Chinese MNE Activity, Location
Choices, and Business Groups
The Effect of Activity Type and Business Group Affiliation
on Location Choice
Tessa Speek
Student ID: 10290559Date of Submission: 27 January 2017
MSc Business Administration: International Management Master Thesis
University of Amsterdam Supervisor: Dr. Vittoria Scalera
Statement of originality
This document is written by Student Tessa Speek who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
1| Abstract ... 4 2| Introduction ... 5 3| Literature Review ... 9 3.1 Location Choice and Global Cities ... 9 3.2 EMNE Debate ... 12 Emerging Market Context Puzzle ... 13 Internationalization Process Puzzle ... 14 3.3 Chinese Firm OFDI Factors ... 16 3.4 Business Groups ... 18 Benefits ... 19 Challenges & Costs ... 19 4 | Analytical Framework and Hypotheses ... 21 4.1 Global Cities and Knowledge Concentration ... 21 4.2 Technology and Knowledge Transmission ... 22 4.3 Business Group Affiliation Effects ... 23 5 | Research Methods ... 24 5.1 Sample ... 24 5.2 Dependent Variable ... 25 Global City ... 25 5.3 Independent Variable ... 26 Investment Activity ... 26 5.4 Moderator Variable ... 27 Business Group Affiliation ... 27 Control Variables ... 28 6 | Results ... 29 6.1 Data Cleaning Checking ... 29 6.2 Statistical Analyses ... 33 6.3 Additional Analyses ... 36 Additional Analysis on Moderating Effect ... 36 Division of Value-Adding Activities ... 37 7 | Discussion and Conclusion ... 41 7.1 Academic Implications ... 43 7.2 Managerial and Policy Implications ... 45 7.3 Limitations and Suggestions for Future Research ... 45 8 | Acknowledgements ... 47 9 | References ... 481| Abstract
Emerging market multinational enterprises (EMNEs) have increasingly been internationalizing successfully, contradicting numerous traditional international business (IB) theories. This study investigates the applicability of IB theories in the specific topic of location choice within the context of an emerging market, China. Creating a link between location choice, investment activity type, and the phenomenon of business groups will help clarify the factors significant in the internationalization processes of EMNEs. The conceptual model states that firms implementing high value-adding activities will be more attracted to locating in global cities than firm implementing low value-adding activities. Furthermore, affiliation to a business group will negatively moderate this relationship because of the added benefits business group bring to affiliates. Analyses of a large sample of Chinese multinational enterprise (MNEs) foreign direct investments using logistic regression models illustrate there is a higher likelihood of locating in global cities when the activities implemented are high value-adding. Furthermore, additional analyses showed a hierarchical structure within this group, where business activities are more likely than knowledge creating activities to locate in global cities. Moreover, while no support was found for the argument that business group affiliation has moderating affect, it does show have a direct effect, exposing that Chinese MNEs affiliated to business groups are more likely to locate in global cities. This study contributes and extends the current literature on location choice and global cities and gives some interesting insights into the factors relevant in the unexpectedly successful internationalization processes of emerging market MNEs.
Keywords: location choice; EMNEs; emerging market; global cities; internationalization process; value-adding activities
2| Introduction
“Third world multinationalism, only yesterday an apparent contradiction in terms, is now a serious force in the development process” (Heenan & Keegan, 1979, p. 109). Throughout the 1950s and 1960s, the idea that the environment of emerging countries could produce firms with enough competitive advantages to succeed in the rapidly globalizing economy was nonexistent (Wells, 1983). However, in the following decades several firms took major steps towards internationalization to become multinational enterprises. While these few emerging market multinational enterprises (EMNEs) were considered irregularities at the time, their appearance in the global economy only sparked more interest from academics and have only become more important. Heenan and Keegan, (1979), Wells (1983) as well as Lall et al. (1983) recognized already in the late 1970s and early 1980s that the quantity of small developing country firms making outward foreign direct investments (OFDI) was not irregular, but growing considerably. Furthermore, even in the face of liabilities of emergingness that these EMNEs experienced (Madhok & Keyhani, 2012), the pace of their internationalization was also increasing (Mathews, 2006).
This new phenomenon of successful EMNEs developing could never have been forecasted by the existing international business (IB) theory and frameworks (Lall et al., 1983). As most of these theories were developed in the context of western and advanced market MNEs, many scholars have begun to question and debate the legitimacy and relevance of these mainstream MNE theories in relation to new emerging market MNEs (Beerepoot & Roodheuvel, 2016; Chittoor, 2009; Lall et al., 1983; Mathews, 2002a, 2002b, 2006; Ramamurti & Singh, 2009; Wells, 1983). As a result of this debate, other theories and frameworks have been created in an attempt to explain the behaviors of EMNEs, such as the linkage, leverage, and learning model put forth by Mathews (2006). Studying how EMNEs internationalize is necessary to further explore this debate in IB theory and is valuable not only academically but also practically regarding the management of EMNEs.
While the whole internationalization process is interesting to study, one specific area that is noteworthy is the location choices that MNEs as well EMNEs make during this process. These decisions that all MNEs make are essential as they have profound impacts on the performance of investments and are costly to adjust (Bartik, 1985; Duanmu, 2012). As will be described in more detail later, traditional IB theories such as the Eclectic Paradigm (Dunning, 2000) and the Uppsala Model (Johanson & Vahlne, 1977) depict that strategic investment
location choices are determined by a multitude of factors including type of investment activity, motivation for investment, ownership advantages, location advantages, internalization capabilities, knowledge (or the lack thereof) of local practices, and characteristics of investment locations (i.e. geographic and cultural distance, and political atmosphere) (Dunning, 2000; Johanson & Vahlne, 1977). Depending on the strategy and goals of the MNE, they will value these factors differently and from there, take steps toward internationalization (Chung & Alcácer, 2002).
However, while these traditional factors have generally taken a country-level unit of analysis, they have often oversimplified the locational differences and features present within countries and the behaviors of MNEs. Topics with sub-national levels of analysis such as regional agglomeration, cluster formation, and the “global cities” phenomenon, which focus much more specifically on location choice, are becoming more important in the domain of IB theory (Goerzen, Asmussen, & Nielsen, 2013). As the world economy progressively globalizes, cities have emerged and have become more interconnected and thus the “global cities” phenomenon has become increasingly relevant. The theory essentially states that the nature of a global city is unique and different from regional clusters, industry clusters, and megacities, and therefore exhibits specific characteristics. Goerzen et al. (2013) describe these characteristics to be the global interconnectedness, cosmopolitanism, and abundance of advanced producer services available in global cities and depict that they have an important impact on the location choices of firms.
Although these studies have explored MNEs location choices, a debate has begun about the relevance and accuracy of these theories pertaining to the internationalization processes of EMNEs. Thus, investigating the behaviors of EMNEs location choices on a sub-national level of analysis of global cities would not only contribute to this “global city” field of study but as well uncover more insights into the EMNE internationalization process. However, because the contexts of EMNEs can vary extensively, it is impossible to generalize and assess all EMNEs as a homogenous group (Ramamurti & Singh, 2009), therefore assessing one emerging market is more feasible and will result in more comprehensible insights into EMNE behaviors.
