What Determines the Profitability of Foreign Direct Investment?

A Subsidiary-Level Analysis of Japanese Multinationals

 

Mariko Sakakibara

Anderson Graduate School of Management

University of California, Los Angeles

 

and

 

Hideki Yamawaki

 

Drucker and Ito Graduate School of Management

Claremont Graduate University

 

 

October 2005

 

 

 

 

 

 

This article is a product of the joint research project with the Ministry of International Trade and Industry Research Institute, Japan.  We would like to thank Yuji Hosoya, Hiroya Tanikawa, Yukio Uchida, Akira Goda and Susumu Yonekawa for administrative support and data compiling.  Tatsuo Ushijima and Heather Berry provided valuable research assistance.  Financial support from the Center for International Business Education and Research at the UCLA Anderson Graduate School of Management and the Academic Senate of the University of California is gratefully acknowledged.  We are also grateful to Henrik Braconier, Marvin Lieberman and seminar participants at the 74th Annual International Conference of the Western Economic Association for helpful comments.  A summary of this paper appears on the Best Paper Proceedings, Academy of Management Annual Meeting 2000, Toronto, Canada.   

Corresponding author:  Mariko Sakakibara, Anderson Graduate School of Management, University of California, Los Angeles.  110 Westwood Plaza, Los Angeles, CA, 90095-1481, USA.  Tel: (310) 825-7831; fax: (310) 206-3337; e-mail: mariko.sakakibara@anderson.ucla.edu.  Hideki Yamawaki, Drucker and Ito Graduate School of Management, Claremont Graduate University.  1021 North Dartmouth Avenue, Calremont, CA 91711, USA.  Tel: (909) 607-8494; fax: (909) 621-8543; e-mail: hideki.yamawaki@cgu.edu.


 

What Determines the Profitability of Foreign Direct Investment?

A Subsidiary-Level Analysis of Japanese Multinationals

 

ABSTRACT

This article identifies key factors that determine the profitability of Japanese firms abroad by using panel-data regression models on new, large-scale, subsidiary-level data over the 1990-1996 period.  The results show that the determinants of subsidiary profits differ across host regions, suggesting that the economic and institutional factors specific to host regions influence significantly the profit performances of overseas subsidiaries.  While the size effect on the subsidiary profitability is present in all the regions, other effects, such as experience, local supplier networks, local sales and macroeconomic conditions affect the performance of subsidiaries in a different manner by region.

 

 

Running headline: What Determines the Profitability of Foreign Subsidiaries?

Key words: Foreign direct investment, profitability, multinational enterprises, overseas subsidiaries 

 


INTRODUCTION

While the traditional literature that examines the performance of the multinational enterprise (MNE) has focused on resources and capabilities possessed by the parent firm, the more recent literature emphasizes the importance of resources and capabilities developed at subsidiaries (e.g. Birkinshaw and Hood, 1998; and Birkinshaw, Hood, and Jonsson, 1998).  When a MNE is composed of a number of subsidiaries and operates under a network of subsidiaries with different characteristics (Ghoshal and Bartlett, 1990; and Nohira and Ghoshal, 1994), individual subsidiaries are likely to contribute to the MNE’s overall performance through the development of new sets of capabilities.  The process through which a subsidiary develops new capabilities and accumulates resources is likely to be facilitated by the establishment of R&D units in the subsidiary, which is, in turn, determined by the subsidiary’s characteristics such as size and organizational experience (e.g. Gerybadze and Reger, 1999; Kuemmerle, 1999; Frost, Birkinshaw, and Ensign, 2002; and Belderbos, 2003).

Parallel to this line of literature, an economic model developed by Markusen and Venables (1996) and Markusen (2002) shows formally that the locations of the MNE’s R&D and production activities are determined by the confluence of factors required to perform these activities and location-specific factors.  What their model suggests is that the MNE shifts its R&D and/or production units from the home country to host country, and the pattern depends on such economic factors as country size, trade and investment costs, scale economies, and factor content of specific activities, among others. 

These two strands of literature clearly indicate that the MNE’s performance is influenced by its subsidiaries’ capabilities and their geographic locations.  If individual subsidiaries contribute to the performance of the enterprise as a whole, an important question that follows is what determines the performance of individual subsidiaries.  While we believe this question is relevant empirically, thus far no study has addressed this issue using systematic data.  The existing literature on subsidiary capability implies that subsidiary characteristics determine its capability and thus its market performance.  The main purpose of this paper is then to address this question and identify the determinants of subsidiary performance in terms of profitability.  

While the issue of subsidiary performance is important in general, this has become quite relevant and important for Japanese firms and their foreign rivals as many Japanese firms currently face challenges which force them to reevaluate their foreign direct investment strategies and restructure their global operations.  This contrasts starkly with the situation during the latter half of the 1980s in which Japanese firms aggressively expanded their presence in North America, Europe and Asia. 

Our study is the first attempt in the literature to examine the profitability of Japanese firms abroad during the period directly following the era of rapid growth.  This article tests several hypotheses to identify key factors that determine the profitability of Japanese firms abroad by using panel-data regression models on a new data sample covering 19,307 subsidiaries of 3,474 Japanese multinational enterprises (MNEs) over the 1990–1996 period.

While determinants of industry and firm profitability have been sought and examined extensively in the fields of industrial organization and business strategy, the profits reported by MNEs are seldom examined empirically in the field of economic analysis of multinational enterprise (e.g., Lecraw, 1983).  The paucity of statistical research on the profitability of MNEs is largely due to the daunting task of obtaining profit data defined at the level of foreign subsidiary.  This article overcomes the difficulty and goes beyond the existing study in several aspects.  First, this article presents the first evidence of the profits reported by Japanese firms at the level of subsidiary distributed across a large cross-section of industries.  The performance of the overseas subsidiary is an interesting research topic from a managerial perspective because it provides measurement for manager evaluation.  Second, the data covers all of the host countries around the world, including North America, Europe and Asia where Japanese firms established subsidiaries.  Previous evidence on the determinants of profitability of MNEs was confined to a limited number of host countries.[1]  Third, the availability of longitudinal data at the level of subsidiary allows us to construct a panel for the period of 1990–1996.

The results of this analysis show that the determinants of subsidiary profits differ across host regions, suggesting that the economic and institutional factors specific to host regions influence significantly the profit performances of overseas subsidiaries.  Some effects are common across markets, while other effects, such as experience, local supplier networks, local sales and macroeconomic conditions affect the performance of subsidiaries in a different manner by region.

The remainder of this article is organized as follows: The second section explains the data and presents some descriptive statistics on the profits at the subsidiary level.  The third section surveys the existing literature and develops the hypotheses.  Empirical tests are provided in the fourth section in the form of regression analysis of subsidiary profits.  The final section summarizes the main findings and concludes the article.

