The influence of proximity on Multinational Enterprise location choice : a multiple case study in Dutch science parks

(1)

Enterprise Location Choice

A Multiple Case Study in Dutch Science Parks

(2)

1 The Influence of Proximity on Multinational Enterprise Location Choice

A Multiple Case Study in Dutch Science Parks Master Thesis

MSc. Business Administration – International Management University of Amsterdam – Amsterdam Business School

Abstract

Foreign direct investments have a significant impact on the location at which they are aimed in terms of workforce skills, employment, productivity and knowledge transfer (Monaghan, 2012). Multinational Enterprises (MNEs) are thus important for knowledge intensive-clusters such as science parks to attract in order to boost innovation and keep up in the global knowledge economy. This study sets out to research the role of proximity with respect to the location choice of knowledge-intensive MNEs. By doing so, it tries to help managers and academics to better understand the role of proximity on knowledge processes within subnational locations and especially with respect to MNEs. This research builds on the concept of proximity as put forward by Boschma (2005), using the knowledge-based view of the firm (Caballero & Jaffe, 1993; Grant, 1996) as a conceptual foundation describing the mechanisms that drive knowledge-intensive MNEs. The study is conducted using a qualitative multiple case study research design with embedded unit of analysis. Data is collected through semi-structured interviews matched by management role (MNE, science park, university, and knowledge institute managers) in the three major science parks of the Netherlands (viz. Amsterdam, Leiden, Utrecht). The findings suggest that there is almost no subnational variation in emphasis on the nature of proximity in these science parks. However, this study has found a couple of interesting results with respect to the role of proximity on knowledge transfer and knowledge sourcing processes. Most notably, low organizational proximity (instead of high organizational proximity as predicted by Boschma (2005)) has been shown to be an explanatory factor of firms collaborating and/or sharing knowledge. This also contradicts the theory of Mariotti et al. (2010) stating that large MNEs tend not to agglomerate with small domestic firms.

Keywords: Science park, proximity, location choice, knowledge-based view (KBV).

Student: Niels le Duc Student ID: 6050611

Supervisor: Dr Johan Lindeque Second reader: Dr Lori Divito

(3)

Statement of Originality

This document is written by student Niels le Duc who declares to take full responsibility

for the contents of this document.

I declare that the text and the work presented in this document is original and that no

sources other than those mentioned in the text and its references have been used in creating

it.

The Faculty of Economics and Business is responsible solely for the supervision of

completion of the work, not for the contents.

(4)

3

Acknowledgements

Although only my name appears on the cover of this study, a great many people have contributed to its production. I would like to use this space to express my gratitude to all of them.

First of all, I would like to thank my supervisor Dr Johan Lindeque for his guidance, constructive feedback and valuable insights. His enthusiasm and expertise have significantly strengthened this study and I believe this input also contributed to making the course ‘International Strategy’ that he taught the best course of the entire master. With his move to Switzerland, the Amsterdam Business School and its students have lost one of their most valued lecturers.

Furthermore, I would like to express a sincere thank you to the eighteen men and women who so graciously agreed to participate in my study. Without your cooperation this study could not have been completed. In addition, I found speaking to all of you one of the greatest joys of writing this thesis (next to a heap of unpleasant experiences, let’s be honest).

I would also like to thank Mirjam Neervoort for taking the time to proofread my work. Her being a native English speaker dispelled any concerns about grammar issues and about having sentences written in ‘Dunglish’.

Last but certainly not least I would like to thank my parents for their support. Not only during this master thesis, or the one before this one, but for their support during all of my six and a half years at the University of Amsterdam. You were my sounding board, proofread my papers, printed articles and financially supported me all that time. You have given me a head start in this world by doing so, thank you.