China in particular has been recognized as a country with a unique development history. Since the 1990s, it has been one of the largest recipients of inward foreign direct investments (IFDI) however in the recent years, it has experienced a strong increase of OFDI (Duanmu, 2012; Kang & Jiang, 2012; Liu, Buck, & Shu, 2005; Ramasamy, Yeung, & Laforet, 2012). Considerable academic attention has been paid to China because continued high levels of OFDI
growth are expected, resulting in high prospects for Chinese multinational enterprises (De Beule & Duanmu, 2012).
Exploring emerging markets, and especially China, also allows for the possibility to investigate the phenomenon of business groups, which has been recognized as playing an important institutional role in emerging economies (Fisman & Khanna, 2004). These business groups are present in numerous emerging markets, such as the grupos in Latin America, business houses in India, chaebol in South Korea, keiretsu in Japan, and the qiye jituan in China (Khanna & Rivkin, 2001; Khanna & Yafeh, 2009). While there is no universal definition for what business groups are and they generally differ per country, there is consensus on one thing; that firms of a business group are legally independent, however are tied together in numerous formal and informal ways (Khanna, 2000). Generally, these groups consist of a diversified portfolio of businesses and often have the ability to share a common labour pools and finances (Chittoor, 2009; Fisman & Khanna, 2004). The existing literature on business groups is considerable and while it is theorized that business groups might have significant impacts on the internationalization processes of EMNEs, little empirical evidence exists to describe the extent of this impact (Chittoor, 2009). Since emerging economies have gained significant standing in the global economy, understanding factors which may influence their internationalization process is vital.
Business groups in China have been prevalent since the time of the economic reforms. Prior to the reforms, most Chinese firms were state-owned and controlled by administrative agencies (Keister, 1998, 2001). Firms relied heavily on the state however constantly experienced shortages of resources. The state having to control enormous numbers of firms resulted in many issues leading to imbalances in distribution of resources and unequal treatments of firms (Keister, 1998, 2001). Starting in 1978, the Chinese government introduced numerous economic reforms and started retracting its control to play only a shareholder role. Firms gained more control over decision making however also received less support from the government, which posed many new problems for firms that used to depend heavily on the state (Keister, 1998, 2001). In the face of competition from state, non-state, and foreign firms, the Chinese government encouraged the formation of business groups during the 1980s in order to give firms stability in these times of transition. The relationships within these business groups created a secure environment in which firms could benefits from greater access to inputs and resources as well as economies of scale (Keister, 1998, 2001). Essentially, the idea is that these business groups would create a context that would enable Chinese firms to flourish,
gradually leading to greater relations and trade ties, and eventually long term increases in firm performance. From what can be seen in today’s global economy, business groups such as Sinopec Group, China National Petroleum Corporation (CNPC), and China State Construction Engineering Corporation Limited (CSCEC) among others currently reside in the top 50 of the Fortune Global 500 companies of 2016, illustrating how far these groups have come and the status these business groups hold (“Global 500,” 2016). However, this would not have happened without successful internationalization strategies.
The present study attempts to combine the above-mentioned theories and phenomena and aims to investigate a factor important in the internationalization process of Chinese MNEs, namely type of investment activity, and the potential impacts of business group affiliation on this process. Thus, the following research questions are proposed: To what extent do different investment activities influence the propensity of Chinese MNEs to locate in ‘global cities’? and How do business groups moderate this relationship?
To test these research questions, a quantitative analysis will be performed based on an extensive dataset. The dataset is comprised of a collection of outward greenfield investments made by Chinese MNEs. Each investment in the dataset includes various details relating to the type of investment, the destination country and city, and whether the MNE is linked to a business group, among others controlling elements. Several logistic regression models will be run on this data to test the presented hypotheses. The results of this study will reveal insights of the location choices of Chinese MNEs which are relevant both academically and managerially. Academically, the results will extend the literature on general location choice factors, emerging market MNEs behaviors, and lastly further investigate the concept of global cities. Managerially, the results will clarify how manager and policy-makers can approach and deal the growing number of EMNEs.
The remainder of this study is organized as follows. Section 3 will provide a review on the literature and debate surrounding traditional internationalization theories regarding MNEs and EMNEs and further will evaluate the potential advantages and disadvantages of business groups affiliation. Section 4 describes the analytical framework and proposed hypotheses. Section 5 dives into the research methods and Section 6 illustrates the results of the analyses. Lastly, Section 7 will reveal some important discussion points and draw some final conclusions, including the implications and limitations of this study.
3| Literature Review
3.1 Location Choice and Global Cities
If the global economy were to be characterized by free trade, perfect goods and factor markets, there would be no need for MNEs as there would be no transportation costs, no differences in preferences, and no scale or scope economies (Rugman, 1980). However, as the economy is characterized by imperfects goods and factor markets, highly versatile preferences, and potential for economies of scale and scope, location choice has become an integral part of the internationalization process for MNEs. Multiple theories have been developed to explain the progression of MNE expansion to the global market. Understanding these traditional theories that until now have been used to explain internationalization processes is necessary in order to assess the applicability of these theories to emerging market MNEs.
Coase (1937) was the first to tackle the subject of MNE expansion and consequently developed the internalization theory that now is the foundation of a considerable amount of literature. The basis of this theory is that firms internalize activities as a reaction to externalities (e.g. tariffs, taxes, etc.), governmental regulations, as well as market failures (Rugman, 1980). Rugman (1980) further builds on Coase’s theory and depicts that the existence of MNEs and the phenomenon of FDI is grounded on two concepts: 1) that MNEs choose the location with the lowest cost for each activity they execute and 2) that MNEs grow by internalizing markets where the benefits offset the costs. Similarly, Dunning (2000) also extended Coase’s theory and developed what is known as the Eclectic Paradigm, where the concept of ‘ownership’ advantages are deemed necessary in addition to location advantages and internalization in order for MNEs to successfully internationalize. This eclectic paradigm has become the prevailing framework used to test the factors influencing FDI of firms (Dunning, 2000).
Contrasting this internalization theory is the Uppsala Internationalization Process Model developed by Johanson and Vahlne (1977). This Uppsala framework describes that firms internationalize through knowledge acquisition and a sequence of incremental decisions over time (Forsgren, 2002). This model describes that entering new contexts can create new opportunities, however it also creates problems and costs because there is usually of a lack of market information, high uncertainty, and liabilities of foreignness. Consequently, firms have to take slow and incremental steps in order to fully be able to incorporate and apply the learning they acquire to the complexities they face (Johanson & Vahlne, 1977). The knowledge that
MNEs acquire greatly influence their location choices and internationalization process. Consistent with this model, Rugman, Verbeke, and Nguyen (2011) illustrate that firms will initially expand to countries that are closer in distance (culture, economic, geographic, and administrative distance) in order to minimize costs. Only when the firm has gained sufficient knowledge to address and outweigh the costs of doing business abroad, will the firm move forward with its internationalization process and expand to more distant locations (Rugman et al., 2011).