 

DATA AND DESCRIPTIVE STATISTICS

Data

The data used for this analysis is from the Survey of the Overseas Activities of Japanese firms (the Survey), an annual survey conducted by the Ministry of International Trade and Industry, Japan (MITI).  The Survey covers 3000+ parent firms and their overseas subsidiaries, whose response rate is approximately 80% every year.  Companies are asked to report the data as of March 31 of each year.  We are able to obtain proprietary subsidiary-level data, matched with their parent firm data.  From the Survey we can identify the country, the industry, the type of activities, and the basic financial data of each subsidiary for each year.  In addition to the basic annual survey, a detailed survey is conducted once in every three years, including more detailed financial data, the breakdown of intra-trade statistics and qualitative aspects of the activities of foreign subsidiaries.[2]  Due to the condition imposed on the use of the Survey, the identity of these firms is not disclosed, prohibiting us from matching these data with firm-specific data from other sources.  To our knowledge, however, this is one of the most detailed data of the foreign subsidiaries to date.

            A panel data is constructed from the 1990–1996 Survey results.  The panel is not balanced due to the missing reporting.  The panel covers 3,474 distinct parent firms and 19,307 distinct overseas subsidiaries, and the total number of observations is 61,705.  The following statistical analysis of the profitability is based on 46,953 observations with which we can obtain the data of the return on sales.  The reporting of the local procurement and local sales is poorer, yielding 21,720 and 32,245 observations, respectively.  These observations include subsidiaries in all the countries.  In our analysis, we focus on the ones located in the United States, the European Union (EU), East Asia (which includes China, Taiwan, Hong Kong and South Korea) and ASEAN due to the importance of these regions for the Japanese FDI and the ease of the interpretation of the results. 

 

Patterns of Subsidiary Profitability

The most striking pattern that emerges from our data of subsidiary profits is that the profits reported for subsidiaries located in Asia are, on average, higher than the profits of subsidiaries in the United States and the EU.  Figure 1 shows the behavior of the profits (gross profits / sales) for subsidiaries located in the United States, the EU, ASEAN and East Asia in all industries between the 1990–1996 period.  The profits of the subsidiaries in ASEAN and East Asia move between 3.4% and 1.4%, while those in the United States and the EU vary between 1.0% and -0.6%.  The profits reported for U.S. subsidiaries are the lowest among the four regions in the figure during the 1990–1992 period.

For Japan’s four major manufacturing industries, in terms of the extent of FDI, chemicals, general machinery, electrical equipment, and transportation equipment, Figures 2–5 show that the profits reported for subsidiaries in ASEAN and East Asia are consistently positive and likely to be larger than those for the United States and the EU, underlying the general pattern observed in Figure 1.  This pattern is confirmed again for the profits for wholesale and retail trade, and restaurants (Figure 6).

Looking at the profits reported for the U.S., European and Asian subsidiaries of Japanese firms, the behavior of profits over time differs across industries.  For example, the U.S. subsidiaries in chemicals, electrical equipment and transportation equipment tend to report negative profit rates for 1990–1992 and positive rates for 1993–1996, while those in general machinery report negative profits only in 1994.  The pattern of inter-industry difference in the behavior of profits is not necessarily the same between different host regions.  Rather, the behaviors of their profits tend to differ distinctively.  Figures 2–5 indicate a pattern that the profits for EU subsidiaries are higher than those of U.S. subsidiaries in 1990 and 1991 but start declining after 1992.  Indeed, the subsidiary profits in the EU are the lowest among the four regions during the 1993–1996 period for chemicals, electrical equipment, and transportation equipment and during the 1992–1995 for general machinery.  This pattern, however, is not observed for wholesale and retail trade, where the U.S. subsidiaries tend to earn low profits.

Previous empirical studies that examine the time-series patterns of industry and firm profits (e.g., Domowitz et al., 1986, for U.S. industry profit-margins; and Machin and Van Reenen, 1993, for U.K. firm profit-margins) have found that industry and firm profits tend to respond to aggregate demand fluctuations and thus aggregate cyclical fluctuations.  Using panel data, the U.S. and U.K. studies found that industry profit-margins and firm profit-margins move pro-cyclically with aggregate demand fluctuations.

While it is not unequivocally ascertained here whether subsidiary profits move cyclically with macro-economic conditions of host countries, we conjecture that our profit data are also influenced to some extent by aggregate demand fluctuations.  For example, the U.S. economy declined in 1990 and 1991, reducing annual real GDP growth rates from 3.9% in 1988 to 1.2% in 1990 and further to -0.7% in 1991.  The economy recovered in the subsequent period, increasing GDP growth to 2.6% in 1992 and 3.1% in 1994.  On the contrary, the European economy was still expanding relatively strong in 1990 at a rate of 3.7%.[3]  Her economy, however, contracted sharply in 1993.  Such difference in the pattern of aggregate demand fluctuations between the United States and Europe may provide a partial explanation to the difference in the behavior of profits for subsidiaries located in these regions as observed above.[4]

In sum, the preliminary analysis thus far suggests that our subsidiary profit data vary across industries and over time.  In addition to these two sources of variance, it varies importantly across host countries and regions.  In the empirical analysis of this article, we will specify estimation equations that determine profits by taking these different sources of variance into consideration.  One of the hypotheses we are interested in testing is the hypothesis that the determinants of subsidiary profits are different among host regions.  We expect that they are different as implied by the descriptive analysis.

---------------------------------------------

Figures 1 through 6 about here

---------------------------------------------

 

 

Subsidiary’s Local Activities

One of the main features of MNEs’ activities is their extensive use of intra-firm trade.  The previous empirical work on the extent to which MNEs trade goods between parents and their affiliates has found that MNEs engage extensively in off-shore procurement (Jarret, 1979), intra-firm trade of components and semi-finished products (Casson and associates, 1986), and intra-firm exports of finished goods to affiliates (Andersson and Fredriksson, 2000).  A number of statistical studies that examine the determinants of intra-firm trade found, in general, that the extent of intra-firm trade is positively related to R&D intensity used as a proxy for the importance of specialized assets involved in the transaction of goods in question (e.g., Lall, 1978; Helleiner and Lavergne, 1979; Benvignati, 1990; Andersson and Fredriksson, 2000).

Generally speaking, the decision to engage in intra-firm trade for a subsidiary should affect its profits through its impact on operating costs.  A subsidiary may engage in intra-firm trade because it enables the subsidiary to operate under a circumstance in which it can choose an optimal combination of local and import procurement.  While the importance of local procurement as a determinant of MNEs’ performance has been implied in the previous literature of Japanese MNEs (e.g., Hackett and Srinivassan, 1998; Belderbos et al., 1998), no study has explicitly examined their effects on the profitability.

The database used in this study allows us to test whether the extent of local procurement affect the profitability of subsidiaries.  Our data sample shows clearly that the local procurement ratio varies not only across industries and host regions but also changes over time.  For example, in transportation equipment whose major industries are automobiles, the local procurement ratio in the United States increased over time from 53% in 1990, 58% in 1992, and 70% in 1996.  This trend is presumably reflecting the response of the U.S. subsidiaries of Japanese auto makers to the provision of the North American Free Trade Agreement (NAFTA), which increased North American content requirement from 50% to 62 % over an eight-year period.

The local procurement ratio varies widely across industries as well.  In 1996, the local procurement ratio was, on average, quite high and exceeded 90% in raw material intensive industries, such as foods, lumber and wood products, and pulp and paper, while it was much lower in industries such as electrical machinery and precision instruments.  Not surprisingly, the local procurement ratio in our data is strongly negatively correlated with the ratio of subsidiary’s imports from Japan to total procurement.

