The choice between standard and non-standard FDI production strategies for Taiwanese multinationals

G Model

ARTICLE IN PRESS

RESPOL-3019; No. of Pages 11

Research Policy xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Research Policy journal homepage: www.elsevier.com/locate/respol

The choice between standard and non-standard FDI production strategies for Taiwanese multinationals Hui-Lin Lin a,1 , Yi-Chi Hsiao b , Eric S. Lin c,d,∗ a

Department of Economics, National Taiwan University, Taipei 100, Taiwan Department of Strategy and Management, National Taiwan University, Taipei 100, Taiwan Department of Economics, National Tsing Hua University, Hsin-Chu 30013, Taiwan d Public Economics Research Center, National Taiwan University, Taipei 100, Taiwan b c

a r t i c l e

i n f o

Article history: Received 15 April 2013 Received in revised form 19 June 2014 Accepted 21 June 2014 Available online xxx JEL Classification: F21 F23 O32 Keywords: FDI Vertical integration Horizontal integration Multinomial logit model

a b s t r a c t In modeling the foreign direct investment (FDI) behavior of firms, existing studies tend to classify multinational enterprises’ (MNEs) production activities into two broad strategies, namely, standard vertical integration and horizontal integration. In practice, the production behavior of multinational enterprises is much more complex. Based on the production relations between parent firms and their foreign subsidiaries, this study divides MNEs’ production behavior into five strategies: vertical integration, foreign concentration, home concentration, horizontal integration, and heterogeneous horizontal integration. We then propose three sets of interesting hypotheses associated with the five different production strategies. A uniquely compiled firm-level data set of Taiwanese manufacturing firms over the 2004–2007 period is adopted to empirically verify the proposed hypotheses. The empirical estimates are in general consistent with the proposed predictions and suggest that the various production behaviors of MNEs depend on the foreign market size, trade costs, factor advantage, fixed investment costs and firm characteristics. Moreover, each production strategy is influenced by different determinants as implied by our hypotheses. © 2014 Elsevier B.V. All rights reserved.

1. Introduction The overseas production strategies of multinational enterprises (MNEs) are becoming increasingly complicated nowadays. It is therefore not sufficient to describe a firm’s foreign direct investment (FDI) behavior by simply dividing it into horizontal and vertical integration strategies. In particular, horizontal multinationals include not only the “replicators” that are involved in producing replicas of the home plants in foreign countries, but also those that produce different final products in different countries. Head (2007) refers to these types of overseas production strategies as consisting of “polycentric” and “polymorphic” forms, where foreign subsidiaries produce different final products for the parent firms.2 The decision between pure replication and heterogeneous

∗ Corresponding author. Tel.: +886 3 574 2729; fax: +886 3 562 9805. E-mail addresses: [email protected] (H.-L. Lin), [email protected] (Y.-C. Hsiao), [email protected] (E.S. Lin). 1 Tel.: +886 2 2322 5657. 2 According to Head (2007), the difference between “polycentric” and “polymorphic” lies in the pattern of final shipments. The “polycentric” form involves cross-border trade, while the “polymorphic” form does not.

production could depend on firm-specific factors (such as firm size and productivity) or demand-side variables (such as market size and consumer preferences), and the determinants of each production strategy might also differ.3 Likewise, vertical multinationals may include not merely firms that geographically fragment production by stages, but also those which concentrate their production in the home (or foreign) country, and set up sales offices in the foreign (or home) country. Head (2007) labels this fragmented production as “vertical specialization”, home production/foreign sales as “home centralization”, and foreign production/home sales as “foreign centralization”. Hence, it would be interesting and important to further categorize horizontal as well as vertical multinationals into more detailed production patterns and to conduct a comprehensive comparison of the various production strategy patterns of MNEs, thereby helping us gain a more thorough understanding of the behavior of multinationals.

3 For example, foreign market size might be the key factor in the formation of pure replication multinationals, while firms engaging in product differentiation in different markets might place more emphasis on consumer preference differentials between the home and foreign markets, and the costs induced by product differentiation.

http://dx.doi.org/10.1016/j.respol.2014.06.005 0048-7333/© 2014 Elsevier B.V. All rights reserved.

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005

G Model RESPOL-3019; No. of Pages 11

ARTICLE IN PRESS H.-L. Lin et al. / Research Policy xxx (2014) xxx–xxx

2

Following the FDI production strategies in Head (2007), this paper first classifies the potential production strategies of MNEs into three vertical and two horizontal choices, namely, vertical integration, foreign concentration, home concentration, horizontal integration, and heterogeneously horizontal integration. The definitions of two vertical-type integration strategies, foreign concentration and home concentration, will be introduced in Section 3. It is not an easy task to empirically disentangle the factors that influence firms’ decision regarding these different integrated production strategies. The difficulty arises since a typical firm-level data set does not allow us to uncover which production strategy a firm is adopting among the five alternatives. This article compiles a firm-level data set (also by collecting and constructing data on industry-level and country-level variables) of Taiwanese multinational manufacturers during the years from 2004 to 2007. This rich and unique data set enables us to control for countrylevel, industry-level, and firm-level determinants and to analyze the firms’ various production strategies in FDI. The data of this paper are taken from the annual survey on the outward FDI of Taiwanese manufacturing firms by the Ministry of Economic Affairs (MOEA) of Taiwan. The targets of the survey include firms that undertake green-filed outward investment or M&A. In addition, the survey does not distinguish between outward FDI firms and multinationals (MNEs or MNCs), since the operational definition of an MNE (or MNC) may not be relatively standard compared to that of FDI (Aggarwal et al., 2011). The MNCs may be defined on the basis of firm characteristics, such as the proportion of foreign sales and foreign assets, the number of foreign subsidiaries, or on the basis of the degree of international activities, such as operating across national borders, employing foreign capital, people and processes. Because of the implementation of the national policy of economic liberalization and internationalization, Taiwanese firms are not required to apply for permission to undertake outward FDI unless the investment is above a certain amount of money. According to the Regulation for Companies Overseas Investment by the MOEA of Taiwan, a firm is required to apply for permission if the amount of overseas investment exceeds NT$1.5 billion (around US$50 million); if the amount of investment is below that level, a firm can report to the Investment Commission of the MOEA within 6 months. Furthermore, according to the statistics report of 2012 from the Investment Commission of the MOEA, the average amount of overseas investment is US$25,229,412, which is much higher than the average business scale in Taiwan (around US$266,667). This means that Taiwanese business entities are very active in overseas investment.4 With the identified FDI strategies from multinational firms, five interesting strategy-pairs are explored and three sets of hypotheses are thus proposed to reveal the crucial factors for each strategypair. In particular, this study addresses the following questions. First, what are the strengths and weaknesses of each overseas production strategy in terms of our targeted five strategy-pairs? In addition to the traditional choice between horizontal integration and vertical integration, we mainly investigate the choice among two types of horizontal integration and the choice among three types of vertical integration strategies. Second, which aggregate (e.g., country-level and industry-level) variables differentiate the choices among various production strategies for multinational enterprises? Can each multinational production strategy have different aggregate determinants? Third, which firm-specific characteristics affect the integration production strategies? Are these firm-level determinants specific to various production strategies for MNEs? These questions are empirically evaluated in this paper.

4

We owe the discussion on Taiwanese MNEs and FDI to an anonymous referee.