These two theories come from considerably different origins. The internalization theory originates from transaction cost economics, where location decisions of MNE activities are a discrete process separate from other decisions, taking first and foremost cost into account. In contrast, the internationalization theory originates from organizational learning theory, where one location decision of MNE activities can very much influence other decisions and where knowledge acquisition and learning from previous experiences are critical. However, while these theories contrast in origin, there are numerous factors that can influence an MNEs process of internationalization, whether it is a discrete or learning process (Benito & Gripsrud, 21992).
The factors that influence location choice can be split into two categories, pulling factors and pushing factors. Pulling factors include the characteristics of the investment destination as well as the locational advantages present. Pushing factors originate from within the firm and include the type of investment activity, investment motivation, ownership advantages, internalization capabilities, and knowledge of local practices (Dunning, 2000; Johanson & Vahlne, 1977).
Johanson and Vahlne (1977) state that the differences in characteristics between host and home country can play a significant role in the decision making process of international expansion. Ghemawat (2001) classifies these characteristics into four different categories, namely cultural, administrative (or political), geographic, and economic distance. Large differences in these categories between the host and home country could cause obstacles that MNEs would need to evaluate during the location choice process (Johanson & Vahlne, 1977). For example, large differences in administrative or political characteristics might mean different trade agreement and policies, higher risks, potential preferential treatment to local MNEs, barriers to entry, among others. The empirical study by Demirbag and Glaister (2010) exactly illustrates this point for R&D investments, that the greater the host country risk characteristics, the less likely that country will be chosen for investments. According the Johanson and Vahlne (1977), the smaller these distances between host and home country, the
more attractive the host country becomes for investments. Furthermore, Dunning (2000) describes additional pulling factors through the location advantages that a country may have such as the potential and growth of markets, the availability and quality of natural and labor resources, and production and transportation costs. Depending on the industry and motivations of an MNEs, these factors will play a significant role in the location choices the MNE makes.
Dunning (2000) describes ownership advantages as the competitive advantages that are specific to a firm, which include scarce, unique, and sustainable capabilities as well as the competencies of management that an MNE possesses in comparison to competitors. Furthermore, he describes the internationalization capabilities as the abilities of MNEs to create, transfer, recombine, and exploit ownership advantages internally instead of through outside parties (Dunning, 2000). Furthermore, another push factor is the motivation of MNE activity, which can be categorized into four main types; market seeking, resource seeking, efficiency seeking, and strategy asset seeking. Marketing seeking activities are to satisfy a foreign market, resource seeking activities are to gain access to natural resources, efficiency seeking activities are to encourage a more efficient division of labour, and strategic asset seeking activities are to defend or improve existing ownership advantages (Dunning, 2000). Lastly, the type of activity to be implemented may also have considerable influence on the location choice. For example, if the activity is knowledge or research intensive, the MNE will be more likely to locate in countries with high R&D intensity versus countries with low R&D intensity (Chung & Alcácer, 2002).
One article by Goerzen, Asmussen, and Nielsen (2013) has taken a very interesting approach to the location choice concept. They explore another potential pull factor however on a sub-national level of analysis, namely the features that pull MNEs towards what they consider to be global cities. What they find is that global cities have three characteristics that distinguish them from other areas, which pull MNEs to locate in these global cities rather than elsewhere. These characteristics are international connectedness, the advanced producer services, and the cosmopolitan environment. What the authors conclude is that MNEs are very likely to locate in these global cities because the characteristics of these global cities can potentially diminish some of the liabilities of foreignness they will experience by doing business abroad. However, while the conclusions are interesting, the authors do not discuss if the results would hold for different types of MNEs or different types of investment activities. Because there can be drastic differences in the MNEs and their motivations as described above, the benefits that such a global city provides may not necessarily be the objective of certain MNEs activities. Goerzen,
Asmussen, and Nielsen (2013) themselves note that further research is needed within this global cities topic in order to understand MNE geographic behaviors. Assessing whether these conclusions are consistent across MNEs in different contexts and implementing different activities would contribute to the literature and extend this interesting concept.
Goerzen, Asmussen, and Nielsen’s (2013) global city builds on the concept of firm agglomeration, where Ellison and Glaeser (1994) initially recognised that firms will locate in similar locations within countries as costs related to transportation of products, people, and knowledge greatly diminish with proximity. Mariotti et al. (2010) further find that agglomeration location decisions are affected by two factors, namely information externalities and knowledge spillovers. Information externalities, meaning the unfamiliarity and lack of knowledge in the local market, leads to MNEs locating in similar locations and imitating other firms which have faced and possibly resolved these issues to decrease potential costs. Knowledge spillovers relates to the inflows and outflows that a firm might experience due to agglomeration. Knowledge inflows are generally a positive learning experience by the firm, where they gain access to information from other firms in the cluster and therefore potentially increase their value and performance. However knowledge outflows are generally negative as knowledge and information is lost to other firms in the cluster (Mariotti et al., 2010). These spillovers can occur through several mechanisms including the mobility of skilled labour and technology, the relationships and connections with the local community such as research facilities, government and individual agencies, and universities (Audretsch & Feldman, 2003). The extent to which firms perceive the benefits of agglomeration will in turn influence the decision whether they will locate in the cluster (McCann & Mudambi, 2004).
In summation, there are numerous factors which can influence the location decisions of MNEs. However, in many cases these factors have been explored in the context of advanced market MNEs and thus, as previously mentioned, many scholars have begun to question and debate the applicability of these traditional theories to the differing contexts of emerging market MNEs. The following section will describe the highlights of this debate.
3.2 EMNE Debate
The debate surrounding IB literature originates from the puzzling behavior that EMNEs have exhibited in the past few decades. Ramamurti (2012) has identified these so-called “puzzles”, first questioning why emerging market contexts are able to produce such EMNEs in the first place and second, how they have internationalized in a way that does not conform
to the traditional IB theory propositions. Consequently, on the one hand, some academics consider that traditional theories cannot comprehensively explain how EMNEs have internationalized, and believe new theorization is necessary. While on the other hand, other academics have argued this is not necessary, and merely an extension of traditional theory is needed (Ramamurti, 2012).
Emerging Market Context Puzzle
Per the traditional IB theories covered in the previous section, certain behaviors are expected from emerging market MNEs. First of all, on top of the liabilities of foreignness that MNEs experience as discussed above, EMNEs also experience the liabilities of emergingness, defined by Madhok and Kayhani to be “the additional disadvantage that EMNES tend to suffer by virtue of being from emerging economies” (2012, p. 28). These disadvantages include problems associated with underdeveloped markets, inexperienced customers and suppliers, low to middle income levels, weak infrastructures, and resources shortages. All of these can result in institutional voids and missing market mechanisms that can hinder the competitiveness of EMNEs (Cuervo-Cazurra & Genc, 2008; Khanna & Palepu, 1997; Madhok & Keyhani, 2012). It is the expectation that emerging markets need to experience years of IFDI before they gain the competency to be able to support firms taking steps towards internationalizations and OFDI (Ramamurti, 2012).
