University of Groningen
Faculty of Economics and Business
Master Thesis
International Economics and Business
“The Determinants of Outward Foreign Direct Investment by EMNEs from
BRIC-Countries”
Name: Mike Kolman
Student Number: s1687360
Email: m.kolman.1989@gmail.com
Date: 09-01-2015
Abstract
This thesis investigates the determinants of outward foreign direct investment of emerging market multinational enterprises from BRIC-countries and the extent to which the Ownership-Location-Internalization model applies to location-decision behavior of these multinationals. With the use of the eclectic model by Dunning (1977), the institution-based view, and the findings of prior research a model was specified based on the gravity equation and hypotheses were tested using uni-directional bilateral outward foreign direct investment flows collected between 2003 and 2012. This thesis finds that BRIC-country investment flows are associated with market-size, research and development spending, and the institutional environment of the host-country. Furthermore, there is relatively strong support that the eclectic model is still applicable for the use of EMNE internationalization behavior.
1. Introduction
Foreign Direct Investment has steadily surpassed trade in importance for delivering services and goods to markets abroad (Sauvant, 2005). In 2011, outward foreign direct investment flows peaked at US$ 1.65 trillion, and currently the FDI stock consists of US$ 23.6 trillion (UNCTAD, 2013). Over the past few decades ample research has been conducted with regards to the determinants of outward foreign direct investment (OFDI) flows. The empirical research conducted, and hence the theory conceived out of this research, mainly came from a focus on developed countries (Dunning and Lundan, 2008). In recent years, more research has been conducted on the outward flows from multinationals of developing and emerging economies (Buckley et al., 2007; Kumar, 2007; Cheng and Ma, 2008; Cheung and Qian, 2008; Kalotay and Sulstarova, 2010). This shift in focus can easily be explained by the substantial increase of foreign spending of emerging market multinational enterprises (EMNEs). Evidence of this can be found in the case studies performed by Höltbrugge and Kreppel (2012), who analyzed eight EMNEs from the BRIC countries. A great example of this increased spending of EMNEs, as mentioned in the case study, is the acquisition of the Canadian MNE Inco by the Brazilian WEG Equipamentos Electricos for US$ 18 billion in 2006.
In 1982, multinationals from emerging and transition economies only held 9.2% of the world’s OFDI stock, whereas in 2011 this number increased to 31.8% in 2010 (IMF, 2011). Of these emerging economies, the OFDI from Brazil, Russia, India and China (BRIC-countries) increased most notably, from 6% of all OFDI in 2000 to almost 20% in 2012 (UNCTAD, 2013). To merely assume that the same determinants hold for EMNEs as for MNEs could lead to wrongful assumptions about EMNE behavior and location decision (Dunning, 2008). As will be discussed in this thesis, one of the reasons why EMNEs could behave differently could be the lack of certain firm-specific advantages (Rugman, 2009). This occurrence will be discussed in further detail in the theoretical section of this thesis. Even though the performance of the EMNEs from BRIC-countries as emerging outward investors might be uneven, all four countries are taking off with regards to increased foreign spending (Sauvant, 2005), which makes it interesting to analyze these countries together in this thesis. Furthermore, empirical studies showing BRIC-country EMNE’s OFDI determinants using bilateral OFDI data and a well-specified gravity model are quite sparse.
BRIC-countries as a group yet does not provide an empirical analysis of the determinants, but merely a qualitative investigation. The reason why it is interesting to study the determinants of EMNEs from BRIC-countries is because these countries are all taking off with regards to increased foreign spending, and it could be beneficial to investigate what makes certain host-countries more attractive for OFDI from EMNEs in countries that are experiencing rapid increases in foreign spending.
In order to describe the determinants of developed country MNE behavior, Dunning (1977) used the location-aspect of his eclectic OLI paradigm, a paradigm which explains a MNE’s competitive advantages, and identified four main determinants of FDI; market-seeking, efficiency-seeking, resource-seeking, and strategic asset-seeking FDI, which can be sub-classified into horizontal and vertical FDI. This paradigm and these determinants will be discussed in more detail in the literature section. While horizontal FDI types of investment are mainly focused on market expansion and serving foreign markets, the vertical FDI types are more focused on obtaining natural resources and strategic assets, which can be seen as a fragmentation across multiple countries (Helpman, Melitz & Yeaple, 2003; Ramondo, Rappoport & Ruhl, 2013). EMNEs can respond differently to the same stimuli as MNEs would, yet the theoretical framework might still apply. In previous studies, empirical evidence has been found for both types of FDI. Buckley et al. (2007) found evidence for Chinese MNEs that both horizontal and vertical types of FDI play a role in the location decision. Furthermore, Kolstad and Wiig (2012) found evidence for horizontal FDI of Chinese firms in OECD-countries (through market-seeking), and vertical FDI of Chinese firms in non-OECD countries (natural resource-seeking). Some authors have argued that the OLI paradigm does not fit for explaining the behavior of EMNEs, as these enterprises lack certain FSAs that help to face the liability of foreignness (Rugman, 2009). Yet, other authors have claimed that EMNEs established alternative FSA that can be used to overcome this problem (Cuervo-Cazurra and Genc, 2008). These alternative ownership advantages will allow for the use of the OLI paradigm in investigating the determinants of OFDI from EMNEs.
In more current literature, it is often mentioned that in order to do a formal empirical analysis it is important to take the changing institutional context in to consideration (Buckley et al., 2007; Dunning and Lundan, 2008; Kolstad and Wiig, 2012; Aleksynska and Havrylchyk, 2013). These authors state that it is necessary to include an institutional perspective in order to fully comprehend the host-country locational advantages and disadvantages. Furthermore, the period of interest in this research has been underpinned by Peng et al. (2009) as the rise of the institution-based view. For this reason, the adoption of the institution-based view when investigating the determinants of OFDI is considered, following in the footsteps of Buckley at al. (2007), Kolstad and Wiig (2012), and Aleksynska and Havrylchyk (2013). What is particularly interesting is the difference in institutions that can be quantified and used in order to measure the institutional distance between countries. This allows for the usage of institutional data from all BRIC-countries, even when large institutional differences might occur between these four countries. Furthermore, Aleksynska and Havrylchyk (2013) investigate the moderating effect of natural resources on institutional distance, while ignoring other possible determinants and moderating effects. Hence, this thesis will focus on all the determinants of OFDI by EMNEs from the BRIC-countries. In order to investigate these determinants, the following research question will be posed:
What are the determinants of OFDI from EMNEs in the BRIC-countries, and to what extent do these determinants differ from determinants of developed country MNEs?
The contribution that this thesis makes is that it takes both the classical determinants into consideration, as well as the institutional environment, therefore supplying a more complete analysis of OFDI determinants from EMNEs in BRIC-countries. The novelty of this thesis lies in a more complete use of possible moderating or mediating effects of determinants on the institutional environment, and the possibility to test the relevance of the locational-aspect of the OLI paradigm, and the OLI paradigm in general, for OFDI flows from EMNEs. In this thesis uni-directional bilateral OFDI flows data from EMNEs in BRIC-countries is used, provided by UNCTAD, and the gravity model with country-specific fixed effect dummies and robust standard errors.