The remainder of this paper is organized as follows. The next section consists of a brief literature review on horizontal and vertical FDI activities. We introduce the firm’s choices among different production strategies in FDI and propose three sets of hypotheses that will be tested in this paper in Section 3. Section 4 describes the data sources. The empirical specifications and estimation strategies are presented in Section 5. Section 6 demonstrates and discusses the empirical results. The final section concludes the paper. 2. Literature review on FDI Boosted by trade liberalization and the economic integration of the world economy, trade and foreign direct investment have been among the fastest growing economic activities around the world over the past two decades. Specifically, multinational sales have grown at high rates and have outpaced the expansion of trade in manufacturing products, highlighting the importance of FDI for firms to maintain their competitiveness in the global market and receiving wide attention from policy-makers, researchers, and practitioners.5 Based on a two-country equilibrium model, most existing FDI theories, e.g., Helpman (1984), Markusen (1984), Helpman and Krugman (1985), Markusen and Venables (1998, 2000), and Carr et al. (2001), classify multinationals into two broad types, namely, vertical integration and horizontal integration, and examines firms’ endogenous choice between vertical and horizontal production structures. Markusen and Maskus (2001) also offer a nice review of the FDI literature that adopts a general-equilibrium trade-theoretic view of multinationals. For vertical multinationals, firms aim to exploit factor price differences across countries by fragmenting the production process internationally and moving production lines to foreign countries where the production cost is lower. On the other hand, horizontal multinationals avoid trade costs by producing the same goods and services in various countries in order to get closer to customers. In practice, the choice between vertical and horizontal integration for a multinational depends mainly on country-level characteristics, such as relative market size, factor prices, and trade and investment costs. Most empirical studies show that: (1) the relative size of the host market has a negative (positive) impact on the formation of vertical (horizontal) multinationals, (2) trade costs have a negative (positive) impact on the propensity of being a vertical (horizontal) multinational, and (3) vertical multinationals engage in production in countries having significant factor price differentials, whereas horizontal multinationals are more prevalent in countries having similar factor prices (see Shatz and Venables, 2000; Markusen and Maskus, 2001; Lipsey, 2003; Barba-Navaretti and Venables, 2004; Bandinger and Egger, 2008 for more detailed discussions). Recent studies on the determinants of FDI and trade alternatively emphasize the role of firm heterogeneity (Antrás and Helpman, 2003; Buch et al., 2005; Castellani and Zanfei, 2007; Head and Ries, 2003; Helpman et al., 2004; Kimura and Kiyota, 2006), and especially focus on how productivity affects firms’ overseas integrated production strategies, e.g., Kimura and Kiyota (2006) and Castellani and Zanfei (2007). The overseas production strategies of multinational enterprises, however, are essentially quite complex, suggesting that the rough classification on FDI into vertical and horizontal integration may prevent us from gaining more insights into the choice of production strategies undertaken by MNEs. Using a sample of U.S.

5 According to the United Nations Conference on Trade and Development (UNCTAD) World Investment Report 2009, global FDI flows have grown at least twice as fast as trade over recent decades. In 2008, the global FDI flows were about US$1697 billion, and the total FDI stock reached US$14,909 billion.

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005

G Model RESPOL-3019; No. of Pages 11

ARTICLE IN PRESS H.-L. Lin et al. / Research Policy xxx (2014) xxx–xxx

multinationals with affiliates in Canada, Feinberg and Keane (2003) reported that 69% of firms are actually “hybrid” in the sense that they are implementing more complex integration strategies. Kao and Lin (1993) classified Taiwanese outward FDI to China into three types, namely, division of production, division of business activities, and division of products, in order to analyze the impact of outward FDI on Taiwanese industries. Shi and Gregory (1998) grouped international manufacturing networks according to three types, namely, the regional approach, the multinational approach, and worldwide. Theoretically, Yeaple (2003) and Ekholm et al. (2003) modeled a firm as having complex integration strategies to explain why MNEs may engage in vertical integration in some countries and horizontal integration in others. Grossman et al. (2006) further developed a theory to shed light on the optimal integration strategies of multinationals when firms face an even richer array of choices among international organizations than those in Yeaple (2003) and Ekholm et al. (2003). 3. Conceptual framework and hypotheses As mentioned in Section 1, Head (2007) argues that both conventional vertical integration and horizontal integration can be further divided into more specific production strategies to reflect more realistic operations of multinationals. In this section we follow Head (2007) closely to propose five multinational production strategies and then come up with several interesting hypotheses to test empirically. 3.1. Types of integrated overseas production strategies 3.1.1. Vertical integration and horizontal integration First of all, we consider vertical integration to be one of the integration production strategies of a multinational firm. Specifically, standard vertical integration is defined as an international fragmentation of production,6 while standard horizontal integration is referred to as duplicating the same production activity in different countries. These two strategies are the most prevalent and widely discussed production forms in the existing literature. Conventionally, many studies refer to firms engaging in vertical integration as vertical multinationals and to firms engaging in horizontal integration as horizontal multinationals (Buch et al., 2005; Carr et al., 2001; Markusen and Venables, 1998, 2000). The main motivation for MNEs to adopt vertical integration is to exploit differences in relative factor costs. By reallocating labor-intensive production stages to countries with lower labor costs, firms can benefit from engaging in vertical integration as long as the costs of fragmentation (e.g., tariffs, transportation costs, additional costs for acting in other countries, etc.) are lower than the cost savings. On the other hand, firms conducting horizontal integration aim to avoid transportation costs or to gain access to foreign markets. Horizontal integration is beneficial if the savings in variable costs associated with tariffs and transportation outweigh the additional fixed costs of setting up a new plant. 3.1.2. Foreign concentration and home concentration To be more precise, vertical multinationals are characterized by firms internationally integrating the production of their inputs or supplies and the distribution of their outputs or finished products (e.g., Avenel and Barlet, 2000). Multinationals can thus fragment

6 There are several studies using different terms to describe the geographical separation of production strategies, for instance, “disintegration of production” in Feenstra (1998), “slicing the value chain” in Krugman (1996), “delocalization” in Leamer (1996), “global integrated manufacturing” in Shi and Gregory (1998), and “vertical specialization” in Head (2007).

3

production activities into several stages to take advantage of foreign factor prices, or they can concentrate the production in one place to exploit economies of scale. In other words, firms may produce, distribute and market the final goods in different locations. Following Head (2007), we distinguish two other forms of (nonstandard) production strategies adopted by vertical multinationals. In this article, we define foreign concentration (hereafter) as a production strategy to emphasize that firms shut down home plants and establish foreign affiliates to manufacture commodities, while leaving parent firms to take charge of distributing or marketing products to the end market (Kao and Tsai, 2006; Protsenko, 2003). It is worth noting that, according to Kao and Tsai (2006), most Taiwanese multinationals treat their subsidiaries in mainland China as major manufacturing bases, while the Taiwanese parent firms are in charge of marketing, financial scheduling and R&D activities. This type of (vertical) multinational activity is very similar to what we refer to as foreign concentration in this study. In contrast to foreign concentration, home concentration (hereafter) represents a strategy by which multinationals set up foreign subsidiaries to market the final products, while using home plants to produce commodities. A concrete example of home concentration is Spec Seats Inc. (a professional chair maker with a world famous brand), which is actually a Taiwanese manufacturer of high-end portable folding seating systems. Spec Seats Inc. engages in R&D, manufacturing and production in Taiwan, and then sells products in the United States, where it has been chosen to be the official supplier for the NCAA. 3.1.3. Heterogeneous horizontal integration As discussed in Section 1, Head (2007) argues that horizontal multinationals comprise not only firms duplicating their production process in each country, but also those producing a variety of final products in different countries. The former corresponds to the standard horizontal integration. For the latter case, this study specially emphasizes an FDI strategy in which the firm produces mid- and high-end products in the home country and produces low-end ones through foreign affiliates. This non-standard form of MNEs’ production activities is referred to as heterogeneous horizontal integration hereafter. Taking bicycle manufacturers in Taiwan as an example, Hsu and Chen (2002) found that, instead of serving the bicycle market in China through exports, Taiwanese bicycle manufacturers set up production lines in China to produce low-end or low value-added bicycles and took advantage of high-skilled Taiwanese workers to simultaneously produce mid/high-end products at home. 3.2. Choosing among production strategy pairs According to the introduction of firms’ integrated overseas production strategies, we let V1 denote a firm that engages in a typical vertical integration production strategy, V2 be the type of foreign concentration strategy, V3 be the type of home concentration strategy, H1 be the typical horizontal integration strategy, and H2 represent the heterogeneous horizontal integration strategy. In other words, we have the set of firms’ production strategies l, l ∈ {V1, V2, V3, H1, H2} . If a multinational chooses a specific strategy, say V1, then it is obvious that it is not better off engaging in other strategies (i.e., V2, V3, H1 and H2). By performing a multinomial type of regression model using an actual firm-level data set, one is able to figure out the determinants that V1 is superior to V2 and others. However, it is also interesting to gain a clearer understanding for whether a firm engages in one of the two types of vertical integration strategies, for instance, the advantage of conducting a foreign concentration, V2 over a home concentration strategy, V3. In principle, we have C25 = 10 strategy pairs for a firm’s choice among any two types of the five production