As the subsidiary’s propensity to procure locally is a potentially important determinant of profits, its propensity to sell locally is also expected to influence profits through local demand conditions.  Therefore, we will introduce a variable that measures the fraction of total subsidiary sales generated from local sales: the local sales ratio.  This variable again varies across industries.  The manufacturing industries with the lowest local sales ratios in 1996 are petroleum products, lumber and wood products, pulp and paper, and publishing and printing, while those with high local sales ratios are transportation equipment, iron and steel, wooden furniture, metal products, and tires and rubber products.  Along with the observation on the local procurement ratios, this provides evidence that is consistent with the pattern that Japanese MNEs are more likely to export raw material intensive products from host countries.

 

EMPIRICAL MODEL AND HYPOTHESES

Specification and Hypotheses

A great deal of previous empirical research on the determinants of firm profitability has been based on two explanations.  The first explanation is based on the observation that a firm earns rents from two sources: superior efficiency and market power.  Since superior efficiency and market power accrue to a firm from its possession of proprietary assets, firm rents derive from such assets.  To the extent that the firm size and market share reflect a firm’s possession of such rent-yielding assets, large firm size and market share are likely to predict higher firm profits.[5]  In the literature which focuses on the resources firms possess, the attention has been paid on the explanation of how differences in efficiency are sustained in the face of competition.[6]  The second explanation is based on the observation that market structure shapes the behavior of firms within the industry, which in turn determines their profits.  Structural characteristics such as concentration, barriers to entry and product differentiation are considered to be important in determining firm profits.[7] 

Previous work on the performance of MNEs has used these approaches of industrial organization and evaluated the issues of market power and efficiency using variables such as market share and concentration (Lecraw, 1983).  While the presence of both firm-specific effects and industry-specific effects in the determinants of MNEs’ profits has been documented in the previous empirical work, country-specific effects which are more inherent to the operation of MNEs need to be elaborated further in empirical analysis. 

Our profit equation incorporates these factors and is given as:

Pi t = a + gj + lk + ft + Xi t ˘b1 + Ym t ˘b2 + ui t                                                                                   (1)

where Pi t is the measure of profitability defined as the ratio of gross profits to sales for subsidiary i at time t, a is an intercept, gj is an industry-specific fixed effect for industry j in which subsidiary i operates, lk is a country-specific effect for host country k in which subsidiary i operates, f t is a time-specific effect, Xi t denotes a set of subsidiary-specific variables, Ym t denotes a set parent-specific variables, and ui t is a random error term.

The effects common to each subsidiary in industry j and host country k are controlled for by gj and lk ,respectively.  Industry effects are modeled in a general way by incorporating a set of industry dummies in estimation.  The time-specific effect, ft, is included to capture macro-economic effects common to each subsidiary.  Both host country and time-specific effects are modeled by including GDP-related measures.

The subsidiary-specific variables, Xi t , are included to test several hypotheses on the determinants of subsidiary profits of Japanese MNEs.  One of the key features in our empirical analysis lies in that we incorporate several new variables which measure the extent and scope of MNEs’ local activities.  This research design is based on the premise that MNEs respond to location forces specific to the host country and adapt their local operations accordingly.  Location forces such as factor price differences, skill formation, local content rules, tax policy, and rules restricting the repatriation of profits all affect the behavior of MNEs and the configuration of their business activities.  In addition, we consider parent-specific variables, Ym t, which are common to firm m’s subsidiaries.  We consider the following subsidiary-specific and parent-specific variables and corresponding hypotheses. 

Scale.  As discussed above, the scale of a subsidiary is a proxy for its market power and efficiency, and is likely to reflect its possession of advantageous resources.  The subsidiary with a large-scale operation can also benefit from the economies of scale on production and distribution activities.  In addition, large scale might help enhance the subsidiary’s reputation in the overseas market, because the large-scale operation is an indicator of the subsidiary’s commitment to a host country.[8]  Thus we hypothesize:

 

Hypothesis 1.  All else being equal, the scale of a foreign subsidiary is positively associated with the profitability of the subsidiary.

 

Experience.  Past studies have suggested that the experience in the local market positively affects the performance of foreign subsidiaries, because foreign firms with experience in a host country usually have more information about the local environment and accumulate better location-specific skills and know-hows than first-time foreign entrants.  Li (1995) shows that first-time foreign entrants to the U.S. computer and pharmaceutical industries are more likely to fail than repeat entrants.  Kogut and Chang (1996) find that the previous investments of Japanese electronics companies in the United States induces following investments, suggesting that initial investment serves as a platform for subsequent entry.  Moreover, the experience of parent firms in operating overseas subsidiaries can be transferable to subsidiaries in different countries (Johanson and Vahlne, 1977), because managers may build organizational routines that allow firms to efficiently expand abroad in many markets (Westney, 1998; Madhok, 1997).  These arguments suggest that operational experience in foreign markets can be accumulated at the firm level as well as the subsidiary level.  Thus we hypothesize:

 

Hypothesis 2-1.  All else being equal, the experience of a foreign subsidiary in a host country is positively associated with the profitability of the subsidiary.

Hypothesis 2-2.  All else being equal, the experience of a parent firm to operate foreign subsidiaries is positively associated with the profitability of the subsidiary.

 

Local Procurement.  The high level of supplier involvement in product development and production has been identified as one of the major sources of superior performance by Japanese automobile assemblers (Asanuma, 1985; Clark and Fujimoto, 1991).  As Japanese firms increasingly invest in overseas markets, many firms successfully replicate the Japanese-style tight supplier networks and practices in host countries (Cusumano and Takeishi, 1991).  The establishment of supplier networks requires relation-specific investments (Dyer, 1996), and so it takes time to successfully develop such networks.  The high local procurement ratio alone does not give us a clear prediction of its impact on subsidiary profitability.  When host countries have factor price advantages, the high local procurement ratio can contribute to the subsidiary profitability.  When Japanese subsidiaries are required to meet the local content requirement, a high local procurement ratio could negatively affect their performance, especially in the early stage of subsidiary development with undeveloped supplier networks.  Thus we hypothesize:

 

Hypothesis 3.  All else being equal, the local procurement ratio of a foreign subsidiary with a long experience in a host country is positively associated with the profitability of the subsidiary.

 

Local Sales.  For foreign subsidiaries, which are established to serve local markets, the greater local sales ratio is an indicator of the degree of penetration to local markets.  When Japanese MNEs have advantages over local firms, the greater penetration should result in higher profits.  Such a case is likely to occur in developed markets.  For subsidiaries which are established to take advantage of low factor costs and are intended as export bases to the third countries, however, the low local sales ratio (i.e., the high export ratio) contributes to the high profitability.  This situation is likely to occur for subsidiaries in developing countries.  Thus we test the following hypotheses:

 

Hypothesis 4a.  All else being equal, the local sales ratio of a foreign subsidiary is positively associated with the profitability of the subsidiary when it is located in a developed country.

Hypothesis 4b. All else being equal, the local sales ratio of a foreign subsidiary is negatively associated with the profitability of the subsidiary when it is located in a developing country.