Furthermore, one of the establishing factors of Dunning’s Eclectic paradigm is that for an MNE to be able to internationalize successfully, it must retain significant ownership advantages, whether they are asset or transactional advantages (Buckley & Casson, 2009). In the case of these emerging markets, EMNEs often do not have these ownership advantages but instead have only common resources and should, according to traditional theory, not be able to internationalize competitively (Madhok & Keyhani, 2012; Ramamurti, 2012).
Numerous academics have tried to explain this phenomenon even though traditional theory proposes these EMNEs shouldn’t exist in the first place. The first argument proposes that EMNEs just do not possess these ownership advantages however are able to use common resources such as cheap labour or natural resources in order to internationalize (Ramamurti, 2012; Ramamurti & Singh, 2009). However, if this were the case, these common resources would not support a sustainable competitive advantage for EMNEs as MNEs could comparably gain access to them. Another argument is the springboard perspective presented by Luo and Tung (2007), who purport that EMNEs actually internationalize in order to gain access to these
lacking ownership advantages and help overcome the institutional voids in the home market (Madhok & Keyhani, 2012; Mathews, 2002b). These two arguments illustrate how academics have attempted to explore alternative theories to explain how EMNEs have internationalized so effectively however the other side of the story is that traditional theories can be extended to explain this phenomenon.
One argument extending Dunning’s OLI model is that EMNEs do possess these ownership advantages, however these appear to be very different to those present in advanced market MNEs (Ramamurti, 2012). While often riddled with institutional voids, the contexts in which these EMNEs develop enable them to acquire alternative advantages. These include the ability to better understand emerging market customer needs, the ability to function and manage in difficult institutional contexts and the ability to provide goods and services reflective of market income levels (Cuervo-Cazurra & Genc, 2008; Ramamurti, 2012). A secondary argument in this line of thought is that EMNEs are simply in a different stage of evolution in comparison advanced market MNEs and they will eventually develop these ownership advantages in time. In essence, the differences between EMNEs and MNEs are a result not of the market context of origin, but because of the difference in stage of evolution (Ramamurti, 2012). A last argument put forward posits that EMNEs first internationalize to acquire such advantages in order to integrate it and exploit them in the home context and subsequently internationalize again with improved ownership advantages (Ramamurti, 2012).
Internationalization Process Puzzle
The traditional IB theories such as the Uppsala model explained earlier describe the specific process that MNEs should undertake in order to internationalize, namely that MNEs should expand incrementally. They should first invest in locations close in “distance” (culture, economic, geographic, and administrative distance) to acquire the necessary knowledge and experience needed to operate successfully. From there, the acquired experience enables the MNE to further expand and make increasingly distant investments (Johanson & Vahlne, 1977). However, past few decades of EMNE internationalization has shown various processes of internationalization. In some cases, at a much faster pace and scope, and in other cases, bypassing this incremental method and rather than investing in “close” countries first, they invested directly to “distant” countries. Furthermore, it has been found that some EMNEs, instead of using low-risk and low-commitment methods, have been implementing aggressive
high-risk, high-commitment methods (Madhok & Keyhani, 2012; Mathews, 2002b; Ramamurti, 2012; Thite, Wilkinson, Budhwar, & Mathews, 2016).
Similar to the first “puzzle”, academics have as well tried to explain this phenomenon of the non-conforming EMNE internationalization processes. Mathews (2002b) argues that while the internationalization pace is accelerated, EMNEs still experience it on an incremental level (Mathews, 2002b; Ramamurti, 2012). Madhok and Keyhani (2012) make the argument that this faster internationalization pace is because of an aggressively competitive EMNE mind-set with the goal of “catching up” to established MNEs. However these arguments are considered to be far from sufficient in explaining EMNE behaviors (Ramamurti, 2012).
An alternative explanation which has gathered more support is that idea that the global context which the EMNEs experience is very different from the global context established MNEs experienced previously. As the global economy has become increasingly integrated and “flat”, access to the necessary resources to internationalize has become easier in comparison to the process advanced market MNEs went through (Ramamurti, 2012; Ramamurti & Singh, 2009). It has also started to “blur” the boundaries between firms and countries, which used to be much more distinct and strict (Narula, 2006). Further supporting this argument is the recognition that not only have EMNEs had accelerated rates of internationalization, but firms from advanced markets have accelerated as well, dubbed the “born global” firms that from very early on seek to derive performance from operating in multiple countries (Knight & Cavusgil, 2004; Mathews & Zander, 2007; Ramamurti, 2012). Essentially, globalization has changed the context in which EMNEs operate and may have significantly altered the factors necessary to successfully internationalize.
Taking the debate a step further, Mathews (2006) developed a new framework called the Linkage, Leverage, and Learning (LLL) framework which attempts to answer the question of how EMNEs internationalized so rapidly, even when they suffer the disadvantages of being late- and new- comers to the global market. Linkage refers to the focus on acquiring resources externally through linkages, such as partnerships and joint ventures with other firms in the global market. Leverage suggests that the resources that have been accessed through linkage should be leveraged to the benefit of the EMNE and lastly, the learning refers to how the EMNE can take away knowledge from the linkage and leverage processes in order to become increasingly more efficient and successful (Mathews, 2006).
The debate surrounding the topic of IB theory of internationalization has shown to include numerous differing opinions, and in some cases, whole new frameworks from
academics. However, in order to further contribute to IB theory and literature, they also need to be supported by empirical evidence. It has continuously been proven that the traditional IB theories represent how advanced market MNE internationalize, however empirical evidence regarding EMNEs is much less prevalent. As the focus of this study is China and Chinese firms, empirical articles in relation to Chinese MNE expansion choices will be reviewed to assess which parts of the traditional theories and emerging theories are relevant.
3.3 Chinese Firm OFDI Factors
There seems to be a general set of pull and push factors that have been extensively researched such as host country factors like political risk, natural resources, cultural and geographic distance, size of market, wealth, and country characteristics (e.g. such as tax, unemployment rate, labour costs, etc.) (Amighini, Rabellotti, & Sanfilippo, 2013; Buckley, Clegg, & Cross, 2007; Castellani, Jimenez, & Zanfei, 2013; Duanmu, 2012; Kang & Jiang, 2012; Kolstad & Wiig, 2012; Ramasamy et al., 2012; Wei, Zheng, Liu, & Lu, 2014; Zhang, Wei, & Liu, 2013). While these factors have been extrapolated from the traditional IB theories, this makes sense as they have been the prevailing frameworks available for analysis. The following paragraphs describe how these different pull and push factors have influenced the outward foreign direct investments of Chinese MNEs and whether they correlate with what the traditional theories anticipate.