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005

G Model RESPOL-3019; No. of Pages 11

ARTICLE IN PRESS H.-L. Lin et al. / Research Policy xxx (2014) xxx–xxx

4

strategies. However, we do not aim to compare all possible pairs of these strategies exhaustively, but rather focus on the following five interesting combinations, which are vertical integration (V1) vs. horizontal integration (H1), vertical integration (V1) vs. foreign concentration (V2), vertical integration (V1) vs. home concentration (V3), foreign concentration (V2) vs. home concentration (V3), and horizontal integration (H1) vs. heterogeneous horizontal integration (H2). In the following, we propose three sets of hypotheses associated with the five FDI strategy combinations. Fig. 1 depicts the structure of our proposed three sets of hypotheses. The first set (Hypothesis Set I) focuses on comparing conventional vertical integration and horizontal integration (V1 and H1). The second set (Hypothesis Set II) investigates three types of vertical integration strategies (V1, V2 and V3), while the last set (Hypothesis Set III) concentrates on the typical horizontal integration and heterogeneous horizontal integration (H1 and H2). We specifically want to know the niche of utilizing, say, V1 relative to H1, which can be reflected in our empirical framework.7 In fact, since vertical integration (V1) and horizontal integration (H1) are the most prevalent multinational production strategies discussed in the literature, considering that the two important choices enable us to compare our results with those in previous studies. In addition, it is noted that the choice among various production strategies involves a consideration of market advantage, factor advantage, trade costs, fixed investment costs and firm characteristics. If the foreign market size is larger, then a firm can get closer to the market and save transportation costs by engaging in horizontal integration. An export-oriented firm is more likely to choose the vertical integration strategy. Since a firm engaging in horizontal integration would produce similar products in the foreign country, it will not be necessary to export, and hence it can reduce its export ratio. A large firm is more capable of undertaking the international fragmentation of production than a small firm. In addition, it may require a higher technology level for a multinational to engage in vertical integration than if choosing horizontal integration, which basically duplicates the same production activities in both host and foreign countries. For the comparison between vertical integration (V1) and horizontal integration (H1), our first set of hypotheses can be summarized as follows: Hypothesis Set I Hypothesis 1a. The foreign market size positively affects the probability of undertaking horizontal integration (H1). Hypothesis 1b. The larger the multinational firm is, the higher the probability is that the firm will have to adopt vertical integration (V1). Hypothesis 1c. The larger the export ratio the multinational firm has, the higher the probability is that the firm will have to adopt vertical integration (V1). Hypothesis 1d. The higher the R&D intensity the multinational firm has, the higher the probability is that the firm will have to adopt vertical integration (V1). Since our unique data set allows a firm to specify which type of vertical integration strategies were actually adopted, the determinants (especially the concern of factor costs) affecting a vertical multinational regarding where to place its production and distribution of products can be explored. The following hypothesis

7 As will be documented later on in Section 5, this comparison can be made through a multinomial logit model by choosing an appropriate reference group. Please refer to footnote 10 for more details.

Table 1 2004–2007 overseas investment areas of Taiwanese MNEs. Area

Number of firms (proportion)

United States Canada Mexico Central &South America Western Europe Eastern Europe Hong Kong China Japan Malaysia Singapore Thailand Indonesia Philippines Vietnam South Asia Australia &New Zealand Africa Others

532 (8.1%) 8 (0.1%) 7 (0.1%) 38 (0.6%) 56 (0.9%) 10 (0.2%) 263 (4.0%) 4713 (71.5%) 58 (0.9%) 107 (1.6%) 50 (0.8%) 127 (1.9%) 77 (1.2%) 66 (1.0%) 192 (2.9%) 9 (0.1%) 12 (0.2%) 21 (0.3%) 242 (3.7%)

Total

6588 (100.0%)

Data Source: Outward FDI Survey in Manufacturing, 2004–2007, MOEA, Taiwan.

associated with different types of vertical integration (i.e., V1, V2 and V3) is of particular interest to us because we can gain more insights into how vertical multinationals make their domestic and overseas production decisions. We note that it can be expected that a home concentration firm (V3) is less experienced in investing overseas than a foreign concentration firm (V2) as well as a typical vertical integration firm (V1). A larger firm is associated with the ability to invest overseas and hence produce in foreign countries. This is because a large firm is able to cover the expenditure on overhead for producing overseas. Moreover, since on average Taiwanese manufacturing firms are high-tech oriented and technology intensive, they produce high value-added products, which are easy to compete with in the global market. It can be shown in Table 1 that about 71.5% of Taiwanese MNEs undertook overseas investment in China from 2004 to 2007, where the technology level is much lower than in Taiwan. Therefore, those high-tech firms tend to stay at home to produce their main products rather than move the production line abroad for a cost reduction. We expect that a firm with higher R&D and capital intensity is associated with choosing the home concentration strategy, V3. A higher export ratio means that a firm seeks to reduce its production costs by moving its production line overseas, implying a higher propensity to engage in V2 or V1. Furthermore, due to the concern of factor price in regard to different vertical integration strategies, a firm has a tendency to undertake V3 if the foreign wage rate is higher than the domestic wage. For the choice among the three vertical-type integration strategies (V1, V2 and V3), we then propose the following hypotheses: Hypothesis Set II Hypothesis 2a. The longer the multinational firm invests overseas, the higher the probability is that the firm will have to adopt foreign concentration (V2) and typical vertical integration (V1). Hypothesis 2b. The larger the multinational firm is, the higher the probability is that the firm will have to adopt typical vertical integration (V1) and foreign concentration (V2). Hypothesis 2c. The higher the R&D and capital intensity the multinational firm has, the higher the probability is that the firm will have to adopt home concentration (V3).

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005

G Model RESPOL-3019; No. of Pages 11

ARTICLE IN PRESS H.-L. Lin et al. / Research Policy xxx (2014) xxx–xxx

5

Fig. 1. Three sets of proposed hypotheses.

Hypothesis 2d. The larger the export ratio the multinational firm has, the higher the probability is that the firm will have to adopt foreign concentration (V2) and typical vertical integration (V1). Hypothesis 2e. An increase in the foreign wage rate has a positive effect on the probability of undertaking home concentration (V3). We are also further interested in comparing pure horizontal concentration (H1) with heterogeneous horizontal integration (H2) because we may be able to see under what conditions the horizontal-type firms produce different quality products. A larger firm that has more resources might be more capable of producing a variety of final products in different countries. However, the effect of the foreign market size on choosing H1 vs. H2 is uncertain since we have no information on the market demand for high-end or low-end products in foreign countries. The high-tech industry in general consists of firms that have a specific asset, attribute or intellectual property right to produce some sorts of exclusive products. This type of firm is thereby considered to produce similar products rather than heterogeneous products in overseas subsidiaries, and so it is more likely for a firm in the high-tech industry to adopt H1 rather than H2. The following hypotheses will then be tested: Hypothesis Set III Hypothesis 3a. The firm size has a positive effect on the probability of performing heterogeneous horizontal integration (H2). Hypothesis 3b. A hi-tech firm has a higher probability of performing standard horizontal integration (H1). Our prediction of the effects from country-level, industry-level and firm-level variables on the production choice for the above three sets of hypotheses will be empirically tested in Section 6. 4. The data In this study we utilize a uniquely compiled firm-level data set of Taiwanese MNEs to empirically test the proposed hypotheses in Section 3.2. FDI activities engaged in by Taiwanese firms before the mid-1980s were few and sporadic. Owing to the enormous changes in the industrial environment, such as a new emphasis on environmental protection, increasing land and labor costs, and volatile fluctuations in the local Taiwanese currency, which emerged as major issues starting in the late 1980s, many labor-intensive firms began to move their production lines overseas. Specifically, to meet