 

            Equity Ownership.  The greater equity involvement of parent companies to foreign subsidiaries allows parent firms to maintain greater control.  Especially, when the source of advantages of MNEs is R&D or marketing know-how, their overseas operation requires detailed product knowledge, brand-development capabilities, updated technological information, and extensive after-sales service.  The assets necessary for these activities, including the specialized knowledge and working relationship, tend to be transaction-specific and develop over time through agent interaction (Williamson, 1979).  These activities also involve frequent exchanges of tacit knowledge and require high-level intra-firm coordination (Kogut, 1988), in turn requiring a high degree of control over subsidiaries.  Therefore, this type of MNE seeks tighter control over foreign activities.  It has been identified that the accumulation of intangible assets by Japanese firms Granger causes FDI (Berry and Sakakibara, forthcoming), motivating Japanese firms to maintain their control over foreign subsidiaries to achieve better performance.  In addition to receiving parent firms’ capabilities, foreign subsidiaries can develop their own capabilities.  Bartlett and Ghoshal (1989) argued that foreign subsidiaries which face various environments have opportunities to learn from these environments and develop competencies.  Such competencies can be utilized within an MNE worldwide through its integrated network.  A high degree of equity ownership would help capability building and retention in subsidiaries, resulting in their better performance.  Thus we hypothesize: 

 

Hypothesis 5.  All else being equal, the Japanese equity ownership ratio of a foreign subsidiary is positively associated with the profitability of the subsidiary.

 

R&D Intensity.  It has been argued that technological know-how constitutes the basis of the competitive advantages of many MNEs (Hymer, 1960; Caves, 1971; Buckley and Casson, 1976).  The R&D intensity of parent firms indicates its superior technological capabilities.  Thus we hypothesize:

 

Hypothesis 6.  All else being equal, the R&D intensity of a parent firm is positively associated with the profitability of the subsidiary.

 

 

Control Variables. 

Host-Country and Time-Specific Effect.  Host-country conditions such as market size, purchasing power, sophistication of markets and skilled workforces affect the profitability of foreign subsidiaries.  Macroeconomic conditions of host countries, most notably business cycles, also affect their profitability.  We control for these effects by including variables which reflect macroeconomic conditions of host countries. When a host country is not a major market for a subsidiary, these effects might be limited.  When a host country is a major market, however, conditions of a host country directly affect the profitability of a foreign subsidiary.  Therefore we include the interaction variable between the variable which reflect host countries’ macroeconomic conditions and the local sales ratio.

We also include a subsidiary’s degree of local manufacturing to test if there is any differential in profitability between manufacturing subsidiaries and others (mainly distribution subsidiaries) due to the possible difference in their margin structure.  Though we do not report the results, in some specifications we included several subsidiary-level variables.  Japanese employment ratio (the number of Japanese employees / total number of employees) was included to control for the degree of centralization of managerial control.  As Bartlett and Ghoshal (1989) argued, centralized organizational structure is suitable to achieve global efficiency, while it can hinder local responsiveness.  Capital-labor ratio was included to control for the difference in profitability due to capital intensity.  R&D intensity at the subsidiary level was included to examine whether the subsidiary-level R&D efforts, which can be used to localize the products or to build technological capabilities of subsidiaries, contribute to their profitability.  In order to control for regulatory restrictions which could prevent subsidiaries from achieving their optimum operation, a dummy variable for the local content requirement and a dummy variable for the restriction on repatriation were included.[9]  These variables appeared not to be significant.

 

Selection of Variables

Dependent Variable:

            The dependent variable is the return on sales, defined here as the gross profit over total sales.  This measure is chosen because this is a pre-tax measure, hoping to minimize the effect of local tax rules.[10]  Alternatively, the net profit over total sales is considered as an independent variable, but the correlation coefficient of both measures are 0.96 and both yield qualitatively similar results.  Therefore, the results with the gross profit over sales are reported here. 

One might argue that the return on investment is a better indicator of the profitability of a foreign subsidiary.  Since only the book value of a parent company’s investment to a subsidiary (or the book value of total assets) is available, this indicator introduces an upward bias to subsidiaries established earlier.  More practically, the data of investment and assets are only available for 1990, 1993 and 1996.  We therefore employ the return on sales as the dependent variable in this analysis.

Explanatory Variables:

            Scale.  The scale of a subsidiary is proxied for by the log of its sales, deflated by using the GDP deflator in 1990 prices.  Log is taken in order to avoid the identity of the equation.

            Experience.  The age of a subsidiary, which is the years passed after the establishment of a subsidiary is used as a proxy for the experience of the subsidiary.  The number of overseas subsidiaries of a parent firm is used as a proxy for a parent firm’s overseas experience.

            Local Procurement.  The local procurement ratio is defined as the share of procurement from a host country in total procurement.  In order to identify the effect of the local procurement for the subsidiaries with a long local experience, an interaction term is included in which the age of a subsidiary is multiplied by the local procurement ratio.

            Local Sales.  The local sales ratio is defined as the share of sales within a host country in total sales.

            Equity Ownership.  The Japanese equity ownership ratio is defined as the sum of the equity ownership by Japanese investors over total equity.

            R&D Intensity.  R&D expenditure over sales is used as a proxy for a parent firm’s R&D intensity.

            Host-Country and Time-Specific Effect.  GDP and GDP per capita are used as proxies for host-country conditions and business cycles.  Both figures are deflated by using the GDP deflator in 1990 prices.  In addition to these variables, interaction terms are included in which these variables are multiplied by the local sales ratio.

            Manufacturing Ratio.  A subsidiary’s manufacturing shipment over total sales is used as a proxy for the degree of local manufacturing.

            All the variables and their predicted signs are summarized in Table 1.

-----------------------------------

Table 1 about here

-----------------------------------

 

 

EMPIRICAL RESULTS

Estimation Methodology

We employ ordinary least square regression analyses with fixed effects.  The fixed effect model assumes that differences across subsidiaries can be captured in the constant term.  Operationally, we define transformations of our variables such that, for each subsidiary in each year, we subtract the mean of the variables for that subsidiary over time.  In the fixed effect model, all the industry-specific effects, and unmeasured country-specific, parent-specific and subsidiary-specific effects which might have been included in the error term fall out in the process of the transformation of the variables (i.e., these effects are controlled for), since they do not vary “within subsidiaries” over time.  This methodology allows us to focus on the variables we are concerned, which change over time.

One might argue that this type of regression analysis does not yield meaningful results because firms have the ability to shift profits from subsidiary to parent (or vice versa) to some extent (e.g., Eccles, 1985).  The fixed effect model eases the potential data problems which arise from the different local taxation policy and the resulting transfer pricing policy by Japanese MNEs.  As long as the taxation/transfer pricing policy does not change for a given subsidiary over the 1990–1996 period, this effect falls out.

The observations described in the second section indicate our data vary across host countries and regions.  In order to address this issue, we run separate regressions for each of four regions we are concerned — the United States, EU, East Asia and ASEAN.  The Chow test rejected the hypothesis that all the coefficients are the same across four regions at the .01 level, supporting the validity to run separate regressions by region.[11]

We also limit our empirical analysis to subsidiaries in manufacturing industries.  Our preliminary regression results indicate there are major differences in the results from manufacturing industries, service industries and primary industries.  By focusing on manufacturing industries, we hope to obtain results that are easy to interpret.[12]

The cost of this methodology is the reduced number of observations.  After these limitations, we are still able to obtain the total of 9,657 observations.  The reporting of parent-level variables is sparser than the subsidiary-level, so parent variables are used in selected specifications.  Summary statistics and the correlation matrix are presented in Tables 2 and 3.  Another cost of the fixed-effect estimates is that the standard goodness-of-fit measures are not informative.  Adjusted R2 values are reported as a reference. 