In Section 3 the data, variables, and methodology will be fully discussed. In Section 4, the findings of the empirical analysis are stated, and in the fifth and final Section the thesis will be concluded and limitations will be provided, as well as suggestions for future research.
2. Literature Review
2.1 Overview of the BRIC-countries
The EMNEs from BRIC countries play an instrumental role in the rise of OFDI flows from emerging economies. In 2008, the BRIC countries accounted for roughly US$ 141.7 billion in flows, which comes close to 40% of the total OFDI from emerging economies (Sauvant, Maschek, and McAllister, 2010). Brazil peaked in 2006 with OFDI flows of US$ 24 billion, which was mostly attributed to a relatively large take-over by a Brazilian firm, and returned to US$ 11.6 billion in 2007. Russia obtained its highest levels of OFDI in 2011 with US$ 66.9 billion, which is a six-fold increase from 2003. India, which is the most modest country of the BRIC countries, saw its OFDI flows grow to 18.3 billion in 2011, while China’s OFDI flows continued to grow up to US$ 87.8 billion in 2012, which is around a twenty five-fold increase from the previous decade (UNCTAD, 2014). In the past few years, the levels of OFDI flows from EMNEs in most of the BRIC countries stagnated, decreased, or even became negative in the case for Brazil, due to the financial crises and the aftermath which followed. The effects of the financial meltdown were felt in most countries around the world, which can also be seen in the decline of incoming FDI in the BRIC countries in the period between 2008 and 2012 (Sauvant, Maschek, and McAllister, 2010).
Changes in foreign trade regime and the economic liberalization over the past two decades has not only led to FDI into the BRIC countries, but also sparked EMNEs in these countries to direct their investments outside the confines of the home-country. Goldman Sachs made the assumption that EMNEs from India and China will play a dominant role as suppliers of services and manufactured goods, while those in Russia and Brazil will simultaneously assume that role in the supply of raw materials. More specifically, Russian multinationals will dominate in the exploitation of oil and natural gas, and Brazilian multinationals will do the same in the markets of iron ore and soy (Höltbrugge and Kreppel, 2012). EMNEs from Brazil mainly target the Latin American region, and primarily focuses itself on the energy and mining industries (Sauvant, 2005; Sauvant, Maschek, and McAllister, 2010). Russian EMNEs focus themselves on the energy and mining industry as well, mainly through the refining and distribution of gas and petroleum. Furthermore, Russian EMNEs have been engaged in the procurement of raw resources and obtaining the access to strategic assets (Sauvant, Maschek, and McAllister, 2010). Among the host-countries of Russian OFDI is the European Union and the United States, but also the Commonwealth of Independent States (CIS). EMNEs from India made the switch from manufacturing toward non-financial services after the period of 2004-2005, which indicates that, the services sector may become of great importance to the trans-nationalization of EMNEs in India. The two main destinations for OFDI from Indian EMNEs were Russia and the United States, mainly for the presence of natural resources at first, and the existence of technology and knowledge in the latter (Sauvant, Maschek, and McAllister, 2010). Chinese OFDI flows are directed mainly to developing economies in close proximity to China itself with almost 75% of the US$ 87.8 billion in OFDI flows in 2012 staying within Asia. Other major recipients of Chinese OFDI are Australia and the United States. The largest part of this OFDI is directed towards the service sector, as well as the natural resources, which substantially increases the demand for raw materials world-wide (Sauvant, 2005; Sauvant, Maschek, and McAllister, 2010; UNCTAD 2014).
A large portion of OFDI from BRIC county EMNEs is directed towards offshore financial centers, which makes it hard to determine where these flows end up. Some of the OFDI is used for round-tripping, while some of it is redirected towards other economies (Sauvant, 2005; Aleksynska and Havrylchyk, 2013). The general direction and sector of the OFDI provides useful insights into some of the determinants of these EMNEs (i.e. a large presence in the oil and gas sector could explain large OFDI flows to countries with a high abundance of natural resources).
2.2 A General Theory of Foreign Direct Investment
one or more countries (Dunning and Lundan, 2008). Buckley and Casson (1976) state that the principles of FDI theory are that (i) MNEs internalize imperfect external markets, such as making use of exporters or licensing, to the point that the costs of increased internalization outweighs the sum of the benefits, and (ii) MNEs choose locations for certain parts of their value chain that minimalize the total costs of doing business. One of the most pre-dominant theories on competitive advantages of MNEs is the eclectic paradigm, or Ownership-Location-Internalization (OLI) model by Dunning (1977), which has been used to provide a general framework for the behavior of MNEs, and through this, the pattern of FDI, and the extent of its reach. This model by Dunning (1977) is the most relevant micro-economic approach to understanding EMNE investment behavior. Most other methods (i.e. currency-model) have taken a macro-economic approach and are hence less relevant for understanding EMNE behavior.
The OLI model stipulates that at least one of three necessary sources of competitive advantage must be present, for the prevalence of MNEs. These multinational enterprises have the opportunity to add value, with their operations abroad, when these MNEs have ownership-advantages, location-ownership-advantages, and internalization-advantages (Dunning, 1977; Dunning, 2000; Hennart, 2012). The first condition is that a firm holds certain ownership-advantages (Dunning, 1977), otherwise known as firm-specific advantages (FSAs) (Rugman and Verbeke, 1990). Examples of these particular FSAs are property rights or strong brand names. The second condition constitutes that a prospective host-country must offer certain location-advantages, otherwise known as country-specific advantages (CSAs). These CSAs consist for instance of the presence of cheap labor, access to growing foreign markets, and natural-resource endowments, as well as less tangible advantages (i.e. tariffs and institutions) (Dunning and Lundan, 2008). The first two sources are not enough for a MNE to establish abroad. The third source, which Dunning (1977) coined the internalization-advantage, points out that in order for an MNE to exploit their FSAs in a host country, through their operations, it must be more efficient to do so than through renting out these intangibles to local firms. This sub-paradigm of internalization can hence explain why an enterprise choses for export substitution through FDI (Dunning, 2000). These specific internalization-advantages emerge due to the imperfections in the global market for FSAs (Dunning, 1977). Dunning (1980) states that the larger the FSAs of an enterprise are, the greater the desire to internalize these ownership-advantages, and that the greater the CSAs are in a host-country, the larger the tendency for an MNE to start engaging in international production.
(i) Natural resource-seeking FDI, (ii) Market-seeking FDI,
(iii) Efficiency-seeking FDI (iv) Strategic asset-seeking FDI
These four types of FDI have evolved from the general theory of FDI, which has been based on the eclectic paradigm by Dunning (1977). This paradigm has been built mostly on the experiences of developed country investors and the empirical research that stems from these investors’ OFDI behavior. The types of FDI can, to a certain extent, be applied to the OFDI behavior of EMNEs, yet this will most certainly lead to gaps (Buckley et al., 2007). In the sub-sequent section more attention will be paid to the application of this general theory of FDI on EMNEs from BRIC-countries. In this section the applicability of these types of FDI to EMNEs in general is explained.