the requirements of being a member of the World Trade Organization (WTO), Taiwan deregulated its regulatory policy enforced in the electronics industry regarding investment in China in 2001, which resulted in a substantial upsurge in FDI (see Yang et al., 2010). For instance, the annual amount of outward FDI increased from US$1.8 billion in 1992 to US$15.2 billion in 2008, culminating in an average growth rate of 19% over the period. The high growth rate reveals that Taiwanese firms are actively investing in foreign markets to meet the challenges of economic globalization. Currently, Taiwan is one of the most important investing countries in the world. In particular, the production strategies of Taiwanese MNEs are so flexible that firms may adopt various integrated production tactics as discussed above, thereby enabling us to test the proposed theoretical predictions. The data set utilized in this study is primarily drawn from the 2004–2007 Survey on the Outward FDI of Taiwanese Manufacturers conducted by the Ministry of Economic Affairs (MOEA) of Taiwan. In order to understand the motivation, business profile, and the difficulties confronted by overseas investments of Taiwanese manufacturers, the MOEA conducts this survey by sampling about 2500 firms every year, and the successful response rate is about 76%. Besides, the reason why we selected 2004–2007 as our study period lies in the consistency of the returned questionnaires during this period. The data set includes not only some firm characteristics, such as revenue, total employment, and R&D expenditure, but also variables related to overseas activities, for instance, investment areas and overseas R&D expenditure. More importantly, the information regarding the product relations between the parent firm and its foreign subsidiaries is also surveyed, which enables us to identify the production strategies of Taiwanese MNEs. Table 1 lists the overseas investment areas of Taiwanese multinational manufacturers during the sample period from 2004 to 2007. As indicated in Table 1, China is the most popular investment destination for Taiwanese multinationals, accounting for about 72% of their outward investment. This high concentration rate could be explained by the proximity of Taiwan to China, the intimate historical ties, and a common language. Of course, cheaper labor and the huge China market are also important factors attracting Taiwanese MNEs. Southeast Asia, which includes Malaysia, Singapore, Thailand, Indonesia, the Philippines and Vietnam, is the second (accounting for about 10%), and the United States is the third (accounting for about 8%) overseas investment destination of Taiwanese firms.

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005

G Model

ARTICLE IN PRESS

RESPOL-3019; No. of Pages 11

H.-L. Lin et al. / Research Policy xxx (2014) xxx–xxx

6

Table 2 2004–2007 product relation between parent (Taiwanese) firm and foreign subsidiaries. Question: What is the product relationship between the parent firm and your overseas subsidiary?

Classification

1. No relationship 2. Parent firm produces intermediate goods, while foreign subsidiaries produce or assemble the final product 3. Foreign subsidiaries produce intermediate goods, while parent firm produces or assembles the final product 4. Foreign subsidiaries are responsible for production, while parent firm is responsible for marketing 5. The product of the parent firm has higher value added than that of foreign subsidiaries 6. The product of foreign subsidiaries has higher value added than that of the parent firm 7. The product of the parent firm is similar to that of foreign subsidiaries 8. The product of the parent firm is exactly the same as that of foreign subsidiaries 9. Both produce with similar quality but the parent firm produces a variety of product types 10. Both produce with similar quality but foreign subsidiaries produce a variety of product types 11. Others

n.a. V2

Table 3 2004–2007 five types of FDI production strategies of Taiwanese MNEs. Production strategies

Number of firms

V1: vertical integration V2: foreign concentration V3: home concentration H1: horizontal integration H2: heterogeneous horizontal integration

405 878 567 655 2251

(Proportion) 8.5% 18.5% 11.9% 13.8% 47.3%

Total

4756

100.0%

Data Source: Outward FDI Survey in Manufacturing, 2004–2007, MOEA, Taiwan. V3

V1

H2

H2

H1

H1

H2

H2

n.a.

Data Source: Outward FDI Survey in Manufacturing, 2004–2007, MOEA, Taiwan.

The eleven product relations between the parent firm and its overseas subsidiary were surveyed for each firm in the questionnaire, as depicted in Table 2. We further classified these eleven production relations into the five production strategies (including standard and non-standard vertical strategies as well as horizontal strategies) that we proposed in Section 3. More specifically, if a firm answered “the product of the parent firm is exactly the same as that of foreign subsidiaries”, then its production strategy is referred to as standard horizontal integration (H1). When a firm responded with “the product of the parent firm has higher value added/quality than that of foreign subsidiaries”, then it is categorized as heterogeneous horizontal integration (H2). The reply that “the parent (foreign subsidiaries) firm produces intermediate goods, while the foreign subsidiaries (parent) produce or assemble the final product”, is referred to as standard vertical specialization (V1). “Foreign subsidiaries are responsible for production, while the parent firm is responsible for marketing” is regarded as foreign concentration (V2). For those firms answering “the parent firm is responsible for production, while the foreign subsidiaries are responsible for marketing”, the corresponding production strategy is classified as home concentration (V3). In addition, we further deleted observations with answers of “no relationship” and “others”, incomplete information, or extreme values (i.e., using the 0.5 and the 99.5 percentiles as the lower and upper thresholds) with regard to the explanatory variables in our empirical model, thus yielding a final sample of 4756 firms. It is noted that in the 2004 questionnaire

a firm could only choose one among the eleven product relations in Table 2. We thus have a clear-cut classification for each firm’s production strategy. However, multiple choices were allowed in the 2005–2007 surveys, where it is possible to have an overlap between H (H1 and H2) and V (V1, V2 and V3). In order to facilitate our econometric estimation, we thus dropped the sample firms that answered with multiple production strategies (around 6.3% of the original data), i.e., only those firms which adopted exactly one FDI strategy were retained (around 93.7% of the original data).8 In other words, the firms associated with the five different production strategies are mutually exclusive in our sample. As Table 3 indicates, if we roughly separate FDI activities into vertical integration (i.e., V1 + V2 + V3) and horizontal (i.e., H1 + H2) integration, then 1859 and 2906 firms are classified into each pattern, respectively. However, the more precise classification indicates that the majority of Taiwanese MNEs engage in heterogeneous horizontal integration (H2), accounting for about 47%. The second most prevalent form of overseas production strategy is foreign concentration (V2), which accounts for 18.5% of the sample firms. On the other hand, the production strategy of (pure) vertical integration (V1) is less adopted by Taiwanese MNEs. As the motivation and investing destinations matter with various production strategies in FDI, we also report the related summary statistics in Table 4. It can be seen that matching the clients’ need is the most important motivation across each production strategy.9 Additionally, the market motivation is highly associated with foreign concentration (V2), while the factor and technology seeking motivations are the most crucial factors for a home concentration firm (V3). As for the relationship between investment areas and production strategies, the statistics in the lower panel of Table 4 display that China is the most popular investment destination for Taiwanese MNEs among the five production strategies, except for home concentration (V3). As for investment in the U.S., which is the country with a higher wage level, Taiwanese MNEs tend to adopt the production strategy of home concentration (V3) by marketing activity abroad and by reallocating production activity domestically. By contrast, when firms invest in China and South Asia, heterogeneous horizontal integration (H2) has been adopted as the main production strategy. In

8 Specifically, if a firm checked #5 and #6 in Table 2, it is classified as an H2 firm. Similarly, a firm simultaneously chose #7 and #8, it is a member of the H1 firm. Nevertheless, we ignore a very small proportion of firms with mixed production strategies such as choosing #4 (V1) and #5 (H2) at the same time. 9 According to Amit and Schoemaker (1993), “the firm’s strategic assets as the set of difficult to trade and imitate, scare, appropriable and specialized resources and capabilities that bestow the firm’s competitive advantage.”(p. 36) Therefore, based on the above definition, one can see that the “strategic asset”, characterized by scarcity, immobility, dissimilation and specificity, is very heterogeneous and unique for each firm with different capability. We acknowledge that it is not an easy task to clearly disentangle each motivation for different type of FDI strategies in terms of our current questionnaire. Nevertheless, we come up with market, factor, client dependence (i.e., following in the footsteps of the domestic and overseas clients’ investment), efficiency, technology seeking, and tax exemption and local government support motivations.

