اثرات فعالیت های چند ملیتی بر سنجش جهت گیری اصلی

Foreign products reach domestic consumers via cross-border trade and by the production of foreign affiliates of multinational enterprises. The conventional measurement of home bias in consumption of national products relative to foreign products does not recognize the role of multinational activities. In this paper, a gravity-based model that accounts for trade and the operation of foreign affiliates as alternative channels of accessing foreign markets is derived. This gravity-based model is used to theoretically demonstrate that disregarding the activities of multinational enterprises leads to an upward bias in the measurement of home bias. Our empirical application is conducted on a subset of OECD countries in the manufacturing sector. The benchmark results indicate that home bias is overstated by a factor of 1.7 when disregarding the multinational activities. J. Japanese Int. Economies22 (3) (2008) 401–416.

The potential role of foreign direct investment (FDI) as an alternative channel to trade in accessing foreign markets is well discussed in the literature (Markusen, 2002 and Barba-Navaretti and Venables, 2004). In the context of the proximity-concentration trade-off hypothesis (Brainard, 1993 and Brainard, 1997),1 FDI is the result of a strategy undertaken by multinational enterprises (MNEs) to “jump” trade barriers when accessing foreign markets. As such, FDI can be seen as a mechanism that lessens the significance of international borders, and hence reflects one prominent aspect of integration of world markets. One common measurement of the extent of integration of world markets is home bias in consumption of national products relative to foreign products. Home bias occurs as a result of border costs and consumer preferences that are inclined toward the national products. Home bias is commonly assessed by employing a gravity equation that relates trade to the size of the source and destination countries, and trade barriers such as distance separating the trading partners (Wei, 1996). As foreign products reach domestic consumers via trade but also by the production of foreign affiliates of MNEs, multinational activities must be accounted for when measuring home bias. Hitherto, the conventional measurement of home bias has disregarded their role. McCallum (1995) pioneered an approach that applies the gravity equation to measure the magnitude of trade barriers between Canada and the US. McCallum (1995) contrasted Canadian inter-provincial trade with trade between Canadian provinces and US states. The difference between the inter-provincial and international trade flows, after controlling for distance and economic size, was dubbed “border effect”. Following McCallum, Wei (1996) employed the initial concept of border effect to measure “home bias” in the Organization for Economic Cooperation and Development (OECD) countries by contrasting their intranational trade to international trade between them.2 These early papers motivated a large volume of literature measuring border effect/home bias for different economic units and at various levels of industrial classification.3 Theoretical and empirical refinements to the measurement of border effect/home bias have been proposed by a number of authors. Hillberry, 1999 and Hillberry, 2002, Wolf (2000) and Hummels (2001) showed that the decision on the production location, which is determined by international trade, industrial mix and agglomeration, is a constituent factor that influences the measurement of border effect/home bias. Anderson and van Wincoop (2003) highlighted critical shortcomings in the conventional measurement of border effect/home bias by recognizing the theoretical basis of the gravity equation. They showed that the conventional measurement of border effect/home bias suffered from an empirical misspecification that lead to the surprisingly large measure of the Canada–US border effect by McCallum (1995). Some gravity literature brought the activities of MNEs into the analysis of border effect/home bias. Hillberry (1999) regressed the estimated border effect between the US and its trading partners on a proxy capturing the activities of foreign affiliates and found no statistical significance. Evans (2001b) investigated whether the border effect is attributable to location factors such as import barriers and transportation costs or to nationality factors inducing consumption biases toward domestic goods. The location factors were identified by measuring the border effect between imports from the US and sales of US affiliates abroad. The nationality factors were identified by comparing the border effect between imports from the US and sales of domestically owned firms and the border effect between sales of US affiliates abroad and sales of domestically owned firms. Evans (2001b) found that the location factors are more relevant than the nationality factors in explaining the border effect. The objective of this paper is to highlight the distortions accompanying the conventional measurement of home bias that disregards the activities of MNEs. In our analysis, home bias is measured in terms of consumption of domestic products relative to consumption of foreign products (whether imported