Niels le Duc

(5)

Index of Tables and Figures

Figure 1: Multiple case study with embedded units of analysis

p. 21

Table 1: Overview of respondents p. 24

Table 2: Matrix interview questions p. 26

Table 3: Coding scheme p. 27

Table 4: Analytical table for case 1 – Amsterdam Science Park p. 31 Table 5: Analytical table for case 2 – Leiden Bio Science Park p. 35 Table 6: Analytical table for case 3 – Utrecht Science Park p. 39

Table 7: Analytical table for cross-case analysis p. 44

(8)

7

1. Introduction

Up until 1963 cows grazed on the meadows just east of Amsterdam. Today, this land is part of the city and the cows have made room for the Faculty of Science of the University of Amsterdam and numerous knowledge institutes and businesses. The site has been turned into a ‘science park’, a “property development aimed at supporting research-based commercial activity” (Quintas et al., 1992: 161). A more specific definition of this geographic clustering comes from the International Association of Science Parks (IASP):

‘’A science park is an organization managed by specialised professionals, whose main aim is

to increase the wealth of its community by promoting the culture of innovation and the competitiveness of its associated businesses and knowledge-based institutions. To enable these goals to be met, a science park stimulates and manages the flow of knowledge and technology amongst universities, R&D institutions, companies and markets; it facilitates the creation and growth of innovation-based companies through incubation and spin-off processes; and provides other value-added services together with high quality space and facilities’’

(IASP, 2002).

Amsterdam is by no means unique with its science park. The first science park was established in 1950 in Stanford, California, and many European countries subsequently started establishing science parks in the 1980s and 1990s (Storey & Tether, 1998). Today, there are over 400 science parks around the world and their number is still growing (UNESCO, 2015).

It could be argued that science parks are (in a way) a result of our knowledge-based economy (Uppenberg, 2009; Dunning, 2000). In our knowledge-based economy the development of knowledge leads to innovation (new processes, services and products) that in turn drives economic progress (OECD, 1996). Of course knowledge and innovation have always been important, but nowadays innovations follow each other in rapid succession (Rutten, 2003). This is due to the process of ‘’creative destruction’’ as put forward by Joseph Schumpeter (1975). According to Schumpeter (1975), knowledge-intensive activities bring about something new that causes the demise of whatever existed before. This is the reason why, for example in the United States, more than half of the economic growth in 1999 came from activities that did not (or scarcely) existed only ten years earlier (Hospers, 2003). Consequently, no firm, country, region or city can be indifferent to the consequences of the process of ‘’creative destruction’’. One way to deal with these consequences is the establishment of science parks. The assertion is that the combination of institutions, clusters of science-based knowledge and a skilled labour pool encourages innovation (Castells & Hall, 1994; Ylinenpää, 2001). Available evidence shows that global R&D foreign direct investments (FDI) have grown significantly in recent years and

(9)

are estimated to continue to grow in the future (UNCTAD, 2005). This is a positive development for locations such as science parks, since foreign direct investments have a significant impact on the location at which they are aimed in terms of workforce skills, employment, productivity and knowledge transfer (Monaghan, 2012). Multinational Enterprises (MNEs) are thus important for science parks to attract in order to boost innovation and keep up in the global knowledge economy.

The science parks in the Netherlands use different labels to describe themselves. For example: the Amsterdam Science Park is home to companies and institutes in the life sciences and IT sectors, the Leiden Bio Science Park mainly accommodates biological-based companies and the science park in Eindhoven focuses on engineering science and technology. It seems logical that firms that depend on innovation and technological change locate in a science park that ensures the best possible knowledge externalities and spillovers and the international business (IB) literature also increasingly argues that these subnational levels of variation really matter to IB strategy location choices (Boschma et al., 2015; Dunning, 1998). In addition to the knowledge spillover argument, the IB literature has also looked at the influence of the cost and availability of premises (Sternberg, 1990), the availability of skilled labour (Oakey, 1981), the distance to major airports, ports, large (capital) cities and local facilities (infrastructure) (Aelen, 2003), and local government incentives (Dunning: 1998) on the subnational location choice of firms.

However, the IB literature has not yet focussed on the role of proximity with respect to the location choice of MNEs. Proximity has been conceptualized by economic geographers (EG) as distance-related subnational variation (Boschma, 2005). Geographical proximity (spatial clustering) is but one dimension among a number of proximity dimensions that can explain interaction between actors (Broekel & Boschma, 2012). Other dimensions are institutional, social, cognitive and organizational proximity (Boschma, 2005). MNEs may choose to be more or less proximate to different subnational locations in order to obtain certain knowledge flow outcomes. Thus, the concept may help managers and academics to better understand the role of co-location (or not) on knowledge processes within subnational locations such as science parks and especially with respect to MNEs. By integrating the EG concept of proximity into the IB literature, this study also responds to the call in the literature to increasingly do interdisciplinary work (Cantwell & Brannen, 2011).