-------------------------------------

Tables 2 and 3 about here

-------------------------------------

 

The correlation matrix reveals that there is no high correlation between the dependent variable and each of explanatory variables.  The Japanese equity ownership ratio is highly positively correlated with GDP per capita, suggesting that Japanese firms would like to commit more to subsidiaries in a host country with favorable macro conditions.  Knowing this correlation, we would like to maintain both variables in the models in order to test which factor explains the profitability better. 

 

Results

            Table 4 reports the regression results when only subsidiary-level variables are used.  Columns 1 through 4 report the results of the United States, EU, East Asia and ASEAN, respectively, with GDP as the indicator of host-country and macroeconomic conditions.  The results show that the determinants of subsidiary profits differ across host regions, suggesting that the economic and institutional factors specific to host regions influence significantly the profit performances of overseas subsidiaries. 

Some effects are common across region.  The coefficient of log(sales) is positive and highly significant at the 1% level in all regions, indicating that size and implied market power and efficiency are positively associated with the profitability.

The local procurement ratio is positive and significant at the 10% level in East Asia.  This indicates that this region is a favorable sourcing base for Japanese subsidiaries.  In EU, local procurement ratio negatively affects the performance of subsidiaries, suggesting the local content requirements might negatively affect the performance.  The age of subsidiaries is also negative and significant, while the interaction term between age and the local procurement ratio is positive and significant in EU at the 5% level, as predicted.  A similar pattern is observed in ASEAN with the weaker significance, where the age is negative and significant and the interaction term between age and the local procurement ratio is positive and significant at the 15% level (but not at the 10% level).  In these two regions, it appears that just operating for a long period does not help profitability; rather it is the experience of dealing with local suppliers that contributes to the profitability.

In EU, the finding on the effect of the local procurement ratio is consistent with the often-cited difficulty of labor relationships in Europe (Fujimoto et al., 1994).  Strong unions, high wages, the difficulty to implement Japanese-style management practices such as the just-in-time system might prevent Japanese manufacturing subsidiaries from being profitable.  Only experienced firms can effectively deal with local suppliers.  It also suggests that in this region, the establishment of manufacturing subsidiaries is partly motivated by meeting the requirements of voluntary export restraints and local content.

The equity ownership ratio positively and significantly affects the performance in ASEAN.  The local sales ratio is not significant in any region.  In ASEAN, contrary to the prediction, the interaction term between GDP and the local sales ratio is negative and highly significant at the 1% level.  It appears that at a given level of GDP Japanese subsidiaries are more profitable if they sell products to Japan or to the third countries.

In East Asia, GDP is negative and highly significant.  This result and a previous finding on the positive effect of the local procurement suggest that this region contributes to the profitability of Japanese subsidiaries through their sourcing, not as their markets.

Columns 5 through 8 report the regression results with GDP per capita.  Overall, the results are similar to previous results with some interesting differences.  The findings of the effect of size, age, the interaction term between age and the local procurement ratio, and Japanese equity ownership ratio remained to be similar.

In ASEAN, the coefficients of the local sales ratio and GDP per capita are negative and highly significant, while the interaction term between GDP per capita and the local sales ratio is positive and highly significant at the 1% level.  This might indicate that local markets in ASEAN countries are attractive to Japanese subsidiaries only if the local markets have a high level of disposable income and sophisticated consumers who can afford and appreciate products Japanese firms produce and sell in these markets.

In Tables 4, the coefficient of the Japanese equity ownership ratio is not significant except for ASEAN.  It appears the degree of the ownership does not affect the profitability in more advanced economies, after controlling for other factors.  This issue requires further investigation since the determinants of the mode of entry to foreign markets have been extensively scrutinized in past literature.

Table 5 reports some results with parent-firm variables.  Columns 1 through 4 report the results with the number of subsidiaries by a parent firm.  Note these results are not compatible with Table 4 because the number of observations is reduced due to the data availability.[13]  As predicted, the number of subsidiaries by a parent firm is positive and significant in East Asia.  Also, the inclusion of this variable improves the explanatory power of some variables in this region, and now, as predicted, the Japanese equity ratio and the interaction term between GDP and the local sales ratio are positive and significant.  This indicates that this region is profitable as markets for Japanese subsidiaries only if the economic condition is favorable.  The inclusion of this variable also improves the explanatory power of some variables in ASEAN, where the interaction term between age and the local procurement ratio becomes positive and significant.  It appears the experience to deal with local suppliers contributes to the profitability, as is observed in EU in Table 4.

Columns 5 through 8 report the results with the R&D intensity of parent firms.  This variable is positive and significant in the United States, indicating Japanese subsidiaries are profitable in the United States when they possess technological capabilities.  The inclusion of this variable improves the explanatory power of the local sales ratio, GDP, and the interaction term between GDP and the local sales ratio.  It appears the United States is an attractive market for Japanese firms, and the favorable macroeconomic conditions in the United States contribute to its profitability.  The interaction term is negative and significant, indicating that these effects independently, not jointly, affect the profitability.

---------------------------------------

Tables 4 and 5 about here

---------------------------------------

 

 

 

CONCLUSIONS

This article has contributed to the existing literature by presenting new evidence on the pattern of subsidiary profits of Japanese MNEs in the 1990–1996 period.  At least two innovative attempts were made in this article: (1) a new panel of profits at the subsidiary level was presented; and (2) determinants of subsidiary profits were examined by using variables measuring subsidiaries’ local activities.  One of the most robust results obtained in the empirical analysis is that the determinants of subsidiary profits differ across host regions.  The estimated coefficients were found different jointly among subsidiaries located in the United States, Europe and Asia.  This result suggests that economic and institutional factors specific to host regions influence significantly the local activities of Japanese firms and hence their profit performances.

The size effect on the subsidiary profitability is present in all the regions, and it appears that subsidiaries in East Asia are profitable when this region is used as a sourcing and export base.  ASEAN is attractive as a market only if they exhibit favorable host-country market conditions, and in general, this region contributes to Japanese subsidiaries as an export base.  In EU and ASEAN, we find the local supplier networks contribute to the profitability of subsidiaries with local experience.  The experience of parent firms to operate overseas subsidiaries has a positive effect on their subsidiaries in East Asia.  Japanese subsidiaries are profitable in the United States when they possess technological capabilities, and the favorable macroeconomic conditions in the United States positively affect the profitability.

 This article has important implications for managers.  It has been argued that a firm’s FDI decision can often be driven by an imitative behavior of close rivals (Knickerbocker, 1973).  There are cases that an FDI decision is made beyond the reasons that are justifiable from profit-maximizing or strategic perspectives.  This research drives home the fundamental reality that the profitability of overseas subsidiaries reflects the market and firm-specific conditions.  A manager needs to take basic conditions into account in the FDI decisions rather than adopting a follow-the-leader behavior.