Natural-resource seeking FDI is aimed to acquire natural resources of a better quality, but at a smaller real cost than available at the home-country (Dunning and Lundan, 2008). For EMNEs this can be used to secure a steady flow of raw materials (i.e. oil or minerals) and energy sources which are harder to find at home (Buckley et al., 2007). Market-seeking FDI is used to invest in a particular country in order to serve the market with goods and services in these or neighboring countries (Dunning and Lundan, 2008). The market-seeking FDI of EMNEs stems from traditional trade-supporting reasons, and is used to gain access to foreign distribution network and to help facilitate export from the host-country (Buckley et al., 2007). Efficiency-seeking FDI occurs when MNEs seek for lower costs of production and operation in other countries, most dominantly in the pursuit of lower labor costs. Strategic asset-seeking FDI constitutes the acquirement of foreign assets (i.e. R&D facilities, strong brand names) in order to augment the enterprise’s portfolio of tangible assets and human competences, and to promote long-term strategic objectives (Dunning and Lundan, 2008). EMNEs tend to have an increased interest in strategic asset-seeking FDI due to their late entrance into the international business arena, and hence tend to lag behind developed country MNEs with regards to their technological capabilities (Li et al., 2012). The question that arises here is whether OFDI from EMNEs, and in particular those from the BRIC-countries, require additional information in order to explain the location decision-making process (Buckley et al., 2007). Furthermore, it is necessary to explain how this eclectic paradigm from Dunning (1977) applies to EMNEs as well, rather than solely to MNEs.
2.3 Application of the OLI Paradigm on EMNEs from BRIC-countries.
required. For the sake of comparison between the determinants of MNE and EMNE behavior there are two potential arguments as to why additional information is necessary to research and understand the determinants of OFDI from EMNEs in BRIC-countries. The first, which is a lack of traditional FSAs by EMNEs in the home-country, needs to be addressed in order to show that the OLI paradigm can still be used for EMNEs. The second is the existence of the institutional environment and institutional distance, which needs to be explained in order to create a more comprehensive model to investigate the determinants.
(i) A lack of traditional FSAs by EMNEs in the home-country
In the past, the OLI model has been used pre-dominantly by International Business scholars in order to examine the OFDI behavior of EMNEs (e.g. Dunning, 2006; Kumar, 2007; Buckley et al., 2007; Rugman, 2009). This general theory of firm behavior assumes that firms will make foreign direct investments when they possess certain FSAs (ownership-advantages) (Dunning, 1977). Rugman (2009), among others, points out that these EMNEs do not seem to carry many FSAs, especially the knowledge-based FSAs that have proven to be important to Western-type MNEs (i.e. specialized know-how about production, strong brand names, etc.). These authors express that the OLI model shows that the current OFDI of EMNEs is ill-advised and that these enterprises ought to wait before they can make sustainable investments until these EMNEs have obtained the proper FSAs (Rugman and Li, 2007). On the other hand, Cuervo-Cazurra and Genc (2008) argue that EMNEs hold a different set of FSAs than their cousin countries’ MNEs, and these alternative FSAs (i.e. knowledge on developing-market clients, production facilities and distribution networks that are better adapted to the conditions of the host-country, etc.) allow for a better understanding of emerging markets and provide an increased benefit when operating in environments categorized by poor institutions (i.e. imperfect contracting environment, inefficient judiciary, etc.). According to Ramamurti and Singh (2009), the alternative FSAs that EMNEs hold come from doing business in a harsher home-country environment, which can later be reinforced when doing business abroad. Furthermore, an important additional subset of CSAs previously less accounted for (i.e. local customer knowledge, natural resources controlled by governments, land hold by local firms) could possibly be not as available on the competitive markets (Hennart, 2012). EMNEs could benefit from the complementary CSAs as these enterprises have fewer costs in acquiring them. These complementary CSAs and alternative FSAs can be used as arguments as to why the OLI model is still relevant in the research of EMNE location-decision behavior.
In more recent research, the OLI paradigm has been questioned, and the demand for the inclusion of the institution-based view has risen (Peng et al., 2009). Institutions, which are more commonly known as “the rules of the game”, and more formally defined by North (1990) as “the humanly devised constraints that structure human interactions”, has received increased attention from a sub-set of authors (Peng et al., 2009; Dunning and Lundan, 2008; Bénassy-Quéré et al., 2007; Luo et al., 2010). These institutions can be seen as the formal rules (i.e. laws and regulations) and informal constraints (i.e. norms, conventions) (North, 1990). According to economic growth literature, a strong institutional environment has a positive effect on FDI inflow. Examples of what constitutes as a strong institutional environment within a host-country are, for instance, voice and accountability, political stability and the absence of violence, government effectiveness, regulatory quality, rule of law, or control of corruption. The presence of these institutions can provide a safer and well-controlled business environment, which will increase the incentive for MNEs to invest in a particular host-country (Acemoglu et al., 2005). Furthermore, FDI is usually characterized with high sunk costs and investors prefer to write long-term binding contracts that mitigate possible risks, and hence institutions in the host-country that enable contract enforcement are of particular importance (Busse and Hefeker, 2007). Poor institutional quality is often acclaimed to be one of the leading reasons of FDI inflow scarcity, yet this does not explain the rather large OFDI flows from EMNEs to countries with relatively poor institutions (Aleksynska and Havrylchyk, 2013). Both Kolstad and Wiig (2012) and Aleksynska and Havrylchyk (2013) argue that the notion of “psychic distance” is of importance, which states that EMNEs choose to enter into markets that are perceived as psychologically similar, as these host-countries provide lower levels of uncertainty. For EMNEs the preferred institutional framework of a host-country might look the same to a certain extent as that for a MNE, yet EMNEs might be able to obtain a competitive advantage in host-countries with more comparable institutions. Cuervo-Cazurra (2006) argues that EMNEs can outperform MNEs when OFDI from home-countries with high corruption target similar host-countries, because EMNEs exploit the familiarity in working within corrupt environments, and hence face lower operating costs. In order to research this several authors have used an alternative way to investigate institutional differences between host and home-countries, which has been dubbed institutional distance (Xu and Shenkar, 2002; Aleksynska and Havrylchyk, 2013; Kolstad and Wiig, 2012).
home-country, as MNEs prefer to operate in countries with a more similar institutional environment. Following this argument, this would entail that EMNEs might be able to outperform MNEs in institutionally weak environments that match their home-country institutions due to prior experiences in dealing with these institutions (Cuervo-Cazurra and Genc, 2008). This phenomenon could explain the surge in OFDI flows of BRIC-countries to other developing countries (Sauvant, 2005).