This study will focus on how science park management, MNE management and the management of scientific institutions located in the three major science parks of the Netherlands (viz. Amsterdam, Leiden and Utrecht) understand the role of proximity with respect to MNE location choice.

(10)

9 The research question guiding this study is as follows:

How does the subnational variation in emphasis on the nature of proximity in major Dutch science parks affect knowledge intensive MNEs’ location choice?

In order to formulate an answer to this question, this paper starts off with an overview of the relevant literature on the knowledge-base view (KBV), knowledge flows, location choice and proximity. This overview results in the formulation of a number of working propositions that reflect four dimensions of proximity. Following the theoretical foundation, the methodology of this study will be discussed. This discussion includes the research philosophy behind this study, the selection of cases, the collection of data and the method used to analyse the data. This study is conducted using a qualitative multiple case study research design with embedded unit of analysis. Data is collected through semi-structured interviews with MNE, science park, university, and knowledge institute managers in all three science parks. The within-case analysis and cross-case analysis that follow the methodology chapter are used to evaluate the validity of the working propositions. The paper concludes with a summary of the key findings, the limitations of this research, managerial implications and suggestions for future research. While the findings suggest that there is almost no subnational variation in emphasis on the nature of proximity in the major Dutch science parks. The study has found interesting results with respect to the role of proximity on knowledge transfer and knowledge sourcing processes, be it sometimes in unexpected ways.

(11)

2. Conceptual foundation

The concept of the knowledge-based economy, the related process of ‘creative destruction’ (Schumpeter, 1975) and the agglomeration of knowledge in science parks (Castells & Hall, 1994) increases the importance of the knowledge-based view (KBV) of the firm (Caballero & Jaffe, 1993; Grant, 1996). The first part of this chapter therefore discusses the knowledge-based view of the multinational enterprise (MNE) (Grant, 2002; Kogut & Zander, 1993). Next, theories from the traditional international business (IB) literature concerning MNE (knowledge-based) location choice are explained (Zaheer, 1995; Beugelsdijk, 2011). Subsequently, I will discuss the newly emerged subnational stream in the IB literature (Beugelsdijk & Mudambi, 2013) and the related concept of proximity (Boschma, 2005). This discussion will lead to the development of some working propositions on the relationship between different kinds of proximity and location choice that will guide the research design and analytical chapters of this study.

2.1 Knowledge-based view of the MNE

A well-established explanation for the existence of MNEs is their capability to transfer and exploit knowledge more efficiently within the firm than through the (external) market (Buckley & Casson, 1976; Hymer, 1960). This suggests that knowledge generated, transferred and exploited within the organization yields an advantage that is not easy to reproduce in the marketplace. As a consequence, firms differ with respect to their knowledge bases and this causes differences in performance (DeCarolis & Deeds, 1999). Knowledge is therefore considered to be a strategically important resource of the firm (Grant, 1996) and this perspective on knowledge is known as the knowledge-based view (KBV) of the firm (Grant, 2002). The knowledge-based view has emerged from the resource-based view by shifting the focus from physical assets (such as machinery and capital) to intangible resources (such as in-house knowledge of technology) (Gassmann & Keupp, 2007).

Of course, differences in the knowledge bases of firms are not the sole reason for differences in performance between firms. Firms that are able to successfully recombine knowledge by multi-country collaborations between firm engineers, employees and inventors (and thus tap into different country’s ‘knowledge stocks’ (Furman et al., 2002)) can achieve competitive advantages that other firms will find difficult to imitate, duplicate or surpass (e.g. Martin & Salomon, 2003; Kogut & Zander, 1993).

The knowledge-based view literature mentions two knowledge processes that describe the recombination of knowledge within the firm. One is the internal transfer of knowledge (the sharing of knowledge between the HQ and the subsidiaries in different countries) to exploit competencies and benefit from ‘location-specific advantages’ (LSAs). The other is the integration

(12)

11

of knowledge from different sources in different countries to create new competencies

(Eisenhardt & Santos, 2001).