The empirical analysis of this article is by no means without deficiencies.  There are several omissions in terms of controls in the empirical specification, which need to be addressed in the future.  First, while we used subsidiary size as a proxy to measure the extent of market power and efficiency, it is crude at best for this purpose.  Ideally, we need to incorporate variables which measure the subsidiary’s competitive position relative to its rivals in the market such as market share.  Second, the article also does not control the extent of local competition in the host country’ market.  The industry concentration ratio such as the Herfindahl Index is available only for a limited number of countries, and the industry classifications do not correspond to what is used for the MITI data, making the matching of both data difficult.

This article suggests the presence of subsidiary- and parent-specific factors as well as the industry-specific and region/country-specific factors which play important roles in explaining the profitability of foreign subsidiaries.  A future research agenda includes an analysis of the relative importance of these factors.  At the same time, this article shows that both the role of host countries for foreign subsidiaries and the history of their local activities affect their performance.  This suggests that treating foreign subsidiaries equally across countries is misguided, and country- or region-specific issues need to be taking into account.  Further investigation into these issues is soundly warranted.









 


Table 1.  Description of Variables and Their Predicted Signs

 

Variable names

Definition

Predicted sign

ROS

Gross profit / total sales

Dependent variable

Log(Sales)

Log of sales in 1990 million yen

+

Age

The years passed after the establishment of a subsidiary

+

Local procurement ratio

Procurement from a host country / total procurement

+/-

Age * Local procurement ratio

Interaction term

+

Local sales ratio

Sales to a host country / total sales

+ Developed countries

- Developing countries

Japanese equity ownership ratio

Sum of the equity ownership by Japanese investors / total equity

+

Manufacturing ratio

Manufacturing shipment / total sales

+/-

GDP

GDP in 1990 billion yen

+/-

GDP per capita

GDP/population in 1990 thousand yen

+/-

GDP * Local sales ratio

Interaction term

+

GDP per capita * Local sales ratio

Interaction term

+

# of overseas subsidiaries by a parent firm

Number of overseas subsidiaries by a parent firm

+

Parent firm R&D intensity

R&D expenditure / sales

+

 

Note: GDP, GDP deflator, population data are taken from International Monetary Fund, International Financial Statistics for countries excluding Taiwan. Taiwanese data are taken from Taiwan Executive Yuan, Industry of Free China.

In the MITI data, sales were reported in current Yen.  They are converted to current local currency, deflated by using the GDP deflator in 1990 prices, and converted back to Yen using the 1990 exchange rate.

 


Table 2.  Summary Statistics

All manufacturing industries, all four regions

Variable names

Number of observations

Mean

Standard deviation

Minimum

Maximum

ROS

9657

-0.047

1.874

-74.000

149.33

Log(Sales)

9657

7.284

1.737

-0.131

13.635

Age

9657

11.653

7.975

1.000

83.000

Local procurement ratio

9657

0.575

0.349

0.000

1.000

Local sales ratio

9657

0.740

0.333

0.000

1.000

Japanese equity ownership ratio

9657

0.771

0.268

0.000

1.000

Manufacturing ratio

9657

0.649

0.501

0.000

12.361

GDP

9657

311.942

407.864

1.400

1000.80

GDP per capita

9657

2013.639

1369.219

51.287

4087.36

# of overseas subsidiaries by a parent firm

6292

22.263

47.166

0.000

439.000

Parent firm R&D intensity

1402

0.021

0.026

0.000

0.329

 

 

Table 3.  Correlation Matrix

All manufacturing industries, all four regions, number of observations: 9657

 

Variable names

1

2

3

4

5

6

7

8

9

10

1

ROS

1.00

 

 

 

 

 

 

 

 

 

2

Log(Sales)

0.08

1.00

 

 

 

 

 

 

 

 

3

Age

0.05

0.26

1.00

 

 

 

 

 

 

 

4

Local procurement ratio

-0.01

-0.14

-0.09

1.00

 

 

 

 

 

 

5

Local sales ratio

-0.01

-0.02

0.02

0.13

1.00

 

 

 

 

 

6

Japanese equity ownership ratio

0.00

0.14

0.00

-0.17

-0.11

1.00

 

 

 

 

7

Manufacturing ratio

-0.02

-0.03

-0.07

0.11

-0.06

-0.11

1.00

 

 

 

8

GDP

-0.01

0.19

-0.12

0.13

0.27

0.31

-0.06

1.00

 

 

9

GDP per capita

-0.01

0.25

-0.05

0.06

0.21

0.46

-0.14

0.80

1.00

 

10

# of overseas subsidiaries by a parent firm

0.00

0.17

0.04

0.01

0.02

-0.04

-0.03

0.04

0.01

1.00

11

Parent firm R&D intensity

0.02

0.11

0.00

-0.04

0.07

0.01

0.11

-0.02

-0.02

0.07

 

Note: Correlation coefficients concerning the number of overseas subsidiaries by a parent firm is calculated from 6,292 observations, and correlation coefficients concerning parent firm R&D intensity is calculated from 1,402 observations.


Table 4.  Determinants of Subsidiary Profitability

Dependent variable: ROS (gross profit/sales)

Fixed effect estimates, manufacturing industries

 

(1)USA

(2)EU

(3)East Asia

(4)ASEAN

(5)USA

(6)EU

(7)East Asia

(8)ASEAN

Log(Sales)

1.250     (16.518)***  

0.914 (16.830)***

0.597  (19.685)***

0.503     (18.658)***

1.211       (16.450)***

0.910  (16.775)***    

0.538  (18.449)***

0.473  (18.554)***

Age

 

-0.016 

(-0.579)

-0.056  

(-3.236)***

0.011  (1.223)

-0.018   

(-1.958)**

-0.034   

(-1.306)

-0.053  

(-2.966)***

-0.006 

(-0.620)

-0.030 

(-3.797)***

Local procurement ratio

-0.165

 (-0.456)

-0.556

(-2.330)**

0.291  (1.929)*

-0.154 

(-1.289)

-0.251   

(-.699)

-0.538  

(-2.255)**

0.238157 (1.553)

-0.095  

(-0.792)

Age * Local procurement ratio

0.0073   (0.272)

0.043 (2.346)**

-0.012

(-1.223)

0.012 (1.537)

0.014       (0.517)

0.041  (2.234)**

-0.008 

(-0.798)

0.0083   (1.070)

Local sales ratio

 

3.004      (0.598)

-0.008

(-0.027)

0.097  (0.460)

0.220  (1.236)

6.947      (0.776)

1.002    (1.226)

0.350  (1.435)

-0.592 

(-4.454)***

Japanese equity ownership ratio

0.452      (1.352)

0.011   (0.049)

0.310  (1.309)

0.268   (1.700)*

0.430      (1.285)

-0.024   

(-0.107)

0.161  (0.674)

0.273    (1.721)*

Manufacturing ratio

-0.010

(-0.111)

-0.089

(-1.274)

-0.020

(-0.514)

0.026   (0.908)

0.0063   (0.067)

-0.091    

(-1.296)

-0.007 

(-0.188)

0.034   (1.177)

GDP

-0.0027 

(-0.544)

-0.0019

(-0.529)

-0.018

(-4.986)***

0.0018  (0.127)

 

 

 

 

GDP * Local sales ratio

-0.0032 

(-0.592)

0.0002  (0.113)

0.0011 (0.267)

-0.049 

(-3.550)***

 

 

 

 

GDP per capita

 

 

 

 

 

-0.0002 

(-0.107)

0.00006   (0.182)

0.00002  (0.125)

-0.0002 

(-2.194)**

GDP per capita * Local sales ratio

 

 

 

 

-0.002 

(-0.769)

-0.0003 

(-1.217)

-0.0001 

(-1.056)

0.0003  (2.884)***

Number of observations

2810

1419

2653

2775

2810

1419

2653

2775

Adjusted R2

0.773

0.475

0.255

0.623

0.772

0.476

0.236

0.620

Note: T-statistics in parentheses.  *** significant at the 1% level, ** significant at the 5% level, * significant at the 10% level, using a two-tailed t-test.