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005

G Model

ARTICLE IN PRESS

RESPOL-3019; No. of Pages 11

H.-L. Lin et al. / Research Policy xxx (2014) xxx–xxx

7

Table 4 2004–2007 motivation and destination for different types of FDI production strategies of Taiwanese MNEs. Production strategies

V1

V2

V3

H1

Motivationa Market motivation Factor motivationb Client dependence Efficiencyc Technology seeking Tax exemption & local gov’t. support

37 (9.1%) 96 (23.7%) 180 (44.4%) 35 (8.6%) 12 (3.0%) 45 (11.1%)

164 (28.9%) 56 (9.9%) 237 (41.8%) 35 (6.2%) 40 (7.1%) 35 (6.2%)

57 (6.5%) 250 (28.5%) 369 (42.0%) 28 (3.2%) 15 (1.7%) 159 (18.1%)

106 (16.2%) 104 (15.9%) 310 (47.3%) 31 (4.7%) 29 (4.4%) 75 (11.5%)

238 (10.6%) 420 (18.7%) 984 (43.7%) 192 (8.5%) 31 (1.4%) 386 (17.1%)

8 (2.0%) 337 (83.2%) 36 (8.9%)

29 (3.3%) 725 (82.6%) 86 (9.8%)

244 (43.0%) 179 (31.6%) 33 (5.8%)

49 (7.5%) 462 (70.5%) 78 (11.9%)

59 (2.6%) 1895 (84.2%) 228 (10.1%)

Destination US China Southeast Asiad

H2

Data Source: Outward FDI Survey in Manufacturing, 2004–2007, MOEA, Taiwan. V1: vertical specialization, V2: foreign concentration, V3: home concentration, H1: horizontal integration, and H2: heterogeneous horizontal integration. a The question for investment motivation is a multiple response question. b Factor motivation includes being able to access the abundant labor force, land resources, or raw materials. c Efficiency denotes using capital efficiently. d Southeast Asia includes Malaysia, Thailand, Indonesia, the Philippines, Vietnam and Singapore.

what follows, we conduct an empirical investigation to see how the country-level, industry-level and firm-level variables influence the decisions regarding the different types of FDI production strategies.

where 1[sij = j] is an indicator function, which is equal to one if the firm i adopts strategy sij and zero otherwise. 5.2. Variables’ description

5. Econometric model and variables’ description 5.1. Multinomial logit model As discussed in the previous sections, the decision regarding production strategies, including vertical integration, foreign concentration, home concentration, horizontal integration, and heterogeneous horizontal integration, is a multinomial choice. This study therefore employs the multinomial logit model to proceed with the empirical analysis and to test the proposed hypotheses in Section 3.2. If we define the set of different types of FDI production strategies as S = {s1 , s2 , . . ., sJ }, then the probability distribution that a multinational firm engages in a type j production strategy is as follows: Pr(sij |Xi ) Pr(si1 |Xi )

= =

exp(Xi ˇj )

1+ 1+

J

exp(Xi ˇh ) 1

,

if j = 2, . . ., J

,

if j = 1,

h=2

J

h=2

exp(Xi ˇh )

where sij denotes that firm i chooses the jth production strategy, and Xi stands for a vector of firm-level, industry-level and countrylevel characteristics. Given the above probability distribution, one can estimate the model by using maximum likelihood estimation (MLE) as follows10 : ˆ = arg max ˇ

J N  

1[sij = j] log[Pr(sij |Xi )],

i=1 j=1

10

Since there are too many parameters to be estimated in the model and not all of them can be identified, the usual identification condition is imposed by setting ˇ1 = 0, that is, the coefficient vector of the reference production strategy is set to bzero. In our empirical application, we estimate a series of multinomial logit models with different reference groups in each regression run to facilitate the comparison of production strategy pairs. For instance, treating H1 (V1) as the reference strategy in estimating our five-choice logit model enables us to explore the factors underlying our picking V1 (V2) relative to H1 (V1).

The dependent variable is the five forms of production strategies between the parent firm of a Taiwanese MNE and its foreign subsidiaries. In regard to the explanatory variables included in our analysis, we consider country-, industry- and firm-specific variables according to the discussion in Section 3 and the existing literature. 5.2.1. Host-country characteristics As emphasized in the research on the determinants of FDI, e.g., Brainard (1997), Helpman and Krugman (1985), Markusen and Maskus (2002), and Shatz and Venables (2000), the aggregate characteristics of host countries are among the major influences. In this study, three country-level characteristics are included. The GDP per capita of the host country enters the empirical equation in the logarithmic form (GDP) that is to serve as a proxy of the foreign market size. The Wage and DTF denote the host country’s wage rate and the degree of trade freedom (measuring the trade cost in the host country), respectively.11 5.2.2. Industry-level variables The first industry-level variable we consider is the industrial capital intensity (KInt) which is computed by the ratio of fixed capital to employees, since it is the most important variable used to describe the level of fixed costs endured by a firm (Buch et al., 2005). Using the Industry, Commerce and Service Census which is conducted by the Directorate-General of Budget, Accounting and Statistics (DGBAS) in Taiwan, we obtained the capital intensity for each four-digit industry.

11 The information on GDP per capita for each country was obtained from the International Financial Statistics, IMF, wage information comes from Euromontitor trade source/national statistics, and the Trade Freedom Index was collected from http://www.heritage.org/Index/. Note that we adopt GDP per capita rather than absolute GDP to better capture the foreign market size in consideration of purchasing power. In studying the potential determinants of FDI, Chakrabarti (2001) argued that absolute GDP may be a poor indicator since it reflects the size of the population rather than the income per capita. In addition, the larger the Trade Freedom Index is, the smaller the trade cost.

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005

G Model

ARTICLE IN PRESS

RESPOL-3019; No. of Pages 11

H.-L. Lin et al. / Research Policy xxx (2014) xxx–xxx

8

Table 5 Mean and Standard deviation of variables included in the regression analysis. Variables

V1

V2

V3

H1

H2

EMP

5.5623 (1.5466) 0.0269 (0.0486) 1.5896 (0.7429) 0.1834 (0.3241) 7.5963 (0.6404) 0.3457 (0.4762) 0.0815 (0.2739) 4209.9 (8246.2) 0.2956 (0.4749) 4.1050 (0.1700)

5.7885 (1.6567) 0.0221 (0.0408) 1.9196 (0.6383) 0.3102 (0.4146) 7.4088 (0.4733) 0.3884 (0.4877) 0.1663 (0.3726) 4652.4 (9302.8) 0.3239 (0.5281) 0.3239 (0.1563)

5.3299 (1.2694) 0.0624 (0.0750) 1.4936 (0.7147) 0.0647 (0.2069) 7.8933 (0.7949) 0.6649 (0.4724) 0.0406 (0.1975) 25367.3 (18115) 1.5920 (1.1738) 4.3138 (0.1757)

5.3207 (1.4377) 0.0224 (0.0506) 1.6148 (0.7498) 0.1334 (0.2833) 7.6628 (0.6581) 0.3404 (0.4742) 0.1252 (0.3312) 8112.8 (13072) 0.5102 (0.7703) 4.1660 (0.1862)

5.6589 (1.4348) 0.0196 (0.0390) 1.6806 (0.7076) 0.1613 (0.3074) 7.6385 (0.6170) 0.2128 (0.4094) 0.1653 (0.3715) 3962.8 (8064.9) 0.2895 (0.4800) 4.1164 (0.1576)

RDRatio InvYrs ExRatio KInt TechInd TradInd GDP Wage DTF

Notes (1) Standard deviations are in parentheses. (2) V1: vertical specialization, V2: foreign concentration, V3: home concentration, H1: horizontal integration, and H2: heterogeneous horizontal integration. (3) EMP: logarithm of the number of employees, RDRatio: ratio of R&D expenditures to sales, InvYrs: logarithm of the years of overseas investment, ExRatio: the export ratio of foreign affiliates to sales, KInt: logarithm of the (4-digit) industrial capital intensity, TechInd: high-tech industry dummy variable, TradInd: traditional industry dummy variable, GDP: logarithm of per capita GDP in the host country, Wage: wage rate in the host country, and DTF: logarithm of the degree of trade freedom.