or produced by foreign affiliates). In order to account for multinational activities, we derive a gravity-based model from a theoretical framework that characterizes trade and sales of foreign affiliates as alternative channels of reaching foreign consumers. This framework implicitly adopts a nationality criterion in defining international transactions. The nationality criterion records international transactions based on the nationality, rather than on the residence, of the economic agents. The merit of the nationality criterion in the measurement of home bias can be illustrated through the following. Consider the Toyota cars that are produced in North America and destined to the North American consumers. When applying the residence criterion, the Toyota cars are treated as domestic North American cars because the production facilities “reside” in North America. When applying the nationality criterion, the Toyota cars are recognized as Japanese cars consumed by the North American consumers. Previous theoretical frameworks used to derive the gravity equation have implicitly adopted a residence criterion when defining international transactions (e.g., Anderson, 1979, Bergstrand, 1985, Bergstrand, 1989, Bergstrand, 1990, Deardorff, 1998, Haveman and Hummels, 2001 and Anderson and van Wincoop, 2003). Empirical studies have generally documented complementarity between trade and FDI when conducted at aggregate levels (Clausing, 2000) and when using firm level data sets (Belderbos and Sleuwaegen, 1998). Meanwhile, substitution between trade and FDI was reported at the product level (Blonigen, 2001) and in the case of firms with marginal intrafirm trade (Head and Ries, 2001). Our theoretical framework that specifies trade and the operation of foreign affiliates as alternative modes of reaching foreign markets, allows not only for substitutability but also for complementarity between them. For example, complementarity may occur when the foreign presence promotes more efficient delivery of exports and induces domestically operating firms to get engaged in exporting to foreign markets. However, complementarity between trade and the operation of foreign affiliates can also be the result of intrafirm trade between the parent firms and the foreign affiliates. This is because an increase in sales of final goods produced by foreign affiliates could induce more imports of intermediate inputs from the parent firms.4 As in Clausing (2000), intrafirm trade must be accounted for to avoid the double counting of intrafirm imports that are also embodied in the sales of foreign affiliates. The remainder of this paper is organized as follows. Section 2 develops a gravity-based model from a theoretical framework that accounts for trade and the operation of foreign affiliates as alternative channels of accessing foreign markets. Section 3 discusses the measurement of home bias and exploits the gravity-based model to theoretically demonstrate the overstatement of home bias when disregarding the activities of MNEs. Section 4 conducts an empirical application for the set of OECD countries reporting the inward activities of foreign affiliates in the manufacturing sector, for the year 1999. This section presents the empirical specification, describes the sources and construction of the data sets, and displays the empirical results. The benchmark results indicate that home bias is overstated by a factor of 1.7 when disregarding the multinational activities. The final section concludes.

Home bias is an assessment of the consumption bias toward national products relative to foreign products. Foreign products are channeled to domestic consumers via cross-border trade, but also via the production of foreign affiliates of multinational enterprises. This fact is disregarded in the conventional measurement of home bias. This paper derives a gravity-based model that accounts for trade and the operation of foreign affiliates of multinational enterprises as alternative channels of accessing foreign markets. This gravity-based model is exploited to theoretically demonstrate the overstatement of home bias when conventionally assessing its magnitude. The benchmark empirical application, conducted on a subset of OECD countries in the manufacturing sector, shows that not recognizing the multinational activities leads to an overstated home bias by a factor of 1.7. It is important to note that more detailed datasets on the activities of foreign affiliates are required for improved estimates of home bias. Due to data limitations, this study did not account for domestic products that are imported from affiliates abroad. In other words, imported national products were treated as foreign products. Consequently, the estimate of home bias is most likely downward biased. One follow-up study to this paper is to evaluate home bias at disaggregated industrial levels and to apply a generalized theoretical framework that describes the decision mechanism on the mode of accessing foreign markets, as in Helpman et al. (2004). The major limitation of empirical applications of gravity-based models that account for multinational activities remains, however, the availability of, and accessibility to data sets covering the transactions of foreign affiliates.