For example, when it comes to the effect of political risk on FDI, there are mixed results. Duanmu (2012) and Buckley et al. (2007) find that countries with higher political risk is associated with lower Chinese OFDI. De Beule and Duanmu (2012) find that firms generally accept some political risk, as long as it is comparable to the level of home political risks. Ramasamy et al. (2012) find that Chinese firms are attracted to countries that experience political risk and lastly, Quer et al. (2012) and Zhang et al. (2013) find political risk to be insignificant and not related to FDI location decisions.
Natural resources is another factor which has shown varying results in terms of influence on OFDI. Duanmu (2012) and De Beule and Duanmu (2012) find that natural resources have no significant impact on Chinese OFDI. However, there is a great amount of literature that finds the opposite, Zhang et al (2013), Kolstad and Wiig (2012), Amighini et al. (2013), and Ramasamy et al. (2012) find that natural resources are a pulling factor for Chinese OFDI.
Buckley et al. (2007) and Kang and Jiang (2012) find cultural proximity to have a positive significant impact on Chinese OFDI however Zhang et al. (2013) and Quer et al. (2012) find the effect of cultural distance to be non-significant. Furthermore, the presence of ethnic Chinese in the host country increased OFDI from Chinese MNEs (Amighini et al., 2013; Buckley et al., 2007; Ramasamy et al., 2012). Lastly, Buckley et al. (2007) and De Beule and Duanmu (2012) find geographic distance to be non-significant in relation to OFDI while Amighini at al. (2013) find a negative impact on Chinese OFDI.
Contrary to the previous dimensions, market size seems to have consistent results throughout the literature. Zhang et al. (2013), Kolstad and Wiig (2012), De Beule and Duanmu (2012), Amighini et al. (2013), Ramasamy et al. (2012), and Duanmu (2012) all find that market size is an important determinants of Chinese OFDI and large markets has attract significantly more OFDI than smaller markets.
Some other factors have also been researched however to a lesser extent. Amighini et al. (2013) and De Beule and Duanmu (2012) determine that Chinese OFDI is not overly attracted to rich or high income countries. Furthermore, Amighini et al. (2013) also finds that human capital attracts OFDI from Chinese firms. Moreover, Duanmu (2012) looks into multiple factors such as high costs, corporate tax, unemployment rates, and favourable exchange rates and finds them significant in deterring Chinese OFDI.
Traditional theories describe the impact that certain factors should have on the OFDI of MNEs however in many instances, the literature on Chinese MNEs and empirical results are not very consistent, as illustrated in Table 1. While in some cases, the results do correlate with expectations of traditional theory, in other cases, the opposite is found. This provides evidence for the notion that the behaviors of Chinese firms do not correlate completely with what the traditional theories state, however this does not mean that traditional theories are completely irrelevant. While this review has not covered all the existing literature, it does illustrate that there are gaps in the literature and additional empirical studies are necessary to paint a more complete picture of EMNE behaviors.
Table 1 – Influence of host country factors on Chinese OFDI
Article Political Risk
Natural Resources Cultural Distance Geographic Distance Market Size
Duanmu (2012) Negative Non-significant - - Positive Buckley et al. (2007) Negative - Positive Non-significant - Ramasamy et al. (2012) Positive Positive Positive - Positive Quer et al. (2012) Non-significant - Non-significant - - Zhang et al. (2013) Non-significant Positive Non-significant - Positive De Beule & Duanmu (2012) Partial Non-significant - Non-significant Positive Amighini et al. (2013) - Positive Positive Negative Positive Kolstad & Wiig (2012) - Positive - - Positive Kang & Jiang (2012) - - Positive - -
3.4 Business Groups
As covered earlier, many factors have been researched in relation to their influence on internationalization and location choice, however, one institutional factor that has not been extensively researched yet is the phenomenon of business groups present in emerging countries. Leff (1978) was the first to investigate the notion of business groups. He bases his conclusions on the Harvey Leibenstein theory, which essentially states that performing activities in developing countries must include the “opening of channels for input supply and for marketing of output” (p. 667) as the institutions that normally provide these services are lacking. The absence of these institutions result in problems or increased costs of EMNEs in acquiring essential inputs such as finance, technology, and management talent (Khanna & Palepu, 2000a). Leff (1978) thus concludes that business groups emerged as an instrument for dealing with the failures of the market and to overcome agency problems. Similarly, Fisman and Khanna (2004) further theorize these failures to be the institutional voids in the market. Ma, Yao, Xi (2006) extend this theory of why groups emerged and explain that there exist ownership voids as a subset of these institutional voids. Considering these different theories and conceptualizations, it seems as though business groups have emerged as a new institutional “level” in developing countries, where failures in the market are very common. This has incited much academic interest into how they work and what their impacts are and there is much debate on whether such business groups contribute to better performance and valuation of its affiliated firms. The literature on the benefits as well as challenges and costs follows next.
Benefits
Leff (1978) suggests that business groups enable the flow of capital beyond the resources of just a single firm but instead from a “pool” of multiple firms. Furthermore, these groups permit for effective communication and allocation of power, thus allowing firms to overcome general problems that size and efficiency may cause. Additionally, because of the common vertical integration in groups, the network allows for more ideal distribution and organization of investments. In line with Leff’s position, Elango, Pattnaik, & Weiland (2014) describe that groups enable the “acquisition, sharing, and transfer of critical managerial, financial, and organizational resources across member firms and provide a means for coordination and interfacing with external stakeholders” (p. 3205). Additionally Manos, Murinde, and Green (2007), Perotti and Gelfer (2001), Yiu et al. (2005), and Claessens, Fan, and Lang (2006) all find that the structure of business groups allow for internal capital markets, better distribution of resources between firms, and create unique access to resources such as government and foreign leans.
Khanna & Rivkin (2001) and Singh & Gaur (2009) describe other benefits of business groups, including the less costly access to capital, labour, product markets as well as raw materials than non-affiliated firms. Affiliated firms can exchange products and services internally to other firms without the risks or costs of transactions. Furthermore, they have access to political resources and greater opportunities because of the structure and connections of the group. The network of these business groups permits economies of scope as these groups are generally diversified (Khanna & Rivkin, 2001; Khanna & Yafeh, 2009; D. A. Singh & Gaur, 2009).
Keister (1998), Chang and Choi (1988), and Khanna and Palepu (2000a, 2000b) find superior performance of business group firms in China as well as Korea, India, and Chile. Chang and Choi (1988) find that business groups improve performance as a result of decreased transaction costs and economies and scale and scope. Chang and Hong (2000) describe that MNEs benefit from affiliation because of the possibility to share intangible and financial assets with other affiliated MNEs. Vissa, Greve, and Chen (2010) furthermore discover that affiliated firms have better responsiveness to performance. The structure of the group results in affiliated firms having an improved sense of problematic search and willing to change to improve performance when compared to non-affiliated firms.
costs associated with affiliation to a business group. Leff (1978) as well theorizes that business groups may create inefficiencies within the group. There is a consensus in the literature that firms experience “diversification discount”, where high levels of diversification results in loss of firm value because of the destruction of shareholder wealth. However, Khanna and Yafeh (2009) find that this is not representative of business groups as in some countries there is a diversification discount while in others, there is rather a diversification premium. Because of the nature of the business groups, in many cases costs of activities are shared between affiliates and while they may be beneficial to some, they could be detrimental to others and because of the formal and informal ties, firms suffering are bound to the group. Because of these powerful ties, firms have the responsibility to support badly performing affiliates and this strong support system may result in weak motivations for managers to run the firm resourcefully (Khanna & Rivkin, 2001).
Singh and Gaur (2009) as well argue that misallocation of resources across affiliates in the group as well as tunnelling of resources from one firm to another could result in losses in performance. Caves and Uekusa (1976) and Nakatani (1984) both find that keiretsus in Japan generally take on lowered risk however to achieve this, they generally forego returns and experience lower performance. Lee, Peng, and Lee (2001) follow chaebols in Korea and discover that groups that used to be traded at a premium were starting to be valued for much less starting in the mid 1990s. Ferris, Kim, Kitsabunnarat (2003) as well find that firms affiliated with chaebols experience lower valuation and performance in comparison to non-affiliated firms. This is because they prefer consistency in returns rather than expansion and do so by mainly investing in low performing industries and supporting members of group that have poor performance. Gopalan, Nanda, and Seru (2007) as well address this matter that affiliated firms have a responsibility to support member firms, even when performance is increasingly poor, thus potentially negatively influencing the performance of the whole group. Gaur and Kumar (2009) find similar results that business groups negatively moderate the positive impact of internationalization on firm performance. Yiu et al. (2005) additionally find that the age of a business group and government ownership is negatively associated with groups performance.
Kwon, Han, and Lee (2016), Singh, Nejadmalayeri, and Mathur (2007), Singh and Gaur (2009), and Bertrand, Mehta, and Mullainathan (2002) all further discover different negative effects of business groups on affiliated firms in comparison to non-affiliated firms. More specifically, Kwon et al. (2016) describe that firms experience financial constraints when
affiliated with a group. Furthermore, Singh et al. (2007) depict that the diversification that business groups generally experience is related to inefficiencies and thus decreased performance. Additionally, Singh and Gaur (2009) notion towards how ownership concentration of the group has negative impacts on performance. Bertrand et al. (2002) find that the controlling owners in business groups tunnel away resources from minority members. Many of these issues can stem from the agency problems often found in business groups (Claessens et al., 2006).
The previously reviewed studies demonstrate that there are numerous positive and negative impacts of business groups, however very few studies have addressed how affiliation to business groups may impact location decisions of MNEs. One study by Fisman and Khanna (2004) investigated firm plant location decisions within India and identified that firms affiliated with groups were more likely to locate in poorer and less developed regions of India when compared to un-affiliated firms. The authors interpreted these results to show that affiliation with a group provides a firm with many of the benefits described above, thus permitting them to perform better in these locations. This study will therefore attempt to incorporate this field of business groups with the factors influencing internationalization as well as global cities to extend the understanding of IB theory and literature on location choice.
4 | Analytical Framework and Hypotheses
There are three main concepts that support the analytical framework and hypotheses of this study.
4.1 Global Cities and Knowledge Concentration
As described earlier, Ellison and Glaeser (1994) find that firms will locate in close proximity to each other within countries because the costs associated with transportation of products, people, and knowledge will greatly be diminished. As a consequence of this proximity, knowledge spillovers occur and create an increasing concentration of knowledge (Mariotti et al., 2010). Furthermore, with the shift of firm value creation from tangible assets to intangible assets such as knowledge, it is necessary that any activities involved in knowledge-creation are successful (Mudambi, 2008). Cantwell and Mubamdi (2000) precisely argue that knowledge creating activities such as R&D should be located in areas with high knowledge concentration in order to be successful, while for example this is not necessary for less value-creating activities like manufacturing.
Taking these two concepts into account, locations with high knowledge concentration and innovative environments would therefore be the most attractive locations for implementing value-creating activities (Audretsch & Feldman, 2003; Cantwell & Kosmopoulou, 2000). Goerzen et al. (2013) contend that global cities are exactly such locations with high knowledge concentration and are conducive to innovation and knowledge creation. The authors describe numerous characteristics that specifically global cities hold (global interconnectedness, cosmopolitanism, and abundance of advanced producer services) which can contribute even more to mitigating potentially high transactions costs.
Therefore, it becomes clear that firms that are investing in value-creating activities will be most attracted to locations with high knowledge concentration, and more specifically, global cities because of the ultimate benefits and atmosphere they provide.
4.2 Technology and Knowledge Transmission
The advances in technology have allowed for more easy transmission of knowledge across and between these knowledge-concentrated locations as well as to other locations where MNEs are operating, even over long distances. While distance has a negative impact on costs for firms implementing activities with for example high transportation requirements (like manufacturing), value-creating activities generally deal with the transmission of knowledge and thus are less affected by distance. However, it is still important that firms become efficient and organized in how they transfer this knowledge across borders to different locations and within their networks. Because of the development in technology, the costs of transmitting this knowledge has only decreased, thus diminishing the constraints it places on a firm (Arora & Gambardella, 1994; Goerzen et al., 2013). Furthermore, firms have been able to develop increasingly efficient methods of transferring this knowledge both within their groups as well as over large distances because of the intense innovation competition (Goerzen et al., 2013; Kogut & Zander, 1992). In conclusion, advancements in technology has decreased the transaction costs of value-creating activities which in turn provides more freedom to firms in their location choices, while costs still have significant impacts for less value-creating activities such as manufacturing.
The two concepts discussed so far illustrate that firms have a desire to locate their value-creating activities in location where the knowledge concentration is high to increase the potential for success and this is increasingly possible because of the abilities and skills of the firm to efficiently transmit this knowledge across borders to the necessary networks. While
these concepts describe the behavior of MNEs looking to make investments across borders, this is especially relevant to EMNEs from China as they experience additional issues such as liabilities of emergingness. Because of the additional difficulties they face when making investments across borders, it is even more necessary to leverage all the potential benefits that such global cities provide to increase competitiveness, particularly when considering activities that are value-creating to the firm. The following hypothesis is therefore proposed and shown in figure 1 below:
Hypothesis 1: Chinese MNEs investing in high value-creating activities are more likely to
locate those activities in global cities than low value-creating activities.