According to Cui et al. (2006), the internal transfer of knowledge is affected by both market and cultural factors. Especially organizational cultural distance and the dynamics of the market emerged as the factors with the strongest effect; close organizational cultures and a dynamic market positively influence the transfer of knowledge. Other determinants of internal knowledge flows are described by Gupta and Govindarajan (2000), who recognize the importance of the dual effect of the willingness and the ability to transfer and receive knowledge within the firm. This understanding translates into five elements of the knowledge transfer process that are important in order to understand the process: (1) the value of the knowledge possessed, (2) the motivational disposition to share this knowledge, (3) the existence, quality and cost of channels to transmit the knowledge, (4) the motivational disposition of the receiving actor to accept the knowledge, and (5) the receiving actor’s absorptive capacity for the incoming knowledge (Gupta & Govindarajan, 2000). The recognition of the dual effect of the willingness and the ability to transfer and receive knowledge has received a lot of attention in the years following the publication of the article by Gupta and Govindarajan (e.g.: Wang et al., 2004; Foss & Pedersen, 2004; Minbaeva et al., 2003). The second knowledge process, the integration of knowledge, depends on the internal transfer process to distribute the knowledge. The success of the integration itself depends on the absorptive capacity and compatibility between the different knowledge bases (Andersson et al., 2015).

Next to recombining knowledge within the firm, some firms also try to recombine their knowledge base with other organizations in countries other than their home country. The resulting strategic alliances can have a variety of organizational arrangements, such as technical exchanges, joint ventures, research and development partnerships, and licensing agreements (Inkpen, 1998). In the KBV literature, the associated knowledge inflows and outflows are combined into one concept: external transfer. While internal transfer is about the transfer of knowledge within the firm (across borders), external transfer refers to those processes in which firms share their knowledge and technology across firm boundaries (Eisenhardt & Santos, 2001). In the case of MNEs, external transfer can involve other firms, but also host country institutions and the quality and nature of innovation in the host country (Andersson et al., 2016). Another knowledge process mentioned in the KBV literature that can be linked to the recombination of knowledge outside the firm is called sourcing. Sourcing is the “process by which

managers identify and gain access to relevant knowledge that is being created in the environment”

(Eisenhardt & Santos, 2001: 14). In order to successfully source knowledge from outside the firm, managers need to be able to recognize and access knowledge. While recognizing knowledge (and knowledge gaps) is something that depends on managers’ assessment and

(13)

learning capabilities (Petersen et al., 2008), to have access to knowledge depends more on the embeddedness of the firm in local networks (Andersson et al., 2002). In this context, embeddedness refers to the extent to which a subsidiary’s direct, individual relationships with costumers, competitors, suppliers, etc. can be used to learn.

In both cases (transferring and recombining knowledge both within and outside the firm) the motive of the MNE to invest abroad (foreign direct investment, FDI) is to seek knowledge. The recombination of knowledge across countries, therefore, not only helps to achieve competitive advantages in the form of new and improved products and processes. It also extends the knowledge base from which the firm can draw in the future (Berry, 2014).

However, combining knowledge across country locations is easier said than done. Managing international knowledge activities and sharing knowledge across countries are both associated with high costs due to (for example) duplication and complex organization structures and this makes it difficult to manage these processes effectively (Berry, 2014). Organizations can also be less responsive to knowledge from abroad because of organizational inertia in innovation. This is the case when firms prefer to maintain the status quo in terms of location and type of R&D they are involved with at home (Berry, 2014; Narula, 2002). In addition, organizations tend to experience internal resistance against external views and ideas, revered to as the ‘not invented here syndrome’ (Katz & Allen, 1982). Likewise, Szulanski (1996) shows that firms have trouble with the transfer of knowledge when the source is not perceived as reliable, trustworthy or knowledgeable. These difficulties (among others) make managers of MNEs reluctant to locate their R&D activities abroad (UNCTAD, 2004), despite the advantages to performance.

2.2 Knowledge-based view of location choice

Adopting the knowledge-based view of the MNE influences the way in which one thinks about location choice. Using the KBV, traditional resources are less important and location choice is all about knowledge spillovers, knowledge transfer and knowledge exploitation (Mariotti et al., 2010). This section provides an overview of how location choice of MNEs is dealt with in the IB literature. The section is divided in two parts: the first part focuses on the IB literature that uses the country as a unit of analysis to explain MNE location choice. The second part discusses the IB literature that looks at within country differences to explain the location choice of MNEs.