Furthermore, we included two industrial dummy variables to capture the other possible industrial effects. TechInd denotes the high-tech industry dummy which equals one if a firm belongs to the computer communication or computer components industry, otherwise it is equal to zero. TradInd represents a traditional industry dummy that equals one if a firm belongs to the food and beverages industry or the textiles and clothing industry, and zero otherwise. 5.2.3. Firm-level variables Firm heterogeneity is widely recognized as one of the important dimensions affecting FDI decisions in previous studies. Thus, two firm-specific characteristics, namely, firm scale and R&D intensity, are included. Firm scale is measured by the logarithm of the number of employees (EMP) and R&D intensity is measured by the ratio of R&D expenditure to sales (RDRatio).12 Since experience in conducting foreign investment is an important factor for firms’ production decision, we also incorporate the length of overseas investment (InvYrs) into the empirical specification. Finally, the export ratio of foreign affiliates (ExRatio) is also included to capture the potential effect of triangular trade on the FDI decision. The summary statistics of the variables for various forms of production strategies are displayed in Table 5.13 One can see that firms engaging in home concentration (V3) tend to be associated with higher R&D intensity, less experience in overseas investment, smaller firm size, more high-tech industry, less traditional industry, as well as lower trade costs, and higher per capita GDP and wage rates in the host countries. It is very intuitive to expect such

12 In the literature, firm productivity has been widely applied to represent the heterogeneity of the firm, such as in Kimura and Kiyota (2006) and Castellani and Zanfei (2007). Due to the data availability regarding the intermediate inputs, we are not able to calculate and include firm productivity. 13 Additional summary statistics in regard to the distribution of industry and firm size can be found in Appendix Tables A.1 and A.2. The Pearson correlation matrix for the dependent and independent variables used in the multinomial logit regression model is reported in Appendix Table A.3.

Table 6 Multinomial Logit Regression Results for Various FDI Production Strategies Variables Reference group EMP RDRatio InvYrs ExRatio KInt TechInd TradInd GDP Wage DTF Constant

(1) V1–H1 H1

(2) V1–V2 V2

(3) V1–V3 V3

(4) V2–V3 V3

(5) H1–H2 H2

0.121*** (0.045) 2.753* (1.457) −0.031 (0.091) 0.386** (0.220) −0.197* (0.108) 0.006 (0.149) −0.631*** (0.227) −0.486** (0.225) 0.111 (0.269) −0.033 (0.696) 4.266* (2.214)

−0.040 (0.043) 1.915 (1.378) −0.607*** (0.093) −1.005*** (0.193) 0.675*** (0.123) −0.639*** (0.143) −0.458** (0.218) 0.295 (0.215) 0.012 (0.270) −3.179*** (0.648) 6.770*** (2.192)

0.162*** (0.053) −3.898*** (1.281) 0.234** (0.104) 1.128*** (0.298) −0.377*** (0.115) −0.588*** (0.165) −0.517* (0.304) −0.613** (0.248) −0.560** (0.267) −0.268 (0.879) 8.408*** (2.700)

0.202*** (0.047) −5.812*** (1.217) 0.841*** (0.097) 2.133*** (0.269) −1.052*** (0.116) 0.050 (0.147) −0.059 (0.264) −0.908*** (0.198) −0.572*** (0.197) 2.911*** (0.796) 1.639 (1.639)

−0.185*** (0.032) −0.456 (1.189) −0.053 (0.065) −0.160 (0.165) −0.040 (0.073) 0.611*** (0.110) −0.010 (0.140) 0.420*** (0.148) −0.023 (0.170) −0.270 (0.497) −2.137 (1.550)

Number of Obs. Log-likelihood Chi-squared

4756 −5746.201 −1846.190

Notes: (1) V1: vertical specialization, V2: foreign concentration, V3: home concentration, H1: horizontal integration, and H2: heterogeneous horizontal integration. (2) There are five pairs of production strategies in the first row, where for each pair the latter one represents the base strategy to be compared with, e.g., in the case of the V1 vs. H1 pair (V1–H1), H1 is the reference strategy. (3) Standard errors are in parentheses. * Statistical significance at the 10% levels. ** Statistical significance at the 5% levels. *** Statistical significance at the 1% levels.

firms to adopt V3 under those circumstances. We also observe that firms focusing on foreign concentration (V2) are featured as having a larger firm size, less R&D intensity, greater firm age in overseas investment, more traditional industries and higher trade costs in the host country. Moreover, those firms engaging in heterogeneously horizontal integration, lower R&D intensity, belonging to less high-tech but more traditional industries, as well as having lower per capita GDP and lower wage rates in the host countries, are exhibited. 6. Empirical results Table 6 provides the estimation results using the multinomial logit model,14 where columns (2)–(4) present the comparative estimates among the three vertical-type FDI production strategies, column (1) displays the comparison between the typical vertical integration and horizontal integration, and column (5) shows the comparison between two horizontal FDI strategies. First of all, for the comparison between V1 and H1, we find that the foreign market size (GDP) negatively affects the probability of conducting vertical integration (V1), suggesting the validity of Hypothesis 1a. It indicates that a larger foreign market is likely to increase the propensity of MNEs to undertake the production strategy of H1. For the impacts of firm-specific variables on the choice between

14 The Hausman tests are performed to examine whether the independence of irrelevant alternatives (IIA) is satisfied. According to the p-values of the test statistics (not shown in the table), we conclude that the IIA assumption is confirmed so that the use of the multinomial logit model can be justified.

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005

G Model RESPOL-3019; No. of Pages 11

ARTICLE IN PRESS H.-L. Lin et al. / Research Policy xxx (2014) xxx–xxx

V1 and H1, firm size (EMP), R&D intensity (RDRatio), and export ratio (ExRatio) have significantly positive effects on the probability of performing V1, confirming Hypotheses 1a–1d. However, the length of overseas investment (InvYrs) exhibits a negative but insignificant effect on the propensity to choose V1. These empirical findings are along the lines of the summary statistics in Table 5, e.g., a larger per capita GDP (overseas market size) is associated with H1, while a larger firm size (EMP) and higher R&D intensity (RDRatio) are in accordance with V1. Moreover, the foreign trade freedom index (DTF) has no significant impact on the probability of performing V1. The insignificant coefficient on the foreign trade freedom index reveals that the choice between V1 and H1 may not depend on the magnitude of costs associated with foreign trade. Next, we turn to discuss the empirical results among the three vertical FDI strategies in columns (2)–(4) of Table 6, i.e., vertical integration (V1) vs. foreign concentration (V2), vertical integration (V1) vs. home concentration (V3), and foreign concentration (V2) vs. home concentration (V3). Regarding the firm-level variables, it is found that the past experience of overseas investment (InvYrs) has a negative impact on the choice of concentrating production at home (V3) compared to the alternative production strategies with V1 or V2, indicating that Hypothesis 2a is supported. It means that years of overseas experience help the formation of multinational activity V2 but do not favor domestic-oriented vertical integration, V3. However, InvYrs positively affects the probability of choosing V2 when compared with V1. In regard to the other effects of firm-level variables, both firm size (EMP) and R&D intensity (RDRatio) have no significant impacts on the choice between V1 and V2. By contrast, in the comparison between V1 vs. V3, or V2 vs. V3, firm size (EMP) exhibits a negative impact on the probability of undertaking V3 and R&D intensity (RDRatio) exerts a positive impact on the choice of V3. Hypotheses 2b and 2c are thus confirmed. These findings imply that firms with relatively smaller size but higher R&D intensity tend to stay in Taiwan to engage in the home-concentration type of vertical integration. Table 6 also reveals that a higher export ratio (ExRatio) is associated with the choice of V1 and V2 (especially V2, see column (2)), which is suggested by Hypothesis 2d. We also find that, compared to either V2 or V1, the country-level variable – foreign wage rate (Wage) positively affects the probability of performing V3, which supports Hypothesis 2e. It is not surprising to observe that a firm would choose a typical vertical integration or foreign concentration (relative to home concentration) when facing a lower wage in the host country and when experiencing more in foreign investment. In addition, columns (3) and (4) in Table 6 also show that a larger GDP is positively associated with performing V3. The reasoning is straightforward in that as the major source of revenue for strategy V1 or V2 is the home market, hence, with the increased foreign market size, MNEs have the incentive to undertake the production strategy of V3. Compared to V1 or V2, while the trade freedom index (DTF, an opposite measurement for trade costs) is expected to have a positive effect on the probability of performing V3, it actually exerts a negative impact on the choice of V3 compared with V2 and has no influence on the choice between V1 and V3. The reason for this could be that DTF does not capture the foreign trade costs particularly well. Turning to the comparison between horizontal integration (H1) and heterogeneous horizontal integration (H2), we find that few variables have significant impacts on a specific production strategy. Hypothesis 3a demonstrates that firm size should have a positive impact on the choice of strategy H2. Our estimates in column (5) of Table 6 display that firm size (EMP) has a positive and significant impact on the probability of engaging in H2, thus supporting Hypothesis 3a. This implies that firms of a larger size are more capable of enduring the additional fixed costs brought on by engaging