On the other hand, this argument might not hold for EMNEs. Most EMNEs do not stem from home-countries with a relatively strong institutional environment, as EMNEs tend to hail from relatively less-developed countries. Therefore, a large institutional distance between an EMNEs home-country and the country can entail a far superior institutional environment in the host-country. Furthermore, some empirical research shows that most OFDI flows from EMNEs can be explained by the presence of strong institutions, which offer transparency to potential entrants (Kolstad and Wiig, 2012; Aleksynska and Havrylchyk, 2013). The duality in empirical results for institutional distance and EMNE behavior requires to be taken into consideration in this thesis. It is apparent that in order to determine the effect of institutions on OFDI, institutional distance needs to be taken into consideration. By using the institutional distance as a determinant in the specification of the panel data model, rather than the institutional environment, it becomes possible to investigate whether or not EMNEs are deterred or attracted by a large positive or negative institutional distance, rather than investigate what institutional factors attract EMNEs. It is important to keep in mind that the institutional distance can increase both positively, as well as, negatively. This is due to the nature of the institutional distance which lies between -2.5 and 2.5, depending on the average institutional environment. The calculation of the institutional distance will be explained more thoroughly in the methodology section.
2.4 Outcome of prior research on OFDI motivations from the BRIC-countries
Havrylchyk, 2013). Furthermore, prior research has indicated that capital market imperfections, which can be found in the BRIC-countries, can cause a semi-permanent disequilibrium in the market for funds, which can be exploited by EMNEs that are considering OFDI. This capital abundance could lead to EMNEs using these market imperfections to create alternative FSAs (ownership advantages) (Buckley, 2004a). This could cause outward investors from BRIC-countries to have advantages in obtaining natural resources as capital becomes available below market rates. Russian and Chinese EMNEs, which are often state-owned, can have access to soft loans as a preferential treatment, in order to pursue political goals (Kalotay and Sulstarova, 2010; Buckley et al., 2007), and Indian and Brazilian firms are often considered to be family owned, which gives them easy access to cheap capital through family members (Kumar, 2007; Sauvant, 2005). These capital market imperfections allow BRIC-country EMNEs to face a smaller probability of financial losses when investing in host-countries with less developed institutions (Kolstad and Wiig, 2012). Especially in comparison with developed country MNEs, where these potential risks when investing in countries with less developed institutions do prevail. These host-countries with poor institutions are usually targeted in order to appropriate the natural resource rents that are present.
Host-market characteristics, such as size of the market, are generally considered to be a highly significant determinant of OFDI from BRIC-countries (Buckley et al. 2007; Kalotay and Sulstarova, 2010; Bénassy-Quéré et al. 2007; Cheung and Qian, 2009; Cheng and Ma, 2007). As the market size increases of a host country, so does the opportunities for efficient use of available resources and the creation of scale and scope economies via FDI (Kalotay and Sulstarova, 2010), which is coherent to the traditional market-seeking argument. Next to this it is important to keep in mind that BRIC-countries are emerging economies and are among the countries with lower on average GDP, which could constitute that these countries have smaller markets (Sauvant, 2005). This would only further drive EMNEs to invest in large markets in the form of market-seeking FDI in order to seek growth outside the home market (Dunning and Lundan, 2008). Although it is important to keep in mind that the BRIC-countries have relatively high growth rates, which could translate into market potential in the home-market in the long-run.
EMNEs have a limited need to invest in a host-country for this particular reason (Buckley et al., 2007). Furthermore, finding the appropriate data in an accredited database is exceptionally hard, and availability of labor cost data is limited. Cheung and Qian (2009) find some evidence for labor costs in the manufacturing sector, but are extremely limited by the use of one sector, a small set of countries, and a timeline of merely a few years. This is why I have omitted the efficiency-seeking type of FDI in this research paper.
The strategic asset-seeking FDI, or knowledge and technology-seeking, is used by EMNEs to justify overseas expansion in order to appropriate capabilities that are possessed by MNEs in developed markets (Li et al., 2012). Examples of these capabilities are brand names embedded in MNEs, R&D capabilities, and design facilities, which can normally only be accessed by EMNEs through the acquisition of these firms (M&As) or strategic alliances (JV) (Dunning and Lundan, 2008). The focus on overcoming their latecomer disadvantage has been coined a “springboard-strategy”, which enables EMNEs to catch-up in technological areas (Luo and Tung, 2007). As mentioned before, EMNEs lack a certain sub-set of FSAs, and through strategic asset-seeking FDI it becomes possible to obtain these advantages (Hennart, 2012).
2.5 The determinants of OFDI from EMNEs in BRIC-countries.
In the following section the determinants of OFDI are reviewed. These determinants are derived from both the eclectic model by Dunning (1977) and the additional information on institutional distance and prior research, and hypotheses on the direction of influence with regards to the distribution of OFDI from the BRIC-countries are made. The traditional determinants following the location-decision aspect of the OLI model are hypothesized as followed:
Natural resource-seeking FDI
The natural resource-seeking type of FDI refers to the acquisition of natural resources at a smaller fraction of the real cost compared to the home-country, or to those natural resources that are unavailable in the home-country (Dunning and Lundan, 2008). BRIC-countries have been characterized with a higher degree of economic development and internalization theory underlines the importance of an equity-based control in the reserves, or exploitation of, natural resources with regards to economic development (Dunning and Lundan, 2008). Single-country evidence has shown a positive relation between OFDI from BRIC-countries (Buckley et al., 2007 – China; Kalotay and Sulstarova, 2010 – Russia), and hence I hypothesize that:
Market-seeking FDI
Market size has been considered to be one of the important host-market characteristics and has been used in many studies on FDI determinants (Buckley et al. 2007; Kalotay and Sulstarova, 2010; Bénassy-Quéré et al. 2007; Cheung and Qian, 2009; Cheng and Ma, 2007). Most of these authors state that an absolute larger market in the host-country leads to better opportunities for generating profits, due to scale and scope economies. Furthermore, theory suggests that a growth in demand of a host-market will be positively related to the attraction of OFDI, as fast-growing markets present even more opportunities for these EMNEs (Buckley et al., 2007). Based on these assumptions I derive the following two hypotheses:
H2a: The OFDI of EMNEs from BRIC-countries is positively associated with absolute market size. H2b: The OFDI of EMNEs from BRIC-countries is positively associated with host market growth. Strategic asset-seeking FDI
A substantial sum of the OFDI that originates from EMNEs is directed towards developed economies in order for the EMNEs to appropriate technology and knowledge from MNEs located in the host-country (Li et al., 2012). This appropriation will give EMNEs a chance to overcome their initial lack of FSAs, as stipulated by Rugman (2009). It is expected of EMNEs from BRIC-countries to direct most of the strategic asset-seeking FDI towards developed economies in order to use the spring-board strategy, and overcome the current gap between both types of multinationals. Developed economies tend to be characterized with high levels of human and intellectual capital, which can boost EMNEs competitiveness on a more global scale by positive technological spillovers (Buckley et al., 2007). Some of the main variables in determining a host-country’s level of human and intellectual capital are R&D expenditure, school enrollment, and patent registration. The most complete data set is currently on R&D Expenditure as a % of GDP, which can be used as a good proxy for the strategic asset-seeking argument. Therefore, I hypothesize that:
H3: The OFDI of EMNEs from BRIC-countries is positively associated with R&D expenditure of the host-country.
