns of Multinational Firms

One of the key factors behind the growth of global trade in recent decades is an increase in intermediate input as a result of the development of vertical production networks (Feensta, 1998). Manufacturing goods are no longer produced in a single country. Production processes are subdivided into several stages, in which respective countries specialize in producing parts and components. Many countries are involved in vertical production networks of producing just a single final good for consumers. It is widely recognized that the production networks have formed due to the expansion of multinational enterprises’ (MNEs) activities. Multinational enterprises have been differentiated into two types according to their production structure: horizontal FDI and vertical FDI. However, a new type of FDI which diverges from the vertical one has been proposed in the context of the recent expansion of more complex multinational activities; it is called export-platform FDI. Horizontal FDI maintain affiliates in home and host countries with the headquarters located in the home country, while vertical and export-platform FDI install affiliates in host countries with the headquarters located in the home country. The difference between vertical and export-platform FDI is where their products are sold: vertical FDI seek to sell their products in both the home and host country, while export-platform FDI seek to sell in a third market through the affiliates in the host country (Ekholm et al., 2007). Theoretical research on MNEs has been conducted since the early 1960s (Hymer, 1976), but it developed dramatically from the mid 1980s as a result of the “new” trade theory. There are two important theoretical models of MNEs: one was presented by Helpman (1984) and the other by Markusen (1984). Helpman’s model treats vertical MNEs with monopolistic competition and without trade costs. On the other hand, Markusen’s model treats horizontal MNEs with one factor, assuming firm-level scale economy. Markusen (1997) combines horizontal and vertical motives in a model, so the model allows two types of MNE to exist at the same time. This is called the “knowledge capital” model. Zhang and Markusen (1999) extended the model to consider vertical MNEs that supply intermediate inputs to a final production plant in a host country. While their models were constructed in a two-region framework, Ekholm et al. (2007) extended the model into a three-region framework to include export-platform FDI. Matsuura and Hayakawa (2008) pointed out that recent explorations of FDI theories have shifted from the two-region setting to the three-region setting (for example, Yeaple, 2003 and Grossman et al., 2006). Ekholm et al. (2007) and other models in the three-regional framework assume that skilled-labor-intensive intermediates are produced only at home, and the host country imports intermediate products and assembles final goods, combining intermediates and unskilled labor. However, those models do not adequately explain observed facts where some kinds of intermediate goods are produced in the host country. Our final goal is to extend Ekholm et al. (2007) to treat the procurement of intermediates from the host country in view of the present situation. We start from the simple model in the two-region framework in preparation for further extension. In this paper, we extend Zhang and Markusen (1999) to include horizontal and vertical FDI in the model with traded intermediates. There are no studies which treat vertical and horizontal FDI with traded intermediates at once, although more evolved models which treat vertical, horizontal and export-platform FDI with traded intermediates, such as Ekholm et al. (2007), do exist. This paper serves to bridge the gap between Zhang and Markusen (1999) and Ekholm et al. (2007) in theoretical studies of FDI. we extend the model presented by Zhang and Markusen (1999) to include horizontal and vertical FDI in a model with traded intermediates, using numerical general equilibrium analysis. The simulation results revealed that horizontal MNEs are more likely to exist when countries are similar in size and in relative factor endowments. Vertical MNEs are more likely to exist when countries differ in relative factor endowments, and trade costs are positive. Based on the results of the simulation, lower trade costs of final goods and differences in factor intensity are the condition for attracting vertical MNEs.