9

in product differentiation between the foreign and home markets. The significant result for foreign market size (GDP) indicates that the aggregate variable has a crucial impact on the choice between H1 and H2. The reason might be due to the fact that both strategies obtain similar revenue from the foreign market, implying that country-level variables have no significant impacts on the choice between these two strategies. In reality, firms engaging in product differentiation might attach different weights to different markets. Hence, the revenue from overseas markets might be different even though both strategies involve selling and producing overseas. In this way, foreign aggregate variables may be influential in determining the choice of H1 or H2. Lastly, high-tech firms tend to have a higher probability of choosing H1 than H2, confirming Hypothesis 3b. Several policy implications can be drawn from our empirical findings. In terms of the firm’s policy, we conclude that larger, older, and export-oriented firms are more capable of moving their production process to foreign countries. By producing/assembling final products overseas and marketing/distributing products in the home country, FDI firms with our specified features can enhance production efficiency and/or reduce production cost to optimize their profit. In addition, this study indicates that a firm engaging in more R&D, and having higher capital intensity and wage levels, tends to adopt the home concentration FDI strategy in comparison to the foreign concentration production strategy. In other words, the issue of “deindustrialization” in Taiwan over the past few years can be mitigated through government-supported policies. For instance, the Taiwan’s government could build a friendly environment such as implementing a preferential policy to encourage innovation activities and help upgrade firms’ technology, which would in turn attract Taiwanese manufacturing firms to stay at home. A final remark is that fewer than a quarter of Taiwanese MNEs have overseas operations in countries other than China and Hong Kong, and this strong intra-regional orientation may suggest that the majority of Taiwanese MNEs are not yet truly multinational. According to our summary statistics, Taiwanese MNEs also tend to show heavy sectoral concentrations, leaning toward electronics manufacturing, as we defined high-tech firms in the Taiwanese setting as those firms in the computer industry. In other words, there is a possibility that our findings from the Taiwan case may not be generalizable to a substantial degree.15 7. Conclusion Most of the existing literature on FDI divides multinational activities roughly into two broad forms of standard vertical integration and horizontal integration, while the production strategy of MNEs is more complex in practice. This study further classifies the production strategies of MNEs into five types (i.e., two standard and three non-standard FDI production strategies), which are standard vertical integration (V1), foreign concentration (V2), home concentration (V3), standard horizontal integration (H1), and heterogeneous horizontal integration (H2), in order to obtain more insightful implications regarding the decisions based on the production strategies of MNEs in FDI. We then propose hypotheses regarding the role of country-level, industry-level and firm-level variables in affecting the choices of various FDI strategy pairs. Based on a firm-level data set of Taiwanese MNEs during the 2004–2007 period, the three sets of proposed hypotheses are in

15

We owe this cautionary note to an anonymous referee.

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005

G Model RESPOL-3019; No. of Pages 11 10

ARTICLE IN PRESS H.-L. Lin et al. / Research Policy xxx (2014) xxx–xxx

general supported. The main findings may be summarized as follows. First, comparing the typical vertical integration (V1) with horizontal integration (H1), we find that foreign market size has a positive effect on the probability of adopting the FDI strategy H1. This finding is consistent with most of the existing literature. In addition, we also find that the length of the overseas investment positively affects the probability of engaging in strategy H1 because “localization” enables foreign subsidiaries to be independent of their parent firms and to then sell and produce overseas. Firm size, R&D intensity, and the export ratio are confirmed to have significantly positive effects on the probability of performing V1. Our empirical results, however, show that the length of overseas investment has a negative but insignificant effect on the probability of adopting strategy H1. Second, we conduct a comparison among three vertical multinational strategy pairs, i.e., vertical integration (V1) vs. foreign concentration (V2), vertical integration (V1) vs. home concentration (V3), and foreign concentration (V2) vs. home concentration (V3). The years of overseas experience in investment were found to have a negative impact on the probability of engaging in V3 compared to either V1 or V2, and to have a positive impact on the probability of undertaking V2 compared with V3. This implies that years of overseas experience help the formation of multinational activity V2. In addition, firm size exhibits a negative impact on the probability of undertaking V3, while R&D intensity exerts a positive impact on the choice of V3. It is intuitive to confirm that the foreign wage rate positively affects the probability of engaging in V3 compared to either V1 or V2. Finally, this study investigates the choice between horizontal integration (H1) and heterogeneous horizontal integration (H2). Firm size is important when there are economies of scale by engaging in product differentiation (Cooper and Kaplan, 1991). We argue that a firm of larger size will have a higher probability of adopting FDI strategy H2. The empirical results basically support that there is a positive effect of firm size on the choice of H2. We also observe that high-tech firms tend to have a higher probability of choosing H1 than H2, confirming that firms with a specific asset, attribute or intellectual property right tend to produce some sorts of exclusive products. However, foreign market size was found to have a negative impact on the probability of choosing H2. Since both strategies involve selling and producing overseas, the foreign market size and wage rate may have no crucial impact on the decision between H1 and H2 in the case of a multinational firm. Note that the classified vertical production strategies, V1 (typical vertical integration), V2 (foreign concentration) and V3 (home concentration), implicitly indicate different stages of product production and location choices. The product production stages are dynamic in consideration of three stages: new product, maturing product and standardized product (Vernon, 1966, 1979). During the new product stage, the production activities are concentrated in the home country (i.e., V3, home concentration), since at this stage the producer needs more flexibility to handle the considerable amount of uncertainty in new product production. Production

at home can offer the producer higher flexibility. Such a decision is beyond simple factor cost analysis plus transportation cost. In the maturing product stage, the factor cost is taken into account. To efficiently utilize and allocate resources, the producer will move the stage with less comparative advantage in the production process (e.g., the labor intensive activity) to other countries (i.e., V1, vertical integration). As the product is standardized, the producer might move all of the production activities out of the home country (i.e., V2, foreign concentration). As far as the governmental authority of the home country is concerned, the least desirable scenario might be foreign concentrated production, since it implies the “hollowing out” of the industry or “deindustrialization”. To keep core and value-added production activities in the home country, instead of prohibiting or regulating outward capital movement, our study suggests that the governmental authority should encourage domestic firms to dedicate more resources to innovation activities, which can help them level up technology, differentiate products and hence maintain their competitive advantages. Acknowledgements We wish to acknowledge the very useful and constructive comments from two anonymous referees and participants of the seminar in the Department of Economics at National Taiwan University. Financial support from the National Science Council in the form of grants NSC 100-2410-H-002-067-MY2 (H-L Lin) and NSC 100-2410-H-007-025-MY2 (Eric Lin) is gratefully acknowledged. The usual disclaimer applies. Appendix.

Table A.1 Distribution of firm size. Firm size

Number of firms

Small Medium Large

1226 870 2660

Proportion 25.78% 18.29% 55.93%

Total

4756

100.00%

Cum. Prop. 25.78% 44.07% 100.00% –

Data Source: Outward FDI Survey in Manufacturing, 2004–2007, MOEA, Taiwan.

Table A.2 Distribution of industries. Industry

Number of firms

Mental Information Tech. Chemical Consumer goods

1094 2089 853 720

Proportion 23.00% 43.92% 17.94% 15.14%

Total

4756

100.00%

Cum. Prop. 23.00% 66.93% 84.86% 100.00% –

Data Source: Outward FDI Survey in Manufacturing, 2004–2007, MOEA, Taiwan

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005

G Model

ARTICLE IN PRESS

RESPOL-3019; No. of Pages 11

H.-L. Lin et al. / Research Policy xxx (2014) xxx–xxx

11

Table A.3 Correlation matrix of variables.