Table 5.  Determinants of Subsidiary Profitability: with Parent Firm Variables

 Dependent variable: ROS (gross profit/sales)

Fixed effect estimates, manufacturing industries

 

(1)USA

(2)EU

(3)East Asia

(4)ASEAN

(5)USA

(6)EU

(7)East Asia

(8)ASEAN

Log(Sales)

1.527

(12.715)***

0.188

(12.400)***

0.374

(15.844)***

0.402

(15.888)***

0.279

(5.735)***

0.179

(3.137) ***

0.138

(2.540)**

0.158

(2.924)***

Age

 

-0.010

(-0.201)

-0.002

(-0.435)

0.011

(1.391)

-0.016

(-1.932)*

-0.031

(-1.105)

-0.010

(-0.521)

0.044

(1.587)

-0.002

(-0.162)

Local procurement ratio

0.001

(-0.002)

0.069

(1.062)

0.074

(0.542)

-0.277

(-2.307)**

-0.399

(-1.582)

-0.369

(-1.625)

0.550

(1.666)

-0.103

(0.530)

Age * Local procurement ratio

-0.002

(-0.044)

-0.004

(-0.834)

0.000

(-0.036)

0.018

(2.486)**

0.026

(1.156)

0.010

(0.606)

-0.027

(-1.111)

0.006

(0.409)

Local sales ratio

 

2.487

(0.323)

-0.064

(-0.610)

-0.186

(-1.091)

0.055

(0.316)

36.408

(3.249)***

-0.052

(-0.140)

0.506

(-0.937)

-0.335

(-0.710)

Japanese equity ownership ratio

0.387

(0.643)

0.131

(1.763)*

0.418

(2.268)**

0.430

(2.440)**

0.213

(0.501)

1.485

(1.836)*

0.346

(0.777)

0.065

(0.291)

Manufacturing ratio

0.109

(0.677)

-0.028

(-1.586)

-0.025

(-0.636)

0.005

(0.160)

-0.038

(-0.420)

0.086

(1.161)

0.052

(-0.494)

0.156

(1.831)*

GDP

-0.006

(-0.765)

-0.001

(-0.826)

-0.018

(-5.997)***

-0.013

(-0.950)

0.042

(3.740)***

0.003

(1.252)

-0.010

(-0.880)

-0.031

(-0.868)

GDP * Local sales ratio

-0.003

(-0.304)

0.001

(0.984)

0.006

(1.754)*

-0.020

(-1.389)

-0.039

(-3.179)***

-0.001

(-0.427)

0.010

(0.814)

0.029

(0.788)

# of overseas subsidiaries by parent firm

-0.0002

(-0.056)

0.0005

(0.872)

0.002

(1.653)*

0.001

(0.968)

 

 

 

 

Parent firm R&D intensity

 

 

 

 

2.373

(1.695)*

1.310

(0.336)

0.615

(0.482)

-0.349

(-0.342)

Number of observations

1838

892

1769

1793

452

213

373

364

Adjusted R2

0.775

0.988

0.440

0.827

0.934

0.998

0.916

0.925

Note: T-statistics in parentheses.  *** significant at the 1% level, ** significant at the 5% level, * significant at the 10% level, using a two-tailed t-test.


Reference

Andersson T, Fredriksson T. 1996. Distinction between Intermediate and Finished Products in Intra-Firm Trade. International Journal of Industrial Organization 18: 773–792.

Asanuma B. 1985. The Contractual Framework for Parts Supply in the Japanese Automotive Industry, and The Organization of Parts Supply in the Japanese Automotive Industry. Japanese Economic Studies 15: 32–78.

Barney JB. 1986. Strategic Factor Markets: Expectations, Luck, and Business Strategy. Management Science 32: 1231–1241.

Bartlett CA, Ghoshal S. 1989. Managing across Borders: The Transnational Solution. Harvard Business School Press: Boston.

Belderbos R. 2003. Entry Mode, Organizational Learning, and R&D in Foreign Affiliates: Evidence from Japanese Firms. Strategic Management Journal 24: 235-259.

Belderbos RG, Capannelli G, Fukao K. 1998. Local Procurement by Japanese Electronics Firms in Asia.  Paper presented at the NBER Ninth Annual East Asian Seminar on Economics, Osaka.

Benvignati AM. 1990. Industry Determinants and Differences in U.S. Intrafirm and Arms-Length Exports. Review of Economics and Statistics 72: 481–488.

Berry H, Sakakibara M. forthcoming. Resource Accumulation and Overseas Expansion by Japanese Multinationals. Journal of Economic Behavior and Organization.

Birkinshaw J, Hood N. 1998. Multinational Subsidiary Evolution: Capability and Charter Change in Foreign-Owned Subsidiary Companies. Academy of Management Review 23: 773-795.

Birkinshaw J, Hood N, Jonsson S. 1998. Building Firm-Specific Advantages in Multinational Corporations: The Role of Subsidiary Initiative. Strategic Management Journal 19: 221-241.

Buckley P, Casson M. 1976. The Future of the Multinational Enterprise. Macmillan: London.

Casson M. and associates. 1986. Multinationals and World Trade: Vertical Integration and the Division of Labor in World Industries. Allen & Unwin: London.

Caves RE. 1971. International Corporations: The Industrial Economics of Foreign Investment. Economica 38: 1–27.

Caves RE, Gale BT, Porter ME.  1977. Interfirm Profitability Differences: Comment. Quarterly Journal of Economics 91: 667–675.

Clark KB, Fujimoto T. 1991. Product Development Performance: Strategy, Organization, and Management in the World Auto Industry. Harvard Business School Press: Boston MA.

Connor JM.  1977. The Market Power of Multinationals: A Quantitative Analysis of U.S. Corporations in Brazil and Mexico. Praeger: New York, NY.

Cusumano MA, Takeishi A. 1991. Supplier Relations and Management: A Survey of Japanese, Japanese-Transplant, and U.S. Auto Plants. Strategic Management Journal 12: 563–588.

Demsetz H. 1973. Industry Structure, Market Rivalry, and Public Policy. Journal of Law and Economics 16: 1–9.

Dierickx I, Cool K. 1989. Asset Stock Accumulation and Sustainability of Competitive Advantage. Management Science 35: 1504–1511.

Domowitz I, Hubbard RG, Petersen BC. 1986. Business Cycles and the Relationship Between Concentration and Price-Cost Margins. Rand Journal of Economics 17: 1–17.