4.3 Business Group Affiliation Effects
The last concept relates to the moderating effect that Chinese MNE affiliation to business groups may have on the relationship between location choice and activity type. It is clear from the literature review that business groups have both positive and negative impacts on the MNEs involved however, the effect of these impacts on the location choices of the Chinese MNEs is still unclear. As discussed above, the reason why a firm would locate its value-creating activities in a global city is because the beneficial knowledge intensive environment the location provides. However, in the case of business groups, certain characteristics of these business groups may provide an alternative method for accessing these benefits. For example, a business group can provide the opportunity for affiliated MNEs to share intangible assets and knowledge resources (Chang & Hong, 2000), share access to capital, labour, and product markets (Claessens et al., 2006; Khanna & Rivkin, 2001; Manos et al., 2007; D. A. Singh & Gaur, 2009). Because of the immense network and benefits business groups provide, they can essentially act as a substitute for locating in a global city and consequently, firms that are affiliated may deem it less necessary to locate their value-creating activities directly in those global cities. On the contrary, firms not affiliated with business groups very likely still face liabilities of emergingness and need to leverage the benefits of locating in a global city to succeed. The following hypothesis is therefore proposed and a full conceptual model is shown in Figure 1.
Hypothesis 2: Chinese MNE affiliation to a business group negatively moderates the
5 | Research Methods
This section provides details on the research methods of this thesis. A quantitative research design was selected to identify the potential influence of business groups on the relationship between location choice and different activity of Chinese MNEs. This section will address the sample, research design, the description of the study variables, and the process of data collection.
5.1 Sample
In order to examine the impact of business group affiliation on the relationship between location choice and activity type in global cities, a dataset was compiled detailing specific greenfield FDI made by Chinese MNEs. The data set created includes data drawn from multiple sources, namely fDi Markets, which is a Financial Times service and online database of greenfield investments throughout all countries and industries around the globe, Orbis database, which contains comprehensive firm-specific information on companies across the globe, the World Bank, which is an international institution that provides country level data in numerous areas of development, and lastly individual firm websites. Each of these databases provides comprehensive and reliable data on individual firms or countries that are part of this study.
The initial dataset was provided by this thesis’ supervisor however the data originates from the fDi Markets database. This dataset includes 2,092 greenfield investments executed by Chinese MNEs throughout the world between 2003 and 2011. Many details are provided for each investment including but not limited to the investing and parent company, the investment origin (country, state, city) and destination (country, state, administrative region, and city), the industry and sector of investment, and the type of industry activity implemented. Considering Figure 1 – Conceptual Model
High value-adding activities Location choice in global city
the focus of this study and hypotheses proposed, the entire dataset was assessed and reduced to include only the investments where the destination city was specified. The dataset was further reduced to include only investments where data could be found on the documented firms. The final dataset includes a sample to 1,022 investment and includes numerous additional details for each investment. The following section will describe all the relevant variables.
5.2 Dependent Variable
Global City
The dependent variable of this study is the location on city level of the Chinese MNE investments. The classification of what cities are considered global or not was developed by Beaverstock, Smith, and Taylor (1999), also used by Goerzen et al. (2013). They depict world cities as having advanced producer services or corporate services and created a list of cities that can be considered to have these services. Table 2 illustrates the list of world cities. For the sake of this study, all cities in this inventory will be considered as a global city regardless of the category they fall in as all categories show evidence of world city formation. To create a useable variable in the dataset, each investment was evaluated by the destination city and compared to the inventory of world cities. A dummy variable was created, 1 indicating the destination city as a global city and 0 indicating the destination city as a non-global city. This dummy variable represents the location choice of Chinese MNEs and serve at the dependent variable in this study.
Table 2 – Inventory of World Cities
Source - Beaverstock, Smith, & Taylor (1999)
5.3 Independent Variable
Investment Activity
The independent variable is the type of the investment activity being executed by the Chinese MNE. The source of this information is the fDi Markets database and classifies the investment into the following eighteen different types of activities: business services; construction; customer contact center; design, development, & testing; education & training; electricity; extraction; headquarters; ICT & internet infrastructure; logistics, distribution & transportation; maintenance & servicing; manufacturing; recycling; research & development; retail; sales, marketing & support; shared services center; technical support center.
The distinction made in this study is the difference between high value-adding and low value-adding activities. High value-adding activities are research & development, design, development, & testing; education & training, business services, headquarters, ICT & internet infrastructure, sales, marketing, & support, and shared services center. Low value-adding
Alpha World Cities Beta World Cities
Chicago Brussels Amsterdam Johannesburg Athens Detroit Mumbai Frankfurt Madrid Atlanta Kuala Lumpur Abu Dhabi Dresden New Delhi Hong kong Mexico City Bangkok Manila Adelaide Dubai Oslo London Moscow Barcelona Melbourne Almaty Dublin Philadephia Los Angeles San Francisco Beijing Miami Antwerp Edinburg Richmond Milan Sao Paulo Berlin Minneapolis Arhus Genoa Rio de Janeiro New York Seoul Boston Montreal Auckland Glasgow Riyadh Paris Sydney Budapest Munich Baltimore Gothenburg Rotterdam Singapore Toronto Buenos Aires Osaka Bangalore Guangzhou Seattle Tokyo Zurich Caracas Prague Birmingham Hanoi St Petersburg
Copenhagen Rome Bogota Helsinki Stuttgart Dallas Santiago Bologna Ho Chi Minh City Tashkent Dusseldorf Shanghai Brasilia Kansas City Tehran Geneva Stockholm Bratislava Kiev Tel Aviv Hambug Taipei Brisbane Leeds The Hague Houston Warsaw Bucharest Lille Tijuana Istanbul Washington Cairo Lima Turin Jakarta Calgary Lisbon Utrecht
Cape Town Luxembourg Vancouver Cleveland Lyon Vienna Cologne Manchester Wellington Colombo Marseille
Columbus Montevideo
activities are construction, electricity, extraction, logistics, distribution & transportation, maintenance & servicing, manufacturing, recycling, customer contract center, and technical support center. A dummy variable was created, 1 signifying high value-adding activities, 0 signifying low value-adding activities. This dummy variable represents the type of activity being implemented by the Chinese MNEs and serves as the independent variable of this study.
5.4 Moderator Variable
Business Group Affiliation
The moderator variable is the affiliation of Chinese MNEs to business groups. The source of this variable is the Orbis company information database and consists of three different criteria, namely the Bureau van Dijk Independence Indicator, the number of subsidiaries the parent firm owns, and the number of different industries these subsidiaries operate in. While each of these criteria individually do not necessarily point to whether a firm is affiliated with a business group, when combined, they may illustrate a clearer picture, therefore if firms fulfill at least two out of the three criteria, it is then considered as being affiliated with a business group. The reasoning behind allowing two of out three is an attempt to not exclude firms which may not necessarily portray all the characteristics of business groups.