2.2.1 Country level perspectives on location choice

Traditional IB literature uses the country as a primary geographic unit of analysis when studying the performance and strategic behaviour of MNEs (e.g. Zaheer, 1995; Dunning 1988). A lot of the work in this stream of the literature focuses on differences between countries and the implications of these differences for MNEs (Zaheer, 1995). The literature looks at firms crossing

(14)

13 national borders and at the economic activities of firms in other countries. The concept of ‘country knowledge stocks’ of Furman et al. (2002), referred to in the preceding section, is thus clearly part of this stream of the IB literature. Key concepts in the literature are ‘location-specific advantages’ (LSA) (or country-specific advantages, CSA) and ‘distance’ (Beugelsdijk, 2011). These two concepts are related to each other. In order to take advantage of LSAs across borders, firms need to overcome the ‘distance’ between the host and home country (Beugelsdijk, 2011). Distance does not only concern geographic distance, it may also refer to measures such as cultural, economic, language, religious, administrative, and institutional differences (Ghemawat, 2001).

Of course, distance will only be bridged when firms perceive a LSA. Studies such as Berry (2006) and Alcacer & Chung (2007) put forward empirical evidence proving this; both studies show that host country attributes influence the attractiveness of foreign R&D locations. Naturally, what constitutes a LSA differs for each industry. Industries that mostly internationalize their R&D are the chemical and pharmaceutical industry, information technology, telecommunications, and consumer electronics (Gerybadze & Reger, 1999). This is because the knowledge bases, regulatory requirements and consumer requirements differ greatly across countries for these R&D intensive industries (Gerybadze & Reger, 1999). Of these three, especially the presence of a highly advanced technological knowledge base influences the internationalization decision of a firm (Gerybadze & Reger, 1999). This is because intangible firm-specific advantages (FSAs) such as knowledge, in line with the knowledge-based view of the firm, are highly important in the globalised knowledge based economy that firms face today (Dunning, 1998). In other words: FDI by knowledge-intensive firms is less geared to exploiting existing FSAs (market-seeking strategy), and more to protecting or augmenting their FSAs by gaining new assets or partnering with another firm (strategic-asset seeking) (Dunning, 1998).

The existence of geographical knowledge bases in a world of digital connections seems paradoxical. Why would geographic location matter to firms, while the cost of communicating has drastically been reduced? Part of the answer lies in the distinction between knowledge and information. Information, such as the value of the Euro, or the price of gold, can simply be codified and can only be interpreted in one way (Audretsch, 1998). In contrast, knowledge is more difficult to codify, it is vague, and is often only recognized by accident (Audretsch, 1998). While information can be easily shared using digital methods, the cost of transmitting knowledge (and especially tacit knowledge) increases with distance (Krugman, 1991). This implies that ‘distance’ is best represented by a continuum scale: a scale that does not start at the border of a country and takes steps to other countries, but one that looks at variation in distance starting from the subnational level. I will discuss this ‘subnational’ stream of the IB literature in the next section.

(15)

2.2.2 Subnational unit of analysis

Literature focussing on the science park phenomenon is part of a newly emerged stream of the IB literature (Porter, 1990, 1998), which has developed in recent years alongside the regionalization literature (Ohmae, 1985; Rugman, 2000), both focusing on the spatial dimension of international business. The stream that is mostly used in the clusters literature (which includes the literature on science parks) has ‘zoomed in’ by looking at subnational territory. The regionalization stream has ‘zoomed out’ by looking at (the Triad) regions as geographic units of analysis (Boschma et al., 2015).

Beugelsdijk and Mudambi (2013) are important authors of the ‘zoom in’ discussion, arguing that the use of the country as the location unit of analysis causes weaknesses in IB research, because firms do no locate in the centre of a country and they also do not employ workers that represent country averages. The actual ‘distance’ firms need to bridge is therefore not clear when just using the country as unit of analysis. Moreover, due to the nature of tacit knowledge, the innovation process and the associated knowledge spillovers are spatially bounded (Malmberg et al., 1996). Firms are therefore inclined to co-locate or agglomerate with other firms, institutes and possible partners and form so-called clusters (Porter, 2000; Mariotti et al., 2010). Companies that are part of such a cluster may enjoy “special access, better

information, special relationships, powerful incentives and other advantages in productivity and productivity growth that are difficult to tap from a distance” (Porter, 2000: 32). This corresponds