The hypotheses H1 to H3 cover the traditional determinants found by Dunning (1977), minus the efficiency-seeking type of FDI, which has been omitted in this thesis. As previously argued, the importance of the inclusion of the institution-based view has become apparent, and will therefore be tested as well. The institutional determinants are hypothesized as followed:
Recent studies on institutional distance state that the multinationals are discouraged to invest in host-countries with poor institutions and tend to invest in host-countries that have a relatively similar institutional framework, and hence the institutional distance is small (Bénassy-Quéré et al., 2007). But, Aleksynska and Havrylchyk (2013) claim that it is misleading to assume that there exists a symmetric preference for better and worse institutions, and that one should not estimate institutional distance between host and home-countries in absolute terms, which has been done in previous literature. In order to determine the relation between institutional distance and OFDI it becomes important to look at the sign of the distance, be it negative or positive. Aleksynska and Havrylchyk (2013) find evidence that EMNEs that invest in host-countries with better institutions tend to do so in countries with the best possible environment – as this is most transparent, and that EMNEs tend to be deterred by a large and negative institutional distance towards the host-country, which confirms the institutional-theory on negative institutional distance. The empirical research that focuses primarily on institutions as a determinant for OFDI flows shows that a positive institutional distance, which means flows towards host-countries with better institutions, tends to have a positive effect on OFDI flows (Aleksynska and Havrylchyk, 2013). Therefore, I hypothesize that:
H4a: The OFDI of EMNEs from BRIC-countries is positively associated with a positive institutional distance.
However, both Aleksynska and Havrylchyk (2013) and Kolstad and Wiig (2012) find a strong relation between the effect of institutional distance on OFDI and the natural resource-seeking argument as mentioned in the previous section. The growing importance of natural resource-seeking FDI by EMNEs has recently appeared to outweigh the stopping power of a bad institutional environment. Investors from emerging economies that are used to operate in an environment with poor institutions are able to compete with developed country MNEs as they face lower costs of operation due to familiarity with corrupt systems, as opposed to the investors from developed countries (Aleksynska and Havrylchyk, 2013). The effect of the presence of natural resources in a host-country moderates the effect of institutional distance; the larger the negative institutional distance is towards the host-country, the more OFDI is attracted due to the presence of natural resources. Therefore, I hypothesize that:
In Aleksynska and Havrylchyk (2013) the moderating effect of institutional distance is only applied to the presence of natural resources within the host-country. If the presence of natural resources has a moderating effect on institutional distance, then it can also be possible that this occurs in the presence of strategic assets. A good institutional environment allows for the presence of strong brand names and better protection of patent rights. Both strong brand names and patent protection are important for the attraction of strategic asset-seeking type of FDI (Dunning and Lundan, 2008). In order to get a complete picture of the effects of institutional distance it is important to include a possible moderating effect between institutional distance and research and development of the host-country. The presence of such institutions (i.e. patent right protection) could mediate, rather than moderate, the attraction of strategic assets; the larger the positive institutional distance towards the host-country, the more OFDI is attracted due to the presence of strategic assets. Therefore, I hypothesize that:
H4c: The OFDI of EMNEs from BRIC-countries is positively associated with a positive increase in institutional distance between the home-country and a host-country with high spending on Research and Development.
The first two hypotheses regarding institutional distances stem from the work of Aleksynska and Havrylchyk (2013), while the final hypothesis tests for a mediating effect of strategic assets on institutional distance, which has previously been unaccounted for.
2.6 Additional determinants of OFDI from EMNEs in BRIC-countries
In previous research a number of additional determinants have been used as control variables in order to enhance the predictability of the model. These determinants were found to be significant, while other authors found some of these same determinants to be insignificant. In order to be as exact as possible, the following determinants are included, all which have been used in prior articles and estimations.
Openness to FDI constitutes as the degree of openness of a host-country to the OFDI from EMNEs of the home-country, which can be measured as the ratio of inward FDI stock to GDP in the host-country. The attractiveness of the host-country increases with the openness to FDI, as it becomes easier to invest in such a country, and hence the likelihood of the host-country to be chosen as a destination will increase (Buckley et al., 2007). Therefore, a positive correlation between the openness of FDI of the host-country and OFDI flows from the home-country is expected.
The host-country inflation rate can play a crucial role with regards to an EMNEs price-setting behavior and profit expectations. Highly fluctuating and unpredictable inflation rates will discourage an EMNE to enter a host-country through market-seeking FDI as it creates uncertainty, which makes long term planning difficult, and the long-term profitability could possibly diminish as the domestic currency could devaluate (Buckley et al., 2007). Furthermore, this inflation rate unpredictability could diminish export oriented FDI as the locally sourced inputs increase in price, making it harder to sustain a cost advantage in third markets. Therefore, the host-country inflation rate is expected to be negatively associated with the OFDI flows from the home-country (Buckley et al., 2007; Kolstad and Wiig, 2012).
In order to further specify the estimation, and to be as concise as possible, a subset of time-invariant dummy control variables will be added that have been used in previous papers on FDI determinants. A common language between the host and home-country increases the ease of communication and understanding, and will most likely lead to an increase in OFDI towards that host-country (Cheng and Ma, 2007; Buckley et al., 2007; Bénassy-Quéré et al., 2007). A colonial history between the host and home-country could indicate historical trade between both countries that could have sustained over time (Bénassy-Quéré et al., 2007). Therefore, a colonial history would have a positive effect on OFDI from the home-country. Contiguity, which is the sharing of a border between the home and host-country, will make trade easier as transportation costs are most likely low and a mutual understanding exists of each other’s culture The expectation here is that a common border will increase the OFDI towards that host-country (Bénassy-Quéré et al., 2007; Cheng and Ma, 2007; Kosltad and Wiig, 2009).
3. Data and Methodology
3.1 Data
research both the OFDI flows and stock has been used for analysis, yet in this paper the focus is on the annual foreign investment behavior of EMNEs, hence the use of flows are a better fit. Furthermore, flows correspond better with the proxies for the determinants as immediate changes can be noticed in the data when using flows rather than stocks (Buckley et al., 2007; Aleksysnka and Havrylchyk, 2013). The data on OFDI flows comes from the UNCTAD database and ranges from 2003 to 2012. Due to limited data availability some home-countries have more data points than others, and not every year is equally represented as China has data from 2003 to 2012, while India has data from 2010 to 2012, which makes the panel dataset unbalanced. The reason for the missing data is the fact that the data was up until recently been poorly reported on, and certain countries (especially developing countries) are only now seeing the importance of proper data collection. When dealing with an unbalanced panel data it is important that certain assumptions hold. Traditional methods of panel data analysis are only valid when the data is missing completely at random (MCAR), which means that missing data is not correlated with any other variable. In order to test whether or not this assumption is violated it is possible to perform a Robust Hausman-test. If this assumption does not hold, the proper way of dealing with the missing data would be to perform a “multiple imputation” in STATA (Kwak, 2011). Both this particular Hausman-test and the multiple imputation are beyond the scope of this thesis, and hence the assumption that MCAR is not violated is taken for granted, which allows for the use of traditional methods - the gravity model. In order to analyze the determinants of the BRIC-countries all tax havens and offshore financial centers have been omitted from this study. Flows towards these host-countries are mainly for taxation purposes or redistribution, and it becomes unclear what happens with these flows upon entering these countries (Aleksysnka and Havrylchyk, 2013; Kolstad and Wiig, 2012). Data on tax havens and offshore financial centers have been taken from the OECD.
has been taken as a variable for natural resource-seeking FDI. All data used in order create these variables to determine the presence of different types of FDI are based on prior research on OFDI determinants (Buckley et al., 2007; Kolstad and Wiig, 2012; Cheng and Ma, 2007; Bénassy-Quéré et al., 2007). All variables contained some missing values for certain years, which have been estimated by looking at the growth rate of the years before or after, depending on the presence of those observation. By looking at the average growth rate over the years that were available it became possible to estimate the missing values. The data used in order to determine the institutional distance between the home and host-country are an average of six different institutional variables, ranging between -2,5 and 2.5. The variables used in order to construct the average institutional level of both the host and home-countries are voice and accountability, political stability and the absence of violence, government effectiveness, regulatory quality, rule of law, and control of corruption. By using the simple averages of all six variables and comparing them between countries a relatively simple variable for institutional distance can be constructed. The same six factors of an institutional environment, and method in order to create the data for institutional distance, were used in Aleksynska and Havrylchyk (2013).
The data for the additional determinants comes from the CEPII database, UNCTAD and the World Bank. The CEPII database provided data on the geographical distance between the capitals of the home and host-country, common language, colonial history (colonial link and common colonizer), and contiguity. UNCTAD provided the data on openness of the host-country to FDI. The data on inflation comes from the World Bank. An overview of the data used in the thesis is displayed in the Table 1 in the Appendix.
3.2 Methodology
ni n i ni
GS
M
X
This equation can be modified in order to allow for FDI flows rather than for trade flows. The term S(i) stands for the “capabilities” of FDI exporting home-country as a supplier to all host-countries, M(n) captures all the characteristics of the host-country n that could trigger FDI motivations from the home-country, and
ni pertains the accessibility of the host-country to thehome-country, which will hold a value between 0 and 1. Lastly, the term G stands for the “gravitational constant”. This general model of gravity holds two conditions; (i) the insistence that each variable enters multiplicatively, and (ii) any possible third country effects must be mediated through the n and I multilateral terms. The origin of the multiplicative form of gravity models stems from physics and hence has mainly persisted due to historical usage. Head and Mayer (2013) therefore redefine the gravity model to a structural gravity model in order to allow other functional forms. In the structural gravity model – Si and Mn are observables - bilateral FDI is given by:
ni n n i i ni
Y
X
X
Here
Y
i
nX
ni is the value of Inward FDI in the host-country, andX
n
iX
ni equals to the outward FDI of the host-country towards the home-countries.
i and
n areconsidered to be the multilateral resistance terms, and are defined according to the following functions:
l l l nl n Y
and
l l l li i X
According to Head and Mayer (2013) the structural gravity model relies on two important assumptions; (i) spatial allocations of expenditure for the host-country, and (ii) market-clearing for the home-country. In order to have a symmetric gravity equation, where FDI is balanced,
X
i
Y
i and
ni
in, which also implies that
i and
n are equal to each other. This would furtherimply that Si equals to Mn in the general model of gravity, as was first presented by Anderson and
van Wincoop (2001).
proportion relative to the host-country GDP or other possible OFDI determinants as specified in the theoretical section (Head and Mayer, 2013). The panel data set used in this thesis contains variables that vary over time, which makes the use of fixed effects appropriate. When using fixed effects there is an assumption that something within the host or home-country may impact or bias the predictor variables and using fixed effects allows for controlling this possible bias. The standard estimating procedure is to take the logarithms of the general model of gravity, which also solves issues regarding normality, as is described here below:
ni n i ni G S M X ln ln ln ln
ln The equation in multiplicative form allows for an estimation that can be used for regression purposes with the inclusion of country-specific dummy variables that measure these fixed effects as mentioned above. Here the estimation relates the natural logarithm of the OFDI flows between the market size of the home and host-country, a composite term measuring a form of barrier (distance) and terms measuring incentives and impediments to investing abroad. This allows one to alter the regression in such a way that it fits with the particular dependent and independent variables used in the data set. In the current study the focus is on the determinants of OFDI from BRIC-countries (home) towards host-countries. In previous studies by Aleksynska and Havrylchyk (2013) and Bénassy-Quéré et al. (2007) the gravity model has been used in order to determine, respectively, the role of institutional distance and natural resources on outward FDI, and institutional determinants of inwards FDI. This thesis continues to build upon these models by creating a more complete overview of determinants of outward foreign direct investment, and furthermore adds an additional interaction term between institutional distance and research and development. In doing so, the following estimation for OFDI flows has been obtained, which allows one to implement all the possible determinants of FDI flows, including certain non-traditional economic determinants:
ijt jt it ij ij ij ij ij jt jt ijt jt jt jt ijt ijt
e
jt
it
D
ComCol
ColHis
ComLang
Contig
Open
Infl
InstD
RD
RES
GDPG
MarketSize
OFDI
)
ln(
)
ln(
ln
12 11 10 9 8 7 6 5 4 3 2 1 0
product of both GDPs has been taken in order to follow the research of Alyksynska and Havrylchyk (2013), and to create a similar basic gravity model. RES is the determinant for natural resource endowment in the host-country, while RD represents the strategic asset-seeking determinant. InstD stands for the institutional distance between the home and host-country, and Beta 5 will answer the hypothesis on institutional distance. Beta 6 to 12 are the control variables where ln(D) is the logarithm of the absolute geographic distance between the host and home-country, Infl. the inflation rate in host-country, and Open shows the openness of the host-country to inward FDI flows. Contig, ComLang, ColHis, and ComCol are non-traditional and time-invariant economic dummy variables that respectively allow for contiguity (common border), a common language, and a historical relationship based on colonization (colonial history and common colonizer). Following Aleksynska and Havrylchyk (2013) and Baldwin and Taglioni (2006), I have included time-variant home and host-country dummy variables. The last two terms in the estimation, it(it) + jt(jt), remove any form of time-series or cross-section correlation that are the result of a possible omitted variable bias, as well as, it allows me to control for the omission of multilateral trade resistance (Anderson and van Wincoop, 2001). These country dummies are binary variables that will capture all country-specific characteristics. In this gravity model that entails one such variable will be set to one whenever the home-country is Brazil and zero otherwise, and another for when the host-country is the United States, and so on. This will happen for all of the host and home-countries. The e(ijt) at the end of the estimation expresses the error term.
ijt jt it jt ij ij ij ij ij jt jt ijt jt jt jt ijt ijt
e
jt
it
AID
RES
D
ComCol
ColHis
ComLang
Contig
Open
Infl
InstD
RD
RES
GDPG
MarketSize
OFDI
*
)
ln(
)
ln(
ln
13 12 11 10 9 8 7 6 5 4 3 2 1 0
And, ijt jt it jt ij ij ij ij ij jt jt ijt jt jt jt ijt ijte
jt
it
InstD
RD
D
ComCol
ColHis
ComLang
Contig
Open
Infl
InstD
RD
RES
GDPG
MarketSize
OFDI
*
)
ln(
)
ln(
ln
13 12 11 10 9 8 7 6 5 4 3 2 1 0
Here RES*AID is an interactive variable between resource endowments and absolute institutional distance and RD*InstD is an interactive variable between research and development and institutional distance. Furthermore, in order to compare the gravity model with a more standardized panel data estimation model the generalized least-square regression will be added as well. Within the generalized least-square regression it is not necessary to take the interaction between both the host and home-countries GDP into consideration and it becomes possible to solely look the host-country determinants, rather than the interaction between both home-host-country and host-host-country determinants. The GLS model will only contain possible determinants and deterrents from the host-country.
3.3 Econometric Issues
In the following section the descriptive statistics and the econometric issues involved in the gravity model are discussed. The econometric issues that need to be addressed in this thesis are normality, multicollinearity, heteroskedasticity, and autocorrelation a small set of procedures and tests have been done. Furthermore, a Hausman-Test will be performed in order to determine whether or not the decision to include fixed effects, rather than random effects, is justified. These tests have been performed in STATA after performing the regression, and will be elaborated on below.
within these variables is high. All dummy variables are binary values, which entails that the value lies between 0 and 1.
Table 2
Obs Mean Median Std. Dev. Min Max
Ln(OFDI) 1197 3.722772 3.73767 2.212395 0 10.84424 Ln(Market Size) 1197 26.95179 27.03825 2.004802 21.68561 32.5265 GDP growth 1197 3.443968 3.39471 4.27882 -17.95499 34.5 N. Resources 1197 9.37856 3.388772 14.25144 0 78.6149 R&D 1197 1.133548 .7567467 1.034828 .01854 4.52323 Inst. Distance 1197 .8684075 .8203253 .9465232 -1.201509 2.62066 Inflation 1197 9.895495 4.110814 55.3488 -4.863278 1096.678 Openness 1197 51.08944 34.0203 70.09764 .6856588 579.7843 Contiguity 1197 .1495405 0 .3567695 0 1 Com. Lang. 1197 .0451128 0 .2076382 0 1 Col. His. 1197 .0701754 0 .2555492 0 1 Com. Col. 1197 .0125313 0 .1112863 0 1 Ln(Distance) 1197 8.600307 8.858085 .7821449 6.532284 9.867729
To take into consideration the non-normal distribution of certain variables I have taken the natural logarithm of all continuous variables, as is also required by the specification of the gravity model. The variables OFDI, GDP and Distance are the three continuous variables in the specification and hence the natural logarithm of all three variables has been taken. By doing so it becomes possible to account for the existence of non-normality in the distribution. All other variables are either in the form of percentages or dummy variables and therefore non-continuous, as the range of these variables are not infinite. In order to proof the presence of a non-normal distribution and the necessity of the natural logarithm a Shapiro-Wilkinson test is appropriate. For the variables OFDI, GDP and Distance the test results can be seen in Table 3. All three continues variables have a high z-statistic, with a p>z of 0.00000, which means that non-normality is present for these variables. Table 3
Obs W V Z Prob>z
OFDI 1197 0.13803 640.427 16.119 0.00000
Market Size 1197 0.31730 507.2335 15.538 0.00000 Distance 1197 0.94524 40.684 9.244 0.00000
collinearity matrix. In general, collinearity exists when the correlation between two variables is 0.8 or higher, as can be witnessed in the collinearity matrix, or when the variance inflation factor is above 10. As can be seen in both tables, none of the variables crosses these limits; hence the analysis is not plagued by multicollinearity. Table 4 VIF 1/VIF Inst. Distance 2.49 0.401076 R&D 2.39 0.417804 Ln(Market Size) 1.71 0.583379 Ln(Distance) 1.68 0.593656 Col. His. 1.60 0.623680 Contiguity 1.54 0.648865 Openness 1.52 0.659562 N. Resources 1.36 0.734426 Com. Lang. 1.32 0.759482 GDP growth 1.23 0.810438 Com. Col. 1.09 0.914505 Inflation 1.05 0.951611
It is possible that the variance of for all observations of an independent is not the same, and that the error term fluctuates more for certain observations than for others. This phenomenon is called heteroskedasticity, and the independent variable and random error are heteroskedastic (Hill, Griffiths and Lim, 2011). Ignoring heteroskedasticity can lead to biased and overestimated results. After performing a Breusch-Pagan/Cook-Weisberg test it is apparent that heteroskedasticity is present in the model. The Null-hypothesis of constant variance can be rejected as the Chi-squared equals to 7.23 and the Prob > Chi-squared equals to 0.0072. In order to prevent the existence of heteroskedasticity, and its effects on the regression, the inclusion of robust standard errors can correct for this.
Autocorrelation is the correlation of error terms in different time periods (Hill, Griffiths and Lim, 2011). Serial correlation can bias the standard errors and cause the actual outcome to be less consistent and efficient (Drukker, 2003). The Wooldridge-test can determine whether or not there is a presence of autocorrelation. The Null-hypothesis of no first-order autocorrelation is not rejected at the 1% or 5% level as the value of F equals to 2.977 with a Prob > F of 0.0863. Therefore, there will be no correction for serial correlation in this thesis.
rejected as the Chi-squared equals to 107.52 with a Prob > Chi-squared of 0.0000. Table 5 shows the results of the Hausman-test. This entails that the fixed effects specification of the model, as mentioned above in the methodology section, is superior over a random effects specification.
Table 6
---Coefficient---(b) (B) (b-B) Sqrt(diag(V_b-V_B)
Fixed Random Difference S.E.
Ln(Market Size) 1.244961 .9466437 .2983176 .0352049 GDP growth -.0127067 -.017902 .0051953 .0019406 N. Resources .0208371 .0108359 .0100012 .0080037 R&D .3055144 -.2641509 .5696653 .2725639 Inst. Distance .2961599 -.151512 .4476719 .4157755 Inflation .0015816 .0027394 -.0011578 .0008476 Openness .0062454 .008896 -.0026506 .0019936
b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg
4. Empirical Results
4.1 Results
Table 7
Model 1 Model 2 Model 3 Model 4 Model 5
Ln(MarketSize) 0.4468*** 0.9775*** 1.2198*** 0.9785*** (0.0284) (0.0616) (0.1070) (0.0615) GDP growth -0.0131 0.0103 -0.0133 -0.0345** (0.0116) (0.0204) (0.0116) (0.0131) N. Resources 0.0091 -0.0077 0.0094 0.0101* (0.0130) (0.0265) (0.0129) (0.0044) R&D 0.5992* 1.3539* 0.6634 -0.0854 (0.3050) (0.6414) (0.3877) (0.0785) Inst. Distance -1.1860*** 1.0049 -1.1081** -0.0694 (0.2980) (1.3964) (0.3992) (0.0905) Inflation 0.0021 0.0012 0.0021 0.0021 (0.0022) (0.0016) (0.0021) (0.0012) Openness 0.0055 0.0110 0.0055 0.0069*** (0.0034) (0.0087) (0.0034) (0.0014) Contiguity 0.0927 0.0260 0.7004 0.0193 0.1066 (0.1790) (0.2729) (0.7286) (0.2759) (0.1697) Com. Lang. 1.7203*** 0.6537* 0.8772 0.6494* 0.8823** (0.2674) (0.3095) (1.5872) (0.3109) (0.2804) Col. His. 0.3876 0.5069 1.3151 0.4980 0.5835** (0.2613) (0.2873) (1.0527) (0.2879) (0.2239) Com. Col. 1.3269*** 0.6109 2.7633*** 0.6150 0.7736 (0.4979) (0.4130) (0.6569) (0.4124) (0.5029) Ln(Distance) -0.9733*** -2.0610*** -0.8845 -2.0689*** -0.8199*** (0.0843) (0.1225) (0.6391) (0.1257) (0.0844) RES*AID 0.0167 (0.0488) RD*InstD -0.0551 (0.2046) Ln(GDP) 0.4918*** (0.0364) _cons -1.8228 -4.5133* -26.8225*** -4.3918* 4.5087*** (1.0795) (1.9969) (6.4746) (2.0619) (0.8769) N 1197 1197 273 1197 1197 R² 0.297 0.596 0.710 0.596 0.315
Note: Robust standard errors are in parentheses.
confirming hypothesis 2a, whereas the growth in GDP (GDPG) appears to be insignificant and negative, which is counter-intuitive to the expectation. Therefore, hypothesis 2b cannot be confirmed. On the other hand, the Research and Development (RD) is significant and positive conforming hypothesis 3. Institutional distance (InstD) Is significant but appears to hold the incorrect sign, hence hypothesis 4a is also not supported. As can be seen in Model 3, the interaction term between Natural Resource Endowments and Absolute Institutional Distance (RES*AID) for all negative observations of institutional distance is insignificant, even though the sign is as expected. Furthermore, the natural resource endowments were insignificant as well. Therefore, there is no evidence that supports hypothesis 4b. The coefficient of the interaction variable between research and development and institutional distance is not only insignificant but also holds the opposite sign. Hence, even though the coefficient of research and development is significant, there is no evidence that supports hypothesis 4c. The next section will discuss in greater detail each of these main findings, and show that the use of a good specification of the model is important to prevent wrongful assumptions about determinants.
4.2 Discussion
The natural logarithm of the product of the GDPs of the home and host-country, which was a proxy for absolute market size, has a positive influence on BRIC-country OFDI flows, with a one per cent rise in the market size the BRIC-country OFDI increases by almost 0.9%. This can be interpreted as evidence that market seeking is a motive for EMNEs in choosing a location and is in concurrence with the findings of both Buckley et al. (2007) and Bénassy-Quéré et al. (2007). Unfortunately, there is no evidence that proofs that market growth also attracts OFDI. On the contrary, it appears that growth in GDP, which is insignificant, causes a slight decrease in OFDI. It is important to keep in mind that the coefficient is rather small, which entails that the effects of GDP growth are barely noticeable, although it might be necessary to try and explain this phenomenon. It could be possible that when EMNEs focus on the long run GDP growth rates of the host-country tend to be of less importance. The GDP growth rates in the home-country tend to be quite high and could cause the attention for market growth potential to shift towards the home-market (Dunning and Lundan, 2008).
insignificant could be the fact that some of the BRIC-countries have a relative abundance of natural resources at home, which could cause EMNEs to focus on other determinants of the location decision. When analyzing the resource endowments it becomes apparent that the proper specification of the model is highly necessary. In Model 5, where the standard GLS method was used to investigate the determinants, the natural resource rents were considered to be significant on a 10% level. By using the proper specification, the assumption that the appropriation of natural resources is significant, and hence a determinant of OFDI, does not hold. As a result hypothesis 1 was rejected rather than accepted.
The expenditure rate on research and development is significant, and a one per cent increase in the percentage of GDP invested in R&D leads to a 1.3% increase in the outflow of OFDI towards the host-country. This could be evidence for the argument of justification of overseas expansion in order to appropriate capabilities that are possessed by other MNEs in developed markets in order to obtain the FSAs that EMNEs currently miss (Li et al., 2012), and gain advantage over MNEs through the springboard-strategy, which enables EMNEs to catch-up in technological areas (Luo and Tung, 2007).
making it even harder to proof that there exists a moderating effect of the presence of natural resources to institutional distance. The fact that the institutional distance is positive in Model 3, where only the observations with negative institutional distances were used, could hint at a possible preference towards an equal institutional environment, rather than a lesser one. Unfortunately, the coefficient remains insignificant, and hence cannot serve as evidence. The interaction variable between research and development and institutional distance in Model 4 is not only insignificant, but also negative, and the institutional distance in general holds the wrong sign. Therefore, it is impossible to proof that the spending of the host-country on research and development mediates the institutional distance.
Both the levels of inflation and the openness to trade of the host-country have a positive sign, yet none of the two variables attain significance. The positive sign of the host-country inflation could suggest that investors from BRIC-countries are more attracted to economies with moderate demand inflation under the assumption that moderate inflation tends to accompany economic growth (Buckley et al., 2007). This cannot be said with certainty as the inflation variable never attained significance. Openness to trade has a positive sign, but also did not attain significance. This makes it impossible to state that OFDI flows increase towards host-countries that are more open to trade. The signs of contiguity, common language, colonial history, and common colonizer are all as expected, but only common language is a significant non-traditional control variable. This could entail that a common language between home and host-country is the last remaining historical variable that is of importance in the decision-making process of location choices by EMNEs. As expected by the gravity model the geographical distance is highly significant and negative, indicating that high transport costs, that accompany investing in countries far away, is still present to this day. One control variable stands out in Model 3, which is the Common Colonizer. In Model 3 only those observations were used with a negative institutional distance. A possible remark here can made that when countries face a negative institutional distance, the relationship of a common colonizer can be of importance.
5. Conclusion and Limitations
5.1 Conclusion
expanded upon with moderating and mediating effects. The main challenge was to fit the eclectic paradigm with both the addition of the institution-based view, and the evidence for the lack of traditional FSAs of EMNEs. Based on the theory by Dunning (1977) and prior research on EMNE location-decision behavior it was possible to devise a model that would test for both the traditional determinants, as well as allow for the institution-based view to be present within the model. Furthermore, it also allowed for the expansion on the work of Alyksynska and Havrylchyk (2013) by adding a possible mediating effect of strategic assets on institutional distance. The up-to-date and large dataset allowed for testing of the OFDI determinants using a wide range of independent and control variables. This thesis provides evidence that that the determinants of OFDI from BRIC-countries is mainly characterized by both traditional types of FDI - with the exception of natural resources-seeking FDI and efficiency-seeking FDI - and institutional distance.