The Bureau van Dijk Independence Indicator specifies the level of independence of a firm with regard to its shareholders (“BvD Independence Indicator,” 2016). There exist five classifications of independence, namely A, B, C, D, and U, A being the most independent while D being the least independent. They are specified per the percentage ownership of shareholders, excluding public, unnamed private and other shareholder which are considered incapable of exercising controlling power over the company. Indicator A is specified to firms where known recorded shareholders do not have more than 25% of direct or total ownership. Indicator B is specified to firms where known recorded shareholders do not have more than 50% of direct or total ownership however where one or more shareholder have an ownership percentage above 25%. Indicator C is specified to firms where a recorded shareholder has a total or calculated total ownership of over 50%. Indicator D is specified to firms where a recorded shareholder has a direct ownership of over 50%. Lastly, Indicator U is specified to have an unknown degree of independence (“BvD Independence Indicator,” 2016).
The distinction made in this study is whether a firm has a recorded shareholder with majority ownership. Thus, there is a division of firms where known recorded shareholders do not have more than 50% direct or total ownership (indicator A and B) and firms where known
recorded shareholders do have more than 50% direct or total ownership (indicator C and D). This study recognizes that one shareholder with majority ownership could potentially signify that the firm is part of a larger network and business group. A dummy variable is created, 1 to signify the independence indicator criteria as fulfilled (indicator A and B), 0 to signify it does not (indicator C and D).
As business groups are considered to be a collection firms with formal and informal connections (Khanna, 2000), the number of subsidiaries that parent firm holds must reflect this. He et al. (2013) and Chen et al. (2015) both find that in terms of number of subsidiaries, a parent firm with more than one subsidiary in its portfolio is enough to establish affiliation. However, having examined the data, it appears that elevating this number to a minimum of five or more subsidiaries is more appropriate in order to categorize the firms more accurately. A dummy variable is created, 1 indicates the subsidiary criteria is fulfilled, 0 indicates it is not.
A business group also generally consist of a diversified portfolio of businesses (Chittoor, 2009; Fisman & Khanna, 2004), therefore, the diversification, or number of industries the parent firm is involved in must also reflect this concept. In this case, the NACE codes available are the source of this variable and every unique 2 digit NACE code is considered as a separate industry. While there is no set distinction of how many industries a firm should be involved in before it is considered diversified, having looked at the data, it appears appropriate to set this limit at 5 or more industries. A dummy variable is created, 1 to indicate the criteria is satisfied (5 or more industries), 0 to indicate it is not (less than 5 industries).
Control Variables
Numerous control variables will be included in this study for more robust results. Certain firm specific characteristics will be controlled for including age of the company, the total assets of the parent company, the industry it operates in, the capital investment amount, the jobs created, and state ownership. Furthermore, several country level characteristics will be controlled for including the GDP, GDP growth, GPD per capita, and distance to country as well as some world bank governance indicators such as voice and accountability and regulatory quality. These data are collected from the Orbis database, the world bank, and the CEPii as well as individual company websites.
Firm Characteristics
The age control variable reflects the age of the company at the time of the investment and ultimately controls for potential differences in experience the company may have. The total
assets variable reflects the assets of the parent company and controls for differences in total company size. The industry control variable will control for whether company function in high-tech or low-high-tech industries, which Gordon and McCann (2000) have found has a potential influences on location choice. This variable is created based on NACE codes of the investing company and the NACE classifications of which codes are considered high technology and knowledge intensive services. All company with NACE codes in the high and medium-high technology sectors as well as knowledge intensive services are combined in this study are considered high tech while all other sectors are low tech. A dummy variable is created, 1 indicating high tech, 0 indicating low tech.
Country Characteristics
All GDP variables, GDP, GDP growth, and GDP per capita will control for differences in wealth of the host countries. The distance variable illustrates the distance between China and the host country in terms of kilometers and controls for the large differences in distance. Lastly multiple parameters of governance are included such as voice and accountability, and regulatory quality. These will account for different stages and quality of institutional development and the economy of the host country. Furthermore, the distance between host country and China will control for the geographical distance differences of investment locations. Finally, the size of the investment and the number of jobs created will control for potential differences in the size of MNEs making the investments.
6 | Results
6.1 Data Cleaning Checking
With a finalized dataset in SPSS, several steps were taken to control for missing or incorrectly entered data and to check for problems regarding multicollinearity. First and foremost, descriptive statistics were run to assess whether any variables had missing data and if there were any abnormalities regarding the dummy variables. What is identified is that the independent, dependent, and moderator variables have no missing cases while several of the control variables do have missing cases, as shown in Table 3. Furthermore, the minimum and maximum confirm that there are no outliers in the dummy variables as all the data points fall between 0.00 and 1.00.
Table 4 demonstrates the difference in distribution of investments located in non-global and global cities between low value-adding activities and high value-adding activities. Out of
the 300 low value-adding activities, the majority were located in non-global cities (205; 68.3%) while the minority were located in global cities (95; 31.7%). Conversely, out of the 722 high value-adding activities, the majority were located in global cities (521; 72.2%) and the minority in non-global cities (201; 27.8%). This initial frequency check suggests a difference between the investment destinations of low value-adding and high-value adding activities.
For the purposes of eliminating any possibility of multicollinearity, a correlation was run to verify that none of the variables are too highly correlated (>.8). The Pearson Correlation Matrix in Table 5 illustrates two highly correlated control variables, namely the Regulatory Quality Estimate with Host Country GDP per Capita (.826). For the purposes of the following analyses, the Host Country GDP per Capita will be left out. In all other cases, the absence of high correlation coefficients between any independent, dependent and control variable alleviates any potential issues of multicollinearity.
Table 3 – Descriptive Statistics
Table 4 – Distribution of activity types
Descriptive Statistics
Variables N Minimum Maximum Mean Std. Deviation
1 Activity Type (High/Low Value) 1022 0.00 1.00 0.71 0.46
2 Destination City 1022 0.00 1.00 0.60 0.49
3 Business Group Affiliation 1022 0.00 1.00 0.67 0.47
4 Tech/Intensity Level 927 0.00 1.00 0.81 0.40
5 State Ownership 1022 0.00 1.00 0.33 0.47
6 Age at Time of Investment 834 1.00 996.00 26.87 45.10
7 Total Assets of Parent Company 839 260.00 18151908955.00 393777292.19 1112138429.94 8 Host Country GDP 1003 249845600.00 14964372000000.00 2768981694802.20 4176186842455.53
9 Host Country GDP Growth 1003 -17.67 26.17 2.99 4.17
10 Host Country GDP per Capita 1003 161.88 82990.07 30224.50 18208.10
11 Voice and Accountability Estimate 1022 -2.20 1.75 0.66 0.85
12 Regulatory Quality Estimate 1022 -2.25 1.99 1.03 0.87
13 Distance between China and Host Country 1022 809.54 19079.88 7208.18 3522.30
14 Capital Investment size in Millions $ 1022 0.03 4586.00 80.77 323.13
15 Jobs Created 1022 1.00 12000.00 183.58 608.79 Low Value-Adding Activities High Value-Adding Activities Non-Global City 205 (68.3%) 201 (27.8%) Global City 95 (31.7%) 521 (72.2%)