V1 V2 V3 H1 H2 EMP RD InvY Ex KI Tech Trand GDP Wage DTF

V1

V2

V3

H1

H2

EMP

RD

InvY

Ex

KI

Tech

Trand

GDP

Wage

DTF

1.000 −0.097* −0.125* −0.106* −0.240* −0.006 0.009 −0.045 0.004 −0.007 0.017 −0.046 −0.067* −0.064* −0.069*

1.000 −0.151* −0.127* −0.289* −0.061* 0.261* −0.101* −0.119* 0.152* 0.255* −0.096* 0.484* 0.514* 0.327*

1.000 −0.163* −0.372* 0.059* −0.030 0.134* 0.176* −0.142* 0.067* 0.037 −0.075* −0.082* −0.017

1.000 −0.314* −0.068* −0.022 −0.046 −0.052 0.030 0.018 −0.013 0.047 0.023 0.039

1.000 0.039 −0.101* −0.026 −0.045 0.035 −0.190* 0.069* −0.195* −0.195* −0.151*

1.000 −0.036 0.127* 0.083* 0.077* 0.144* 0.011 −0.034 −0.016 −0.025

1.000 −0.091* −0.052 0.091* 0.245* −0.129* 0.222* 0.240* 0.120*

1.000 0.155* −0.118* −0.145* 0.153* −0.016 −0.019 0.009

1.000 −0.124* −0.031 0.127* −0.154* −0.109* −0.304*

1.000 0.235* −0.162* 0.143* 0.125* 0.088*

1.000 −0.272* 0.279* 0.273* 0.169*

1.000 −0.142* −0.113* −0.069*

1.000 0.915* 0.779*

1.000 0.613*

1.000

Notes: EMP: logarithm of the number of employees, RD: ratio of R&D expenditures to sales, InvY: logarithm of the years of overseas investment, Ex: export ratio of foreign affiliates to sales, KI: logarithm of the (4-digit) industrial capital intensity, Tech: high-tech industry dummy variable, Trad: traditional industry dummy variable, GDP: logarithm of per capita GDP in the host country, Wage: wage rate in the host country, and DTF: logarithm of the degree of trade freedom. * Correlation coefficients significant at the 1% level or better.

References Aggarwal, R., Berrill, J., Hutson, E., Kearney, C., 2011. What is a multinational corporation? Classifying the degree of firm-level multinationality. International Business Review 20 (5), 557–577. Amit, R., Schoemaker, P.J.H., 1993. Strategic Assets and organizational rent. Strategic Management Journal 14, 33–46. Antrás, P., Helpman, E., 2003. Global Sourcing. Harvard Institute of Economic Research, Discussion Paper, No. 2005. Avenel, E., Barlet, C., 2000. Vertical foreclosure, technological choice, and entry on the intermediate market. Journal of Economics & Management Strategy 9, 211–230. Bandinger, H., Egger, P., 2008. Horizontal versus Vertical Interdependence in Multinational Activity, CESifo Working Paper No. 2327. Barba-Navaretti, G., Venables, A.J., 2004. Multinational Firms in the World Economy. Princeton University Press, Princeton. Brainard, L.S., 1997. An empirical assessment of the proximity-concentration tradeoff between multinational sales and trade. American Economic Review 87, 520–544. Buch, C.M., Kleinert, J., Lipponer, A., Toubal, F., 2005. Determinants and effects of foreign direct investment: evidence from German firm-level data. Economic Policy 20, 51–110. Carr, D.L., Markusen, J.R., Maskus, K.E., 2001. Estimating the knowledge-capital model of the multinational enterprise. American Economic Review 91, 693–708. Castellani, D., Zanfei, A., 2007. International, Innovation and productivity: how do firms differ in Italy? World Economy 30, 156–176. Chakrabarti, A., 2001. The determinants of foreign direct investment: sensitivity analyses of cross-country regressions. Kyklos 54 (1), 89–114. Cooper, R., Kaplan, R., 1991. The Design of Cost Management Systems. Prentice-Hall. Ekholm, K., Forslid, R., Markusen, J., 2003. Export Platform Foreign Direct Investment, NBER Working Paper No. 9517. Feenstra, R.C., 1998. Integration of trade and disintegration of production in the global economy. Journal of Economic Perspectives 12, 31–50. Feinberg, S.E., Keane, M.P., 2003. Accounting for the Growth of MNC-Based Trade Using a Structural Model of U.S. MNCs. University of Maryland (manuscript). Grossman, G.M., Helpman, E., Szeidl, A., 2006. Optimal integration strategies for the multinational firm. Journal of International Economics 70, 216– 238. Head, K., 2007. Elements of Multinational Strategy. Springer, New York. Head, K., Ries, J., 2003. Heterogeneity and the FDI versus export decision of Japanese manufacturers. Journal of the Japanese and International Economies 17, 448–467. Helpman, E., 1984. A simple theory of international trade with multinational corporations. Journal of Political Economy 92, 451–471. Helpman, E., Krugman, P.R., 1985. Market Structure and Foreign Trade. MIT Press, Cambridge, MA.

Helpman, E., Melitz, M.J., Yeaple, S.R., 2004. Export versus FDI with heterogeneous firms. American Economic Review 94, 300–316. Hsu, C.M., Chen, P.J., 2002. The Innovation and Financing of Taiwanese SMEs, and Cooperation with China’s Industries? NPF Research Report, No. 096-023 (in Chinese). Kao, H.C., Lin, T.C., 1993. The Impact of Taiwanese Outward FDI to China on the Domestic Industries Upgrade and the Cross-Strait Division of Labor, Research Report. Industrial Development Bureau, MOEA, Taiwan (in Chinese). Kao, C., Tsai, Y.F., 2006. The trend of bilateral trade, investment and industrial division between Taiwan and China. Economic Circumstances & Comments 12, 75–96 (in Chinese). Kimura, F., Kiyota, K., 2006. Exports, FDI, and productivity: dynamic evidence from japanese firms. Review of World Economics 124, 659–719. Krugman, P., 1996. Does third world growth hurt world prosperity? Harvard Business Review 72, 113–121. Leamer, E.E., 1996. In Search of Stolper-Samuelson Effects on U.S. Wages, NBER Working Paper No. 5036. Lipsey, R.E., 2003. Foreign direct investment and the operations of multinational firms: concepts, history and data. In: Choi, E.K., Harrigan, J. (Eds.), Handbook of International Trade. Blackwell Publishing, Malden. Markusen, J.R., 1984. Multinationals, multi-plant economics and the gains from trade. Journal of International Economics 16, 205–226. Markusen, J.R., Maskus, K.E., 2001. General-Equilibrium Approaches to the Multinational Firm: A Review of Theory and Evidence, NBER Working Paper No. 8334. Markusen, J.R., Maskus, K.E., 2002. Discriminating among alternative theories of the multinational enterprise. Review of International Economics 10 (4), 694–707. Markusen, J.R., Venables, A.J., 1998. Multinational firms and the new trade theory. Journal of International Economics 46, 183–203. Markusen, J.R., Venables, A.J., 2000. The theory of endowment, intra-industry and multinational trade. Journal of International Economics 52, 209–234. Protsenko, A., 2003. Vertical and Horizontal Foreign Direct Investment in Transition Countries (unpublished doctoral dissertation). Shatz, H.J., Venables, A.J., 2000. The geography of international investment. In: Clark, G.L., Feldman, M., Gertler, M.S. (Eds.), The Oxford Handbook of Economic Geography. Oxford University Press, Oxford. Shi, Y., Gregory, M., 1998. International manufacturing networks – to develop global competitive capabilities. Journal of Operations Management 16, 195–214. Vernon, R., 1966. International investment and international trade in the product cycle. In: Dunning (Ed.), The Theory and Transnational Corporations. (chapter 2). Vernon, R., 1979. The product cycle hypothesis in a new international environment. Oxford Bulletin of Economics and Statistics 41 (4), 255–267. Yang, C.H., Wu, Y.Y., Lin, H.L., 2010. Investment to China and local innovation of Taiwanese manufacturing firms. Japanese Economic Review 61, 538–557. Yeaple, S.R., 2003. The complex integration strategies of multinationals and cross country dependencies in the structure of foreign direct investment. Journal of International Economics 60, 293–314.

Please cite this article in press as: Lin, H.-L., et al., The choice between standard and non-standard FDI production strategies for Taiwanese multinationals. Res. Policy (2014), http://dx.doi.org/10.1016/j.respol.2014.06.005