Dyer JH. 1996. Specialized Supplier Networks as a Source of Competitive Advantage: Evidence from the Auto Industry. Strategic Management Journal 17: 271–291.

Eccles RG. 1985. The Transfer Pricing Problem: A Theory for Practice. Lexington Books: Lexington, MA.

Fujimoto T, Nishiguchi T, Sei S. 1994. The Strategy and Structure of Japanese Automobile Manufacturers in Europe. In Mason M, Encarnation D (eds.), Does Ownership Matter? Japanese Multinationals in Europe. Oxford University Press: Oxford.

Gerybadze A, Reger G. 1999. Globalization of R&D: Recent Changes in Management of Innovation in Transnational Corporations. Research Policy 28: 251-274.

Ghoshal S, Bartlett C. 1990. The Multinational Corporation as an Interorganizational Network. Academy of Management Review 15: 603-625.

Hackett SC, Srinivasan K. 1998. Do Supplier Switching Costs Differ Across Japanese and US Multinational Firms. Japan and the World Economy 10: 13–32.

Helleiner GK, Lavergne R. 1979. Intra-Firm Trade and Industrial Exports to the United States. Oxford Bulletin of Economics and Statistics 41: 297–311.

Hymer, S. 1960. The International Operations of National Firms: A Study of Direct Foreign Investment. Ph.D. Dissertation, Massachusetts Institute of Technology.

Jarret, JP. 1979. Offshore Assembly and Production and the Internalization of International Trade within the Multinational Corporation. Ph.D. dissertation, Harvard University.

Johanson J, Vahlne JE. 1977. The Internationalization Process of the Firm. Journal of International Business Studies 8: 23–32.

Knickerbocker FT. 1973. Oligopolistic Reaction and Multinational Enterprise. Division of Research, Graduate School of Business Administration, Harvard University: Boston MA.

Kogut B. 1988. Joint Ventures: Theoretical and Empirical Perspectives. Strategic Management Journal 9: 319–332.

Kogut B, Chang SJ. 1996. Platform Investments and Volatile Exchange Rates: Direct Investment in the U.S. by Japanese Electronic Companies. Review of Economics and Statistics 78: 221–231.

Kumar N. 1990. Multinational Enterprises in India: Industrial Distribution, Characteristics, and Performance. Routledge: London.

Kuemmerle W. 1999. The Drivers of Foreign Direct Investment into Research and Development: An Empirical Investigation. Journal of International Business Studies 30: 1-24.

Lall S. 1978. The Pattern of Intra-Firm Exports by US Multinatioanls. Oxford Bulletin of Economics and Statistics 40: 209–222.

Lecraw DJ. 1983. Performance of Transnational Corporations in Less Developed Countries. Journal of International Business Studies 14: 15–33.

Li J. 1995. Foreign Entry and Survival: Effects of Strategic Choices on Performance in International Markets. Strategic Management Journal 16: 333–351.

Machin S, Van Reenen J. 1993. Profit Margins and the Business Cycle: Evidence from UK Manufacturing. Journal of Industrial Economics 41: 29–50.

Madhok A. 1997. Cost, Value and Foreign Market Entry Mode: The Transaction and the Firm. Strategic Management Journal 18: 39–61.

Markusen JR. 2002. Multinational Firms and the Theory of International Trade. MIT Press, Cambridge, MA.

Markusen JR, Venables AJ. 1996. The Increased Importance of Multinationals in North American Economic Relationships: A Convergence Hypothesis. In Canzoneri MW, Ethier WJ, Grilli V (eds) The New Transatlantic Economy. Cambridge University Press, London, 169-188.

Mancke RB. 1974. Causes of Interfirm Profitability Differences: A New Interpretation of the Evidence. Quarterly Journal of Economics 88: 181–193.

Nohria N, Ghoshal S. 1994. Differentiated Fit and Shared Values: Alternatives for Managing Headquarters-Subsidiary Relations. Strategic Management Journal 19: 491-502.

Rumelt RP. 1991. How Much Does Industry Matter? Strategic Management Journal 12: 167–185.

Schmalensee R. 1985. Do Markets Differ Much? American Economic Review 75: 341–351.

Wernerfelt B. 1984. A Resource-Based View of the Firm. Strategic Management Journal 5: 171–180.

Wernerfelt B, Montgomery CA. 1988. Tobin’s q and the Importance of Focus in Firm Performance. American Economic Review 78: 246–250.

Westney DE. 1988. Domestic and Foreign Learning Curves in Managing International Cooperative Strategies, in Contractor FJ, Lorange P (eds), Cooperative Strategies in International Business, 339–346. Lexington Books: Lexington.

Williamson OE. 1979. Transaction Cost Economics: The Governance of Contractual Relations. Journal of Law and Economics 22: 233–262.



[1] The existing research is confined in particular host countries such as Mexico and Brazil (Connor, 1977), Southeast Asia (Lecraw, 1983), and India (Lall, 1976; Kumar, 1990).

[2] All financial figures are answered in million yen, by using the nominal exchange rate table provided by MITI.

[3] Average real GDP growth rates for EC countries excluding Greece

[4] By the same token, the observation that the profits reported for subsidiaries located in ASEAN and East Asia are larger than the profits for subsidiaries in the United States and EU, may reflect the underlying difference in aggregate growth between these regions.  As is well known, the economies of ASEAN and East Asian countries grew rapidly during the period in question compared to those in the United States and the EU.  Annual growth rates of real GDP for countries such as Malaysia, Thailand, Indonesia, Singapore, Hong Kong, Taiwan, South Korea, and China were at least twice as high as those in North America and Europe, ranging between 6% and 12% during the 1990–1996 period.

[5] See Demsetz (1973), Mancke (1974), and Caves et al. (1977).

[6] See Wernerfelt (1984), Barney (1986), and Dierickx and Cool (1989).

[7] The presence of industry-and firm-specific effects on profitability has been examined previously.  See, for example, Schmalensee (1985), Wernerfelt and Montgomery (1988), and Rumelt (1991).

[8] One might argue foreign subsidiaries could exhibit diseconomies of scale.  Given that the average age of subsidiaries in the sample is 11.7 years, it is unlikely that these subsidiaries have reached the scale that would exhibit the diseconomies of scale.  Emprically, we tested specifications including the quadratic term of a scale measure.  Though the coefficient of the quadratic term has a negative sign, the correlation coefficient between the scale measue and its quadratic term is 0.98, making it difficult to draw appropriate inference on the individual signs.

[9] Capital-labor ratio and R&D intensity are available for observations in 1990 and 1993 only.

[10] Of course, if firms want to minimize foreign tax payments, they will manipulate the pre-tax accounting figures. 

[11] The test statistics are 7.90 and 6.80 for specifications in columns 1 through 4 and columns 5 through 8 in Table 4, respectively, which exceed the critical F value of 1.74 at F(27,5401) required to reject the homogeneous coefficient hypothesis at the .01 level.

[12] We also excluded pure distribution subsidiaries from the sample because we expect a different set of determinants of profitability for these subsidiaries.

[13] The number of overseas subsidiaries is available in 1990 through 1993 and 1996, and the parent firm R&D intensity is available in 1990 and 1993 only.