with the findings of Baptista and Swann (1998), showing that firms that are located in strong clusters are more likely to innovate. However, the attitude of MNEs towards knowledge spillovers is not necessarily positive. Their attitude can vary depending on multiple factors of which I will highlight three below. The first one is the effect of the structure of the industry (McCann & Mudambi, 2005). For example: in an oligopolistic industry, firms realise that unintentional knowledge outflows to rivals can be extremely costly and cannot be offset by potential knowledge inflows from their competitors (McCann & Mudambi, 2005). The second factor concerns the nature of the firm itself. Although Dunning (1998) states that knowledge-intensive firms are less geared to exploiting existing FSAs and are more geared to protecting or augmenting their FSAs (see previous section), competence-exploiting MNEs (naturally) still exist. According to Mariotti et al. (2010), these competence-exploiting MNEs are less interested in spillovers and the associated transfer of tacit knowledge than competence-creating MNEs due to the risk of technological leakages1_{. The third, and final, factor influencing the attitude of MNEs} towards knowledge spillovers/clusters is the nature of the companies to which MNEs locate close. Mariotti et al. (2010) put forward empirical evidence showing that MNEs’ perception of knowledge spillovers (positive or negative) depends on the competitive position of the MNE

(16)

15 towards local competitors and on the expected balance of knowledge inflows and outflows. According to Mariotti et al. (2010), this explains why MNEs tend not to agglomerate with domestic firms (unless these firms enjoy some comparative advantages), but are willing to agglomerate with other MNEs. This being said, it is important to note that the knowledge externalities and spillovers related to science parks are not only a result of relationships between firms, but also of the relationships between firms and knowledge institutes, start-ups and university students. Be it in the form of partnerships, (future) employees or as clients. This enhances the location specific advantage of a science park vis-à-vis other locations where only companies co-locate (Westhead & Batstone, 1998).

What becomes clear from the above is that not ‘distance’ (not even distance between a subnational location and a firm instead of between a country and a firm), but ‘proximity’ is an important pre-condition for knowledge sharing and knowledge transfer. After all: “business is a

social activity, and you have to be where important work is taking place” (James Niedel in: Labich

& Graves, 1993: 44). The IB literature has not thought about proximity as of yet, but looking at ‘proximity’ can help to better understand the location choice of firms.

2.3 Proximity and MNE subsidiary knowledge flows

‘Proximity’ is a concept from economic geography (EG) and can be seen as the opposite of ‘distance effects’ (Boschma, 2005). Just as the focus on distance resulted in numerous measures of distance, the focus of economic geographers on proximity has led to the development of different measures of proximity. Central to this development has been the notion that geographical proximity is not the only explanation for the observed interaction between actors who are geographically close to one another (Broekel & Boschma, 2012). In fact, as mentioned in the preceding section, spatial proximity is not the same as interaction and interaction does not imply positive spillovers. Some firms are not able to receive knowledge and some firms are not willing to produce spillovers (Mariotti et al., 2010). The critical view on the role of geographic proximity has been initiated by the French School of Proximity Dynamics (Torre & Gilly, 2000). They introduced two categories: geographical and organizational proximity. However, these two categories are too loosely defined to be operationalized (Moodysson & Jonsson, 2007). Zeller (2004) contributed to the discussion by introducing the categories: spatial, institutional, cultural, organizational, relational, technological and virtual proximity. While the work by Zeller (2004) is analytically sharper, the categories overlap each other on too many points to provide a useful framework. A third major contribution to the discussion is Boschma (2005). Boschma proposes four measures of proximity (in addition to geographical proximity) that have an impact on the exchange of knowledge between actors. These four dimensions of proximity are: cognitive, social, organizational and institutional proximity. While the lines between these

(17)

categories are fine, they do not overlap. I will, therefore, use the dimensions of proximity as provided by Boschma (2005) in this study. In doing so, I will adopt Boschma’s claim that geographical proximity “is neither a necessary nor sufficient condition for learning to take place...

[but] facilitates interactive learning most likely by strengthening the other dimensions of proximity” (Boschma, 2005: 61). The following sections will go deeper into the four dimensions.

In doing so, I will present some working propositions, which will be used to structure the analysis for understanding how the subnational variation in emphasis on the nature of proximity in major Dutch science parks affects knowledge intensive MNEs location choice.

2.3.1 Cognitive proximity

Cognitive proximity indicates the extent to which actors posses the same absