Multinationals and their foreign business activity: Does Distance Matter?

Multinationals and their foreign business activity:

Does Distance Matter?

- A Multidimensional distance approach –

Master Thesis

by

Benjamin Ihbe

Dual Masters Award in Advanced International Business and Management

University of Groningen

Faculty of Economics and Business

Supervisor: Prof. Dr. Sjoerd Beugelsdijk

Newcastle University

2

Word count

14.080

3

Abstract

Previous studies have shown inconclusive findings about the influence of cross-national distance dimensions on foreign business activity. This paper argues that U.S. Multinational Enterprises (MNEs)’ affiliates’ value added activity is negatively affected by distance. The concept of Liability of Foreignness (LOF) and the general acknowledged negative relation of distance to foreign business activity lead to the hypotheses that value added (VA) and value added per Employee (VApE) are negatively influenced by distance. Using Ghemawat’s (2001) multidimensional framework of Cultural, Administrative, Geographic and Economic Distance (CAGE), the topic is approached. On the basis of three distance concepts a single measure for each dimension of the CAGE framework is established. The following panel analyses of 52 countries over the period of 1997-2008 shows partial support to these hypotheses. Cultural and Economic Distance were found to have a negative influence.

4

Table of content

Abstract ... 3 List of Tables ... 5 List of Figures ... 6 Abbreviations ... 7 1. Introduction ... 8

1.1 Purpose of the Study ... 8

1.2 Approach to Develop the Hypotheses ... 10

2. Theory and Hypotheses ... 11

2.2 Liability of Foreignness... 11

2.3 CAGE – Framework and Distance Constructs ... 14

2.3.1 Cultural Distance ... 15

2.3.2 Administrative Distance ... 16

2.3.3 Geographic Distance ... 17

2.3.4 Economic Distance ... 17

2.4 Literature review about cross-national distance research ... 19

2.5 Hypotheses Development ... 22

3. Methodology ... 23

3.1 Data Description and Calculation Approach ... 23

3.2 Dependent Variables... 25

3.3 Independent Variables ... 26

3.4 Control Variables ... 28

4. Results ... 30

4.1 Descriptive Statistics and Correlation ... 30

4.2 Factor Analysis – Operationalization of the CAGE framework ... 31

4.3 Panel Analysis – Empirical Illustration ... 34

5. Final Section ... 38

5.1 Discussion ... 38

5.2 Conclusion and Future research direction ... 39

References ... 42

5

List of Tables

Table 1. Correlations of Distance Measures ... 30

Table 2. Correlation Matrix of Berry et al.’s Distance Measures Distance (basic and averaged values) ... 31

Table 3. Factor Analysis of Distance Measurements ... 31

Table 4. Rotated Table of Factor Analysis of Distance Measurements ... 32

Table 5. Descriptive Statistics and Correlations of Panel Variable data. ... 34

Table 6. Panel Analysis Value Added (VA). ... 35

6

List of Figures

Figure 1. Connection between the reasons for LOF and distant dimensions. ... 12

7

Abbreviations

BEA U.S. Bureau of Economic Analyses

GDP Gross Domestic Product

MNE Multinational Enterprise

VA Value Added

VApE Value Added per Employee

8

1. Introduction

1.1 Purpose of the Study

The rationale of Multinational enterprises (MNEs) is to engage in business activity across country borders (Nachum & Zaheer, 2005). This type of activity, foreign business activity, can be distinguished between trading relationships and foreign direct investment (FDI)1. The Internationalization theory, established in the 1970’s, emphasized that MNEs engage in FDI “whenever they perceive that the net benefits off their common ownership of domestic foreign activities, and the transactions arising from them, are likely to exceed those offered by external trading relationships” (Dunning & Lunden, 2008, p. 93-94). In more practical terms MNEs try to increase their profits and their competitive position by investing directly in foreign markets, e.g. establishing an affiliate in the country. MNEs do so for various reasons, for example to participate from local rapid economic development, to reduce overall costs, to gain access in economic unions or to acquire know how (Rugman & Collinson, 2009). By entering foreign countries MNEs are prone to host country specific influences and suffer Liability of Foreignness (LOF). LOF are “the costs of doing business abroad that result in a competitive disadvantage for an MNE subunit” (Zaheer, 1995, p. 342) in relation to local companies. This concept represents an underlying building stone of this paper. It has high relevance due to the continuous interweaving of the world economy and MNEs global activity. Campbell et al. (2011) state that since the 21st century research has become increasingly aware of incrementing effect of distance between countries on LOF (e.g. Eden & Miller, 2004; Kostova & Zaheer, 1999). This period of time is also characterised by a growing interest in International Business (IB) research regarding cross national distance dimension as a key reason (e.g. Tsui, Nifadkar & Ou, 2007).

Contradictory findings exist in the academic discourse about their influences on foreign business activity (e.g. Kirkman, Lowe and Gibson, 2006; Tihanyi, Grifith and Craig, 2005, Tsui, Nifadkar and Ou, 2007; Sousa and Bradley, 2008). Some authors present evidence for an influence whereby others do not. This master thesis aims to contribute some clarification to this research field.

To assess the influence of distance dimensions FDI stocks have often been used (Blonigen & Tilger, 2011). This is not an undisputed approach following the critique of Beugelsdijk et al. (2010)2. To gauge

1

“Investment involving a long term-relationship and reflecting a lasting interest and control by a resident entity in one economy (direct investor) in an entity resident in an economy other than that of the investor. The direct investor’s purpose is to exert a significant degree of influence on the management of the enterprise resident in the other economy” (IMF, 1993).

9

the relation of foreign business activity and different distance dimensions, this work uses U.S. MNEs’ affiliates Value Added (VA). Dunning (1993, p. 7) describes VA as “the best indicator of the overall or sectorial economic significance of MNE activity”. Derived from VA this paper also includes U.S. MNEs’ affiliates Value Added per Employee (VApE) as a second measurement of foreign business activity to gain even more insights. So far no analysis has been conducted using this unbiased type of data to assess this topic. Its utilization contributes to the existing discussion about different distance dimensions and their influence on foreign business activity and aims to add some degree of clarification. In consequence of the line of thought my research question develops as follows:

Does cross-national distance between the home and host country affect the amount of Value Added (VA) and Value Added per Employee (VApE) being created by U.S. MNEs’

foreign affiliates in host countries? Method to answer the Research question

To address this question I use the CAGE distance framework of Ghemawat (2001) which breaks down cross-national distance into four dimensions: Cultural, Administrative, Geographic and Economic. I allocate a key measurement to each dimension. This is realized by utilizing cross-national measures out of a pool of three cross-national distance concepts: Cultural Distance by Kogut & Singh (1988), Psychic Distance Stimuli by Dow and Karunaratna (2006) and Institutional Distance measurements created by Berry et al. (2010). The three cross-national distance concepts were chosen in accordance to their heavy usability on the country level where the research question is situated. Hofstede’s cultural dimensions, on which the Kogut and Singh index is based, are supposed to be used on the country level (Hofstede, 2001). The latter two, due to their elicitation of data on the macro level, meet also this prerequisite. After identifying the most influential measurements for each dimension of the CAGE framework, taking insight from the theory and means of a factor analysis into account, a new measurement for each distance dimension is created. The newly created distance measures are then included as independent variables into a panel analyses for VA and VApE as dependent variables. The panel analysis uses a multidimensional dataset created by the author. The dataset consists of 118 U.S. MNEs’ foreign affiliates host countries for the period of 1997-2008. Taking full data availability for all years as a prerequisite the panel data analysis was conducted with 52 countries, providing a considerable basis to answer the research question.

Target group

10

markets practitioners can take measures to minimize LOF, becoming more competitive.

Structure of the paper

The paper is organized as follows: After explaining the approach to develop the hypotheses an understanding for LOF is generated. Afterwards the CAGE framework in conjunction with theory about the three chosen cross national distance constructs is explained. The following literature review evaluates existing cross national distance research and consequently hypotheses are developed. The second part introduces the methodology of this paper including a description of the created dataset and its variables and presents the panel analyses and their results. The final section comprises a discussion of findings, a conclusion as well as limitations and future research directions.

1.2 Approach to Develop the Hypotheses

First the concept of Liability of Foreignness (LOF) will be described to familiarize the reader with the underlying theoretical construct of this master thesis. Secondly the CAGE-Framework of Ghemawat (2001) is introduced. To each dimension theoretical considerations of the Kogut and Singh Index (1988), Psychic Distance Stimuli (Dow and Karunaratna, 2006) and Institutional distance measures (Berry et al., 2010) are allocated. The allocation was conducted in accordance to the effects captured by each measurement. The following literature review provides an overview of the academic field of international business research relevant to distance dimensions. It provides key insights and findings of articles and survey papers to assess the research field. Furthermore it is enriched by a review of relevant articles in the field of cross national empiric distance research published in three top business journal from 2005-2011.This is done in pursuance of presenting a comprehensive and contemporary picture of the academic research field. The aim of the literature review is to assess the relevance of the topic in the field of International Business (IB) and to give an overview about relevant research and theories. I decided to concentrate on the time period 2005-2011 for the detailed review of journals due to the availability of survey papers for prior time periods. Following the literature review approach of survey papers (e.g. Kirkman et al. 2006, Morrison, 2010) I conducted a key word search using distance, cultural

and cross-national for the time period of 2005-2011 using the acknowledged database EBSCO – Business

Source Premier, journal specific databases and consulted journals individually if the abstract was not available or contingent questions arose. The reason to focus on top journals to gauge latest research is described by Tsui et al. (2007, p. 427) by stating “it informs us about major advances and helps us to identify the challenges evident in even the best work”. The journals I decided to review for articles in the field of cross national empirical distance research are the Academy of Management Journal (AMJ), the Strategic Management Journal (SMJ), and the Journal of International Business Studies (JIBS), which are available via the library of Groningen. The reason why I decided to review these journals is as follows: Firstly, in the Journal Citation Report (JCR) 2011 of Thomson Reuter (Thomson Reuter, 2011) all three Journals are placed within the top ten English journals out of 103 listed in the business3 category in

11

2010. Depending on the listing criteria such as Total Cites, Impact Factor or 5-year impact factor the actual rank of each journal varies slightly. Nevertheless, all three are in either category in the top 10. JIBS qualifies thereby especially because, as the name implies, it emphasizes international business and cross-border distance research and is the only top journal amongst the international business journals (Pisani, 2009) and consequently provides a multitude of articles in my research field. Examples are some of the main reference articles of this Master Thesis such as Berry et al. (2010), Dow and Karunaratna (2006), Slangen and Beugelsdijk (2010) and Beugelsdijk et al. (2010). Secondly, the review of international management research by Pisani (2009) also ranked the AMJ, SMJ and JIBS within the top-20 management journals. Other more on international business research focused journals exist but I follow the argumentation of Pisani (2009) that these are not considered to be amongst the top business journals. Consequently, a review of AMJ, SMJ and JIBS should provide a reliable and contemporary overview about relevant cross-border distance research. A detailed summary of the journals reviewed can be found in Appendix E.

2. Theory and Hypotheses

2.2 Liability of Foreignness

In foreign host markets affiliates of MNEs are confronted with the Liability of Foreignness (LOF) (Campbell et al., 2011). This has been subject of academic research in a multitude of articles (e.g. Mezias, 2002; Nachum, 2003; Zaheer, 1995). In the following paragraph the construct of LOF is explained within its context.

What are the main reasons to engage in business activity in a foreign host market and make business activity prone to LOF? The key determinant can be found in are location specific advantages. Dunning and Lundan (2008) distinguish thereby a variety of aspects, e.g. natural and created resource endowments, economies of agglomeration and spillovers, and economic and/or political systems. A clear definition of LOF is controversial in the academic field. I follow the general accepted approach used by Beugelsdijk (2010) who refers to the first definition of LOF by Zaheer (1995), already mentioned in the introduction but included here for the sake of completeness, which utilizes insights of Hymer’s Costs of Doing Business Abroad (CDBA) (1960 / 1976).

LOF are “the costs of doing business abroad that result in a competitive disadvantage for an MNE subunit” (Zaheer, 1995, p. 342)

12

market which is not known in detail. Thirdly, costs arise because of the host countries’ environment such

as “the lack of legitimacy of foreign firms and economic nationalism” (Zaheer, 1995, p. 343). The fourth source identified by Zaheer (1995) is costs related to the home country environment that means for example trade restrictions from the home country to the host country. Beugelsdijk (2011) connected thereby reasons for LOF with cross national distance dimensions as shown in Figure 1 below indicating their direct connection.

Figure 1. Connection between the reasons for LOF and distant dimensions.

There are differences in the effect of LOF throughout industries and countries. Nevertheless, the concept refers to lower profitability of MNEs’ activity in a foreign market compared to host country companies and that survival in the host market is more difficult for foreign firms compared to local ones. With these determinants the concept is applied on the firm and country level Zaheer (1995).

Consequently foreign companies are deemed to overcome higher costs connected to an activity abroad. They have to overcome drawbacks of LOF and implement their Firm Specific Advantage (FSA) to build up a prosperous business activity. In general it is acknowledged in IB literature that with larger distance successful foreign activity is more difficult (Beugelsdijk, 2011). Mezias (2002) states the negative relationship between LOF and firm performance but waken awareness that measuring the relationship is not flawless.

LOF is a complex construct which is not as straight forward as it may seem. For example, relatively high business failures rates might also occur for comparatively close countries, in relation to different distant dimensions (Lee & Shenkar, 2008). This was found by O’Grady (1996) who noticed a relatively high failure rate of US firms with an investment in Canada. Lagendijk and Oinas (2005) emphasize also, with reference to Beugelsdijk and Cornet (2002), that a ‘close neighbour is not as good as a distant friend’, meaning that, in their underlying case, geographic proximity leads not inevitably to a profitable business activity if other dimensions are too far from each other. Consequently the concept of ‘farther equals harder’ is not definite per se. A degree of distance potentially leads to competitive advantages obtained by new ideas or approaches which establish while being active in foreign markets (Lagendijk & Oinas, 2005). LOF is explained in more detail, by showing its development within the academia and placement in today’s theory, due to its underlying implications for this work.

Hymer (1960, 1976), referred to as the founder of cross-national business studies in IB (Berry et al., 2010), was the first to develop a concept about the disadvantages foreign firms face in comparison to local firms while engaging in business activity in a host market. He is considered as the founder of the IB

Reasons for LOF Distance Dimensions

Spatial Distance  Geographical Distance

Unknown market  Cultural Distance

Host countries environment  Cultural Distance

13

field of cross-national distance studies by Berry et al. (2010). Hymer refers to this concept as the Costs of Doing Business Abroad (CDBA) because these disadvantages are translated in higher expenses for MNEs compared to local companies. Hymer (1960, 1976) defined CDBA as “all the additional costs faced by a home country firm connected with its market-based (selling and/or buying) activities in a foreign country, relative to the costs faced by a local firm engaged in similar activities. These activities could be as minimal as exporting into a host country market (where the local firms also sells in the host market) to the extensive activities involved in extraction and processing raw materials (where the local firm also extracts and processes)”. Since then CDBA has been a key concept of MNE theories (e.g. Hymer, 1960, 1976) and consistently present in academic literature (e.g., Mezias, 2002; Zaheer, 1995) as shown in Campbell et al. (2011). In this context Zaheer (1995) introduced Liability of Foreignness (LOF) as a concept standing next to CDBA (Zaheer, 2002). Consecutive academic research has been inconclusive about the usage, meaning and differentiation of these two concepts leading to a special issue of the Journal of International Management (2002) delineating the constructs of CDBA and LOF. Following this Eden and Miller (2004) proposed a different perspective towards the relationship between CDBA and LOF. They see LOF as the essential part of CDBA and refer to it as the social costs of conducting business in a foreign country. Zaheer (2002, p. 357), in contrast, interprets LOF and CDBA as two separate concepts whereby CDBA is an “economic concept consisting primarily of market-driven costs related to geographic distance” and LOF is a social concept which is based primarily on structural, relational and legitimacy costs. Eden and Miller (2004) emphasize that the main drivers of LOF are social costs on the basis of institutional distance, which comprises cognitive, normative and regulatory distance, between ‘home’ and ‘host’ country. LOF is thereby more affected by the cognitive and normative dimensions than the regulatory. Campbell et al. (2011) emphasize in this regard that the social component of LOF refers to a lack of information by the host environment, which triggers the utilization of stereotypes and different criteria to assess foreign MNEs, which again leads to higher costs because of slower acceptance of legitimacy and lower trust levels toward the market penetration of foreign MNCs. Eden and Miller (2004) further argue that CDBA is a broader concept which incorporates LOF and economic activity-based costs. Economic activity-activity-based costs are occurring due to production, marketing, distribution, trade barriers and foreign exchange transaction, following mainly from the physical distance. These costs related to value-adding activities by the MNE can be “anticipated and measured, and may well be finite, the core issue for MNE managers remain LOF” (Eden & Miller, 2004, p. 2).

Summarizing the discussion, LOF was integrated as one major building block of CDBA, which is still the accepted view today (Beugelsdijk, 2011). With the work of Zaheer (1995) and the paper of Eden and Miller (2001) “Opening the Black Box of CDBA” a shift towards LOF as a central strategic issue for MNE Managers (Eden & Miller, 2004, 2) took place.4

The concept has been developing over time, whereby positive and multinational components have been added. I relegate here to other papers due to the focus of my research question (e.g. Sethi & Jung, 2009; Beugelsdijk, 2011; and Gaur, Kumar & Sarathy, 2011).

4

14

LOF is a key determinant of business activity in a foreign market. Distance frameworks have been developed to gauge its complexity. The next section introduces the CAGE framework and thee three distance constructs.

2.3 CAGE – Framework and Distance Constructs

In his article “Distance Still Matters – The Hard Reality of Global Expansion” Ghemawat (2001) the CAGE framework is developed. He shows that distance still matters for foreign trade and that different dimensions have an influence. He argues that, despite unifying effects due to information technology and global communication, the world does not shrink to a “small and relatively homogenous place” (Ghemawat, 2001, p. 138). Companies not taking that into account still struggle while expanding to foreign markets. Ignoring different dimensions of distance and engaging in foreign business activity by underestimating the costs of multinational business activity leads to business failures, e.g. Star TV, Wal-Mart Germany (Ghemawat, 2001).

According to Ghemawat (2001) distance can be divided into four underlying dimension: cultural, administrative, geographic, and economic. The dimensions should help to assess the influence and complexity of distance and are described below. Additionally theoretical assumptions about measurements of the Cultural Distance by Kogut and Singh (1988), the Psychic Distance Stimuli by Dow and Karunaratna’s (2006) and the Institutional measures of Berry et al. (2010) are allocated to one of the four distance dimensions in accordance to underlying theory. This is done to justify the operationalization of the CAGE framework in the second part of this paper.

The Kogut and Singh Index was chosen due to its relevance in the academia and its endorsed utilization on the country level (Hofstede, 2001), being on the same level with the research question. The other two constructs meet the latter criteria because of their operationalization on the country level. The comprehensive coverage of different distance by these constructs was another criterion. Before integrating the latter two constructs theoretically in the CAGE framework they are briefly introduced. Due to the limited scope of the paper it is not possible to introduce further distance models or frameworks here.

Psychic Distance Stimuli by Dow and Karunaratna (2006)

15

Wiedersheim-Paul, 1975, p. 308). First measures of Psychic Distance were constructed (Vahlne & Wiedersheim-Paul, 1977) and consequently certain applications found, for example predicting export, FDI market selection and choice of entry mode (Dow &Karunaratna, 2006). Nevertheless, the literature lacked measurements of the construct for a long time and discussion have been evolving its usage. One of the key discussions has been evolving about measuring perceived psychic distance of top manager prior to big decisions (Dow & Karunaratna, 2006). Therefore Dow and Karunaratna propose an approach which distinguishes between Psychic Distance Stimuli and Perceived Psychic Distance. The Psychic

Distance Stimuli refer to macro-level influences which, besides other factors which are subsumed under

the individual perception of the Psychic Distance Stimuli, determine the individual perceived psychic

distance of top managers. I use the term Psychic Distance Stimuli in this paper accordingly. Institutional Distance by Berry et al. (2010)

Berry et al. (2010) provide a database of the following distance measures capturing differences in institutions: Economic, Financial, Political, Administrative, Cultural, Demographic, Knowledge and Global Connectedness. Additionally, they use Geographic distance, as a non-institutional dimension in their calculation because it has been acknowledged in the literature to effect trade, foreign investment and “other types of economic activity taking place between countries” (Berry et al., 2010, p. 1468). Berry et al. approach distance by capturing different types of cross-national distances along multiple dimensions because decisions are made under consideration of a variety of different influence factors. The theory behind six of their dimensions is integrated in the CAGE framework below. Not all existing dimensions are included due to their irrelevance to any of the four dimensions of the CAGE – framework or unavailability of data.

2.3.1 Cultural Distance

Culture has an impact on people’s behaviour and their interaction amongst each other, with companies and with formal institutions (Ghemawat, 2001). It consists of obvious and rather subtle determinants. The first category inherits easily perceivable issues such as different languages, predominant religions or race. Gauging the second category is more arduous to conceive. The iceberg model explains this reasonably. It conveys that the bigger part of the iceberg in nature cannot be seen on the surface, the same pertains for the subtle part of culture and inherent norms. Ghemawat (p. 142) characterizes it as “the deeply rooted system of unspoken principles that guide individuals in their everyday choices and interaction, and are often nearly invisible”.

The Cultural Distance Index by Kogut and Singh (1988) can clearly be allocated to this dimension of

16

concept of psychic distance, Hymer's (1960) discussion of the disadvantages facing foreign subsidiaries, and research from an economic perspective that noticed a multitude of non-geographic factors affecting economic activity” (Mezias et al., 2002, p. 408). Mezias et al. also underline that the Kogut and Singh index links the concept of cultural distance dimensions with the concept of LOF, as shown in the previous section. It offers an indication of the potential severity of LOF companies face while expanding to another country. The Kogut and Singh Index provides data available for a magnitude of countries. Furthermore is has a high level of clarity, parsimony, and resonance with managers (Kirkman et al., 2006). It conveys further relevance for the research question due its position on the country level (Hofstede, 2001)5. It has also been utilized in abundance of academic publications. Following Kirkman et al. (2006) Hofstede’s cultural dimensions are the most influential (Sivakumar & Nakata, 2001) and most cited one (cited 1800 times through 1999; Hofstede, 2001).

As mentioned above Ghemawat regards language as one of the obvious cultural distance components. For this reason Language Distance by Dow and Karunaratna is included here. It has been one of the key measures of psychic distance from its first appearance in the academic literature until the present (Dow and Karunaratna, 2006). Welch et al. (2001) argue that companies stay mostly in their language group to minimize risk connected to language. When companies extend to foreign countries with a different language the risk of misunderstanding grows. Consequently overall transaction costs increase due to effect mitigating measures. Therefore, language differences influence the international expansion patterns of international companies (Dow and Karunaratna, 2006). Dow and Karunaratna also argue that despite its importance language has rarely been incorporated into international studies. They see the reason in the complexity of the matter and in the lack of a universal scale which measures language differences independent of a specific population.

Religious distance, compared to the other psychic distance dimensions, is not cited very often in the

literature. Nevertheless it is acknowledged to being a distinct factor relating closely to culture, attitudes and norms. Religion is one of the key determinants of conflict between cultures as seen continuously in world history. It determines behavioural characteristics of people within region or group (Down & Karunaratna, 2006). Different manifestations of religion in two countries potentially lead to higher transactions costs and a higher risk of misunderstanding between them, which potentially affect the trade intensity (Dow & Karunaratna, 2006). For these reasons Religious Distance is allocated Cultural Distance dimension within the CAGE framework.

2.3.2 Administrative Distance

This dimension, also referred to as Political Distance, implies that friendly historical and political concordances are having a positive impact on business activity. Ghemawat (2001) shows significant increases in trade in accordance to existing colonial ties, trading arrangements or political union (such as the EU). Another argument is protection of the home market by imposing competitive disadvantages to foreign companies in order to protect the home market by governments. Negative effects on foreign business activity are for example triggered by weak institutional infrastructures. In this context it is distinguished between the level and the quality of formal institutions, meaning existing laws and the

5

17

political system and their enforcement. Findings show a negative impact of low developed institutions on business activity in the respective country (Ghemawat, 2001). Following this specification an allocation of the following distance measurements of Berry et al. and Dow and Karunaratna is conducted.

Political System by Dow and Karunaratna offers accounts for the communication in either direction

between business and government or vice versa. Companies confronted with a different type of government in a host country carry a bigger risk. They experience a rise in costs due to the foreign government’s potential enforcement of distinct rules between businesses different to the ones endorsed in the home country government of the foreign company. Examples are the enforcement of contracts or the continuous observation of anti-competitive behaviour (Dow & Karunarathna, 2006, p. 583). Therefore, the risk of foreign firms engaged in business abroad are higher and trigger more costs due to uncertain behaviour of host companies and host governments.

Political Distance, by Berry et al., refers to a variety of scholars (e.g. Henisz, 2000) proofing that

countries are distinguishable in accordance to their politics. Berry et al. (2010) characterize countries along continuous political dimensions. In this context political differences are regarded to influence foreign market selection, the type of entry mode and flows of direct investments (Berry et al., 2010). These variables correlate “with the choice of foreign markets to enter, the choice of entry mode and foreign direct investment flows”

2.3.3 Geographic Distance

Basically, the Geographic Distance dimension exists in conjunction with the concept of ‘the farther the harder’ (Ghemawat, 2001). Geographic Distance between countries as a wider concept is a key determinant of cross-national distance. Following Ghemawat (2001) the pure spatial distance in kilometres is thereby not the only relevant one. Accompanied by other means of distance such as the average within-country distances to borders, access to waterways and the ocean, and determinants created by men, such as country’s transportation and communication infrastructure which are highly important for cross-economic activity, make the Geographic Distance dimension complex. Ghemawat underlines that the effect influences foreign business activity independent of its type (i.e. trade or FDI).

Geographic Distance by Berry et al. was the only dimension of Geographic Distance existing in the three

chosen distance constructs. Research has thereby shown (Anderson, 1979; Deadorff, 1988) that physical distance influences, for example, foreign trade and investment patterns and “other types of economic activity taking place between countries” due to the fact that greater distance means higher costs of transport and communication (Berry et. al, 2010, p. 1468).

2.3.4 Economic Distance

18

significance. It shows also the existence of various approaches relating to the same underlying dimension.

I allocate the following distance constructs of Berry et al. (2010) to this dimension: Economic, Financial, Demographic and Global Connectedness.

Economic Distance influences the purchasing pattern of consumers. Also countries’ macroeconomic

stability is considerable related. Different economic systems also trigger constraints about the openness to external influences from one country to another.

Financial Distance refers also to economic distance due to different financial systems in countries, which

are connected to country specific ways for companies to “fund/capitalize” their projects. This dimension has been used in “cross-national financial system literature” (Dow and Karunaratna, 2006, p. 1467). The developed world inherits a common financial system, meaning a relatively equal wealth or income throughout developed nations. This leads to the allocation of this measurement to the economic distance between countries.

Demographic Distance is a rather subtle characteristic of a country’s economy. It characterises a

countries’ population taking into account information about consumers and related relevant market insights. Countries on the same economic level exhibit the similar demographic patterns. This type of data has been used by researches to investigate international corporate expansion and share price (Berry et al., 2010).

Global Connectedness Distance is also allocated to the Economic Distance dimension. It “captures the

ability of resident individuals and companies to interact with other parts of the world, obtain information, and diffuse their own activities” (Berry et al, 2001, 1468). This ability is similar in countries on a comparable economic level.

Two determinants of Dow and Karunaratna can also be attributed to the Economic Distance of Ghemawat: Industrial Development and Educational Distance.

Educational Distance by Dow and Karunaratna has been part of the underlying psychic distance over the

past 25 years and is a fundamental factor in the academic discourse (Dow & Karunaratna, 2006). Following this dimension, people’s proper understanding of information and their corresponding reaction depends on the education level in their home-country. There are enormous variations of education levels between countries. Greater differences trigger higher risks of misunderstanding, erroneous decisions and consequently raise the overall risk level of cross-border business activity between cross-national partners. I allocate education to the economic category because the average educational level of a country is related to its economic development and the wealth or income of people within these countries.

Industrial Development according to Dow and Karunarathna (2006) clearly matches this distance

19

studies. The most striking individual level difference in industrial development between countries is the personal employment. That means that in dependence of their employment people understand and react to information in a certain way, potentially causing misunderstandings in cross-border activities. Furthermore, the level of business to business communication depends (amongst other) on the industrial development of a country. This is mainly due to false interpreted communication of businesses originating from different industrial developed countries. For example a company with heritage from an agrarian economy might encounter problems doing business with companies from more developed economies and vice versa. Therefore the costs connected to a business activity abroad increase if a high industrial distance exists. This influences consequently also the location decision of international companies (Dow & Karunaratna, 2006).

The following literature review about cross-national distance research assesses research and findings in the literature and provides in conjunction with a thorough review of three top business journals contemporary and comprehensive insights into the academic field.

2.4 Literature review about cross-national distance research

Campbell et al. (2011) provide a brief literature overview about several distance dimensions which are relevant in this master thesis. It comprises the concept of psychic distance (e.g. Ellis, 2008; Evans & Mavondo, 2002; Johanson & Vahlne, 1977; O’Grady & Lane, 1996), cultural distance (e.g. Brouthers & Brouthers, 2001; Dikova, Rao Sahib, & van Witteloostuijn, 2010; Kogut & Singh, 1988; Morosine, Shae & Singh, 1998; Shenkar, 2001; Tihanyi, Grifith, & Russel, 2005), institutional distance (e.g., Berry, Guillen, & Zhou, 2010; Eden & Miller, 2004; Xu & Shenkar, 2002), and geographic distance (e.g. Egger & Pfaffermayr, 2004; Li & Vashchilko, 2010). Campbell et al. (2011) point out that distance is acknowledged as a multidimensional construct. That means the underlying distance concept between home- and host country in IB literature refers to many dimensions capturing many different aspects. Campbell et al. (2011) state that distance in general but especially institutional and geographic distance, has been found in academic research to increase Liability of Foreignness (LOF) by provoking higher unfamiliarity and discriminatory hazards for foreign firms doing business in host countries (Eden & Miller, 2004; Kostova & Zaheer, 1999).

20

cultural sensitive pays (i.e., with higher returns on assets, sales, and higher bonuses)” (p. 302).This goes also in line with the statement in the IB literature that it is generally approved that with growing distance success of business activity in a foreign country is more difficult (Beugelsdijk, 2011).

Another meta-analysis about cross-national distance was conducted by Tihanyi, Grifith and Craig (2005) who examined the effect of cultural distance on three key variables of international MNE activity: entry mode choice, international diversification, and performance. They state, after analysing 66 independent samples distributed to 55 articles, while culture “is an important determinant of organizational actions and performance, both empirical and theoretical concerns abound” (p. 270). In other words cultural distance is not fundamentally proved in the existing literature to be directly related to any of the three variables. A further finding is that the adoption to foreign culture, which is channelled by nations’ political economy, education, religion, and language, potentially creates concerns for MNEs in foreign markets (Schwartz, 1999). Tihanyi et al. also restate the proposition by Ricks et al. (1990) that the construct of cultural distance has gained respectable interest in IB research. In this context Shenkar and Li (2008) assert that most studies engaging in the field of cultural distance concentrate on the entry mode decision of MNEs in foreign markets (e.g. Barkema et al., 1997; Li & Guisinger, 1991). Nevertheless, the construct of cultural distance has potential limitations which question findings of studies in the field (Brouthers & Brouthers, 2001; Evans and Mavondo, 2002, Shenkar, 2001, McSweeney, 2002).

Tsui, Nifadkar and Ou (2007) review cross-national research between 1995 and 2005. They identified and analysed 93 empirical studies about organizational behaviour with culture as a dependent (explanatory) variable, published in 16 leading management journals. They refer to the 21st century as the century of international management research due to the acceleration of global business. This is also emphasized by Kirkman and Law (2005) who found, in their analyses of published research in the AMJ between 1970 and 2004, that since 2000 the proportion of international business research increased implying a growing relevance of the research field. Tsui et al. (2007) see a special need for consolidation of different cultural frameworks and their measurements. They also call for more implementation of various distance measures toward a multidimensional approach, which is followed by this study.

Sousa and Bradley (2008) affirm that findings about cultural distance and psychic distance have been mixed. Some studies have found significant influence on the foreign business activity of firms (e.g. Barkema & Vermeulen, 1997; Evans &Mavondo, 2002; Pak and Park, 2004; Sousa & Bradely, 2005), whereby others could not identify an influence (e.g. Stoettinger & Schlegelmich, 1998; Mitra & Golder, 2002; Sethi et al., 2003). Also Ellis (2008) refers to (empirical) studies about Psychic Distance as having inconclusive results. He states that some scholars identify and influence for Psychic Distance (Dow, 2000; Johanson & Wiedersheim-Paul, 1975) while others do not (Benito & Gripsrud, 1992; Engwall & Wallenstal, 1988). Therefore claims arise that “the construct of Psychic Distance has been misused (Child, Ng, & Wong, 2002), miss-measured (Evans &Mavondo, 2002) or not properly tested at all (Dow, 2000)” (Ellis, 2008, p. 351).

21

the research question. The reason is the clear distinction of Dow and Karunaratna’s approach dividing Psychic Distance in Psychic Distance Stimuli and Perceived Psychic Distance.

Sousa and Bradley (2008) also emphasize that distance has a negative influence on MNEs’ performance because it is assumed that similarities are easier manageable than differences. They name a number of studies proofing that firms have a better performance in countries which show a smaller distance to each other (e.g. Joahanson & Wiedersheim-Paul, 1975; Johanson & Vahlne, 1977; Lee, 1988; Li et al., 2002). Psychic distance and cultural distance have also been used to explain FDI patterns. Companies chose to invest in rather close countries, in terms of distance dimensions, before they increase activity in more distant countries. This follows the idea of the Uppsala internationalization model (Joahanson & Wiedersheim-Paul, 1975; Johanson & Vahlne, 1977) which postulates a development of the type of market penetration for MNEs over time, engaging close countries at the beginning and more distant countries at later stages of internationalization.

The emphasis in Sousa and Bradley’s (2008) article lies in the refinements and conceptualization of cultural distance and psychic distance. Following them (and due to empirical support by Sousa and Bradely, 2006) it is inappropriate to use the terms psychic distance and cultural distance interchangeably such as done in various studies (e.g. Kogut & Singh, 1988; Shoham & Albaum, 1995; Fletcher & Bohn, 1988; Eriksson et al., 2000; Sethi et al., 2003). The construct of Psychic Distance is assessed and used on the country level in this paper research and is not used interchangeable with the term cultural distance.

The last part of the literature reviews consists of a brief summary of findings, identified in articles of the examined journals, which have not been included in this section so far but meet the prerequisites of my literature search. Luo (2005) considers the Kogut and Singh-Index (1988), implementing all five cultural dimensions of Hofstede (2001), in his research. He includes cultural distance researching the perception of procedural justice within international alliances. Cultural distance as a moderator of shared procedural justice becomes thereby more important to profitability when cultural distance is at a high level. Tsang and Yip (2007) used economic distance to assess its relationship to hazard rates of FDIs. Their findings prove that cultural distance increases the hazard rates of FDIs and therefore back up findings by Barkema et al. (1996). Yu and Cannella (2007) utilize geographic distance to make assertions towards the speed of a MNEs response in a foreign market in consequence of a rival’s action (business activity of its affiliate) on site. They conclude thereby that an influence of a firm’s geographic distances to the respective host market and response speed exists. Hutzschenreuter and Voll (2008) state a negative relationship between MNE performance and added level of cultural distance6. They find also that an irregular expansion rhythm to other countries, meaning MNEs expand to countries which are culturally rather far apart from the prior country they expanded to, has a negative influence on the performance of the firm due cross national cultural distance. Lee, Shenkar and Li (2008) utilize cultural distance and the direction of investment flow as independent variables. They assess their relationship to the extent of control modes MNEs put in place abroad. Results show that cultural distance alone is not

6

22

connected to a preference of higher control mode but when incorporating the direction of investment flow as a moderator cultural distance has an influence on the control mode of firms. Reus and Lamont (2009) use cultural distance, amongst other independent variables, to assess its influence on acquisition performance. They emphasize that the effect of cultural distance is dependent on increasing understanding and the maintenance of communication within the MNE. The last finding included in the literature review refers to the article of Yu and Cannella (2009) who use cultural distance to assess the competitive aggressiveness of MNE in foreign markets. They prove that a greater cultural distance between an MNE’s home country and a host country triggers weaker rivalry-dampening influence of multimarket contact with competing companies. This consequently leads to greater competitive aggressiveness of their subsidiaries in the respective country market.

More recent articles relating to empirical cross national distance research in the three reviewed journals in the period 2005-2011, were expected. Firstly because statements of Tsui et al. (2007) and Kirkman and Law (2006) referring to the 21st century as the century of international business and management research. Secondly the identification of 24 articles employing empirical measure of psychic or cultural distance by Dow and Karunaratna (2006) in five years prior to their publication in the JIBS alone. I could identify 14 articles meeting my premises in the defined period of 2005-2011. I assume a reason for this relatively small number lies in the choice of the reviewed journals. Other journals more focused on international business and management research publish probably more articles in the research field. Notwithstanding I defend the choice of journals because it gives a clear picture of the contemporary status of the research topic. Furthermore the ratio is in line with the findings of Kirkman, Lowe and Gibson (JIBS, 2006), Tsui, Nifadkar and Ou (2007) and Pisani (2009) about the ratio of this research type in the entire business and management field.

The literature review made the relevance of cross-national distance research in IB apparent, cultural distance was thereby identified to be especially utilized. It was also shown that inconclusive findings in the field exist (e.g. Sousa & Bradely, 2008; Ellis, 2008). Nevertheless, a general accepted negative influence of cross-national distance on foreign business activity can be acknowledged.

Following the theory and findings of research communicated so far the subsequent section introduces the hypotheses of this paper.

2.5 Hypotheses Development

The theoretical relations between LOF, Distance and MNE’s foreign business activity are depicted in Figure 1 below. LOF presenting a negative influence on MNE’s affiliates’ business activity in host countries assessed by overarching distance dimensions between the home country of the MNE and the host country in which the MNE invested.

23

Following the information provided so far in the theory section about LOF, the CAGE framework and the general acknowledged view in the literature about a neglecting effect of distance on foreign business, the hypotheses are develop. Thereby a negative influence of distance on MNE’s affiliates’ foreign business activity can be assumed. Hence my hypotheses about the influence of cross national distance dimensions on VA and VApE are established as follows:

Hypothesis 1: Distance between the home and host country has a negative influence on the amount of value added being created by US MNCs.

Hypothesis 2: Distance between the home and host country has a negative influence on the amount of value added per capita being created by US MNCs.

3. Methodology

3.1 Data Description and Calculation Approach

I decided to use panel data, also known as longitudinal data, to answer my research question because it is characterized as having both cross sectional and time series dimensions. Dougherty (2007, p. 402) points out its complexity, but also that it is “increasingly used in applied work” for the following reasons:

It offers a “solution to the problem of bias caused by unobserved heterogeneity, a common problem in the fitting of models with cross-sectional data sets”.

It “may be possible to exploit panel data sets to reveal dynamics that are difficult to detect with cross-sectional data”.

Another advantage is the analysable magnitude of observations in panel data sets.

To answer the research question I created a comprehensive dataset. Due to its size the dataset cannot be included in the Appendix7. The time series data between 1997 and 2008 about Value Added and

Value Added per Capita used for the panel analysis was obtained from the U.S. Bureau of Economic

Analyses (BEA) and its database U.S. Direct Investment Abroad: Financial and operating data. As stated

7

24

before, the data used in the analyses only relate to Majority-Owned Foreign Affiliates. The database aggregates financial and operating data on all U.S. parents and their foreign affiliates on a yearly basis whereby participation in terms of information sharing of the companies is compulsory (BEA, 2011).

Country Only (all countries) data was selected to obtain an appropriate and considerable dataset to

engage the research question. The independent and control variables used for the regression analyses were gathered by utilizing a variety of online databases which are specified in the respective categories below. The final dataset consists of 118 countries. Due to a lack of data in specific categories for certain countries only 52 could be implemented in the panel data analyses. The data of these countries exhibit availability in almost all required variables creating a panel with 608 observations of determinants over the time period 1997-2008. With representing 44 % of all countries the dataset is fairly representative. Such a level “provides a reasonable good level of justifying a meaningful” (Gachino, 2007, p. 14) assessment whether distance has an impact on VA and VApE created by U.S. MNEs foreign affiliates in host countries.

Before conducting the panel analysis I make use of factor analysis including all independent measures introduced. A factor analysis is a statistical technique for data reduction which is often used in research. I implement this measure to identify key composite measures for the distance dimensions of the CAGE framework8. In other words different operationalizations of the same concept are ruled out (William, 2011). With this procedure overlapping effects of distance measures are minimized and that imparts clear interpretations at the end. The resulting factors are created in accordance to the CAGE-framework; meaning for each distance dimension one factor is established. The factors evolve out of the measurements of distance introduced so far. A distinct procedure of factor analysis in Stata is chosen to standardise the respective outcome.

I conduct my panel analysis using the “Stata Longitudinal-Data/Panel Data Reference Manual – Release 11”, the training paper of Princeton University “Panel Data Analysis – Fixed and Random Effects (using Stata 10.x)” and the paper of Gachino (2007) “Foreign Direct Investment and Firm Level Productivity A Panel Data Analysis” as guidelines.

I decided to conduct my panel data analyses with the GLS (generalised least squares) - Random Effect Model (GLS-Random effects model using Stata11.x.). The Random Effect model allows including time invariant variables, which are included in the panel analyses.

The prerequirement for using the Random effects model is that “variation across entities is assumed to be random and uncorrelated with the predictor or independent variables included in the model” (Princeton.edu, 2011). “Random effect models can be estimated based on the maximum likelihood (ML) or GLS” (Gachino, 2007, p. 17). I make use of the latter one and perform my calculation under the

8 _{A factor analysis “reduces the number of variables in an analysis by describing linear combinations of the}

25

assumption of heteroscedasticity and no autocorrelation, as the latter one was proven by the Wooldridge Test for autocorrelation in panel data9_.

Below the two dependent variables, value added (VA) and value added per employee (VApE), of the panel analysis, using the GLS-random effects model, are described. The subsequent part of the methodology section introduces measurements of the independent variables, used for the factor analysis, and control variables.

3.2 Dependent Variables

Value Added (VA):

With value added as a dependent variable the first of the two panel analyses is realized. The data used for value added was obtained from the BEA database and is described as “the portion of the goods and services sold or added to inventory or fixed investment by a firm that reflects the production of the firm itself. It represents the firm’s contribution to a country’s gross domestic product” (Definition from BEA, 2011). Compared to firm’s sales it displays the firm’s own production in a host country, excluding activity from intermediaries, and not the overall sales of all products which potentially are created part wise in different countries. The data is available at the BEA database and is calculated as the “sum of the costs incurred (except for intermediate inputs) and the profits earned in production” (BEA, 2011).Another advantage from value added is that it determines the value-adding activity in a specific time whereby for example some parts of sales data might result from production in a different period of time (Beugelsdijk et al., 2010, 1448).

Value Added per Employee (VApE):

The second panel analysis is conducted with VApE as a dependent variable. VApE is the value added created by each employee on the pay roll of U.S. foreign affiliates. Consequently it measures the productivity of MNE’s affiliate’s business activity in a foreign host country and provides an indication to the type of work conducted in the foreign market in reliance to cross-national distance dimensions. The data was computed by dividing value added by the number of employees of majority-owned foreign

affiliates of U.S.MNCs. The number of employees was gathered by using the BEA database. It represents

the number of full-time and part-time employees at the end of the year. In the case of an abnormal high or low number of employees the number available in the database reflects the amount of employees during normal operation or a yearly average (BEA, 2011).

By inspection of the scatterplots I identified that that both variables are skewed. Following Mickey et al. (2004) I chose to transform my dependent variable data by using LOGe (LN) transformation. To use the

natural logarithm is a standard practice in economic research for skewed variables (e.g. Tsang & Yip, 2007).

9

26

3.3 Independent Variables

Different distance dimension measurements are briefly described below. Cultural distance of Kogut and Singh (1988) was calculated by me following the cultural distance formula shown below and the openly available data of Hofstede’s distance dimensions. The calculation is included in the Dataset file accompanying this master thesis for reference10. For a minute description of the other dimensions I refer to the corresponding articles of Dow and Karunaratna (2006) and Berry et al. (2010).

Cultural Distance (Kogut & Singh, 1988)

I use only the firstly established four dimensions of Hofstede for my calculation because data about Long-Term Orientation, as the youngest dimension, is not available for a multitude of countries and would have reduced the number of observations utilizable for my analyses significantly. Below the Euclidean distance formula of Kogut and Singh (1988) is elucidated.

„CD is defined to be the cultural distance of the jth country from the U.S. I indicates Hofstede's scores on each of the four dimensions for each country j investing in the United States, u. The difference between the jth country and the U.S. on each of the i dimensions is squared. The V in the denominator indicates that the squared difference between the jth country and the U.S. on each dimension is deflated by the variance of that dimension. These four squared differences weighted by the inverse of their variances are then summed and divided by four to yield an equally weighted average; this is the measure of cultural distance” (Mezias, 2002, p. 409).

Psychic Distance Stimuli Dow and Karunaratna (2006)

Language. They measure language using a new approach consisting of three determinants. Firstly

developing a five point scale which concentrates on key differences between the languages of two countries. This captures that some groups of languages are closer to each other which allows a hierarchy classification. The second and third determinants “focus on the reported incidence of one country’s major language(s) within the other countries” (Dow &Karunaratna, 2006, 585). The data is directly downloaded from the Database of Dow and Karunaratna (2006)

Religion. The calculation of the religious distance follows the language distance approach. They

implement three indicators to measure the religion dimensions. The first one is a five dimensional scale which determines difference in religion between two countries. Both the second and third dimensions “focus on the reported incidence of the originating country’s dominant religion within the receiving country, and vice versa” (Dow and Karunaratna, 2006, p. 586). The data was obtained from Barett (1982).

10

27

Political System. Dow and Karunarathna (2006) choose two dimensions to measure the political

distance between two countries. The first one is Democratic Political System Distance representing the degree of democracy or political freedom between countries. It is computed by using the POLCON scale11 (Henisz, 2000), the POLITY IV12 instrument (Gleditsch, 2003) and the Freedom House’s (2000) Political Rights and Civil Liberties scale between countries. The Social Political System Distance measuring differences in political systems is the difference in Beck’s Political Ideology scale between countries and uses Beck’s (2001) Right-Centre-Left scale, which measures the measures the “policy preferences or ideological leanings of the decision-makers” (Beck et al., 2001, p. 13). In consequence of the argumentation and analyses of Dow & Karunarathna (2006) I use the absolute values of these dimensions in the following analyses.

Industrial Development. Dow and Karunarathna (2006) calculate the level of industrial difference by

using ten different indicators. These indicators are the difference in GDP, the differences in the consumption of energy, vehicle ownership, the percentage of employment in agriculture, the percentage of GDP from manufacturing, the difference in the degree of urbanization and the development of the communication infrastructures by distinguishing between newspapers, radios, telephones and televisions per 1000 population. These data are available in the United Nations data (1995 a, b).

Education. The data used by Dow and Karunaratna (2006) to compute an educational distance between

two countries are the difference in literacy rate between countries and the differences in the proportion of the population enrolled in second and third-level education respectively, after adjusting for the proportion of the population under the age of 15. Data were obtained by Dow and Karunaratna (2006) from the Statistical Yearbook and the Social Indicator Data from the United Nations database.

Distance measures of Berry et al. (2010)

For the calculation method of the dataset I relegate to the article of Berry et al. (2010). The data about administrative distance is not available online and can therefore not be included as an independent variable. Berry et al. (2010) also provide a scale for cultural distance. This measure can also not be included in the panel analyses due to its insufficient availability.

Political Distance. Measurements of institutional checks and balances, democratic character, the size of

the state in relation to the economy, and external trade association are considered here. The data was established by using POLCONV, Freedom House, WDI and the World Trading Organization (WTO). Geographic Distance. Geographic Distance represents the Euclidean great circle distance (in kilometres) between the geographic centre of countries, which means here between Washington, DC, and the respective capital of the host market (Slangen & Beugelsdijk, 2010) in which the majority-owned foreign

affiliates of US MNCs are active. I include the LN of the geographic distance in kilometres in the

calculation, following utilization in the academia (e.g. Yu & Cannella, 2007).

28

Economic Distance. Countries distinguish themselves by the level of income (GDP per capita/person),

the existing inflation rates, and the trade intensity with other countries, which in other words refers to exports and imports of goods and service as a percentage of the gross domestic product (GDP) (Berry et al. (2010). Data about these factors was obtained by Berry et al. (2010) from the Word Development Indicator (WDI) database.

Financial Distance. The financial distance index includes as percentage of GDP, the market capitalization

of listed companies, the number of listed companies, and the amount of private credit. The data was gathered by using the WDI database.

Global Connectedness Distance. This dimension is computed by using as a percentage of GDP measures of international tourism expenditures and international tourism receipts. Also included in this measure are internet users as a percentage of the population of a country. Data was gathered by Berry et al. from the WDI database.

Demographic Distance. Berry et al. (2010) establish their measure by using life expectancy rates, birth rates and the age structure of the population. The data was gathered from the WDI database.

3.4 Control Variables

I control for a variety of other factors. Data for the variables described below was collected for 12 years, from 1997 to 2008. I also control for time dynamics in my data by implementing a dummy variable for each year but 2008. The dummy variable for 2008 is not included because the entire data would then be explained and one of the variables omitted because of collinearity, as already broadly described in 3.1 Data and Sample.

Host Market Size (population). The population is relevant as a control variable to rule out patterns of

majority-owned foreign affiliates of US MNCs related to the mere population size of a host country and

not to institutional distance measures per se. The Host Market’s population size for each year was obtained from the World Development Indicators (WDI). I also use the LN of the Host Market Size (population) following the argument of Mickey (1984). Dividing the most populated country (China, 2008) by the country with the smallest population (Lichtenstein, 1997) the result is with 41774 bigger than two and a LN-transformation is reasonable.

Annual GDP per capita growth. Data was also obtained from the WDI. It is included in the calculation to control for influences of a country specific abnormal economic growth or shrinkage on the dependent variables, value added and value added per capita.

Natural-resource abundance. The natural resource abundance is computed by adding up host market’s fuel export (as percentage of merchandise exports) and Ores and Metal export (as percentage of merchandise exports). This determinant is included to rule out influence of resource endowments. The WDI database was used to obtain the data.

29

the countries inward FDI stock as a percentage of its GDP (Habib & Zurawicki, 2002; Kumar, 1994). Inward FDI represents “the net inflows (net investment inflows less disinvestment) in the reporting economy from foreign investors.” (WDI, 2011)

GNI per Capita. The WDI database provides the GNI per capita (Atlas method) of a multitude of countries over the relevant time period for the panel analyses of this paper (1997-2008) to control for economic development of host-countries. It is calculated by dividing the gross national income by the country’s midyear population. Prior to the calculation the gross national income is converted to US-Dollars by using the Atlas Method. This Method is used by the World Bank to smooth “fluctuations in prices” by utilizing a conversion factor “that averages the exchange rate for a given year and the two preceding years, adjusted for differences in rates of inflation between the country” (WDI, 2011, Series long definition) and, since 2001, the Euro area, Japan, the United-Kingdom and the United-States (WDI, 2011, Series long definition). I make use of four dummy variables in my analyses to account for the four different levels of GNI per Capita whereby ‘1’ stands for being a member of a group and ‘0’ for not being a member. Logically one country can only be accounted to one income category per year. The dummy variables distinguishes between low income ($1005 or less), low middle income ($1006-3975), upper middle income ($3976-12275) and high income ($12276 or more) for the relevant time period (1997-2008) for the host countries. The income classes were chosen following the classification of The World Bank (2011)13. As also done for the control of time dynamics, I have to exclude one of the dummy variables in the panel analysis. Otherwise the entire data would be explained and one of the variables omitted because of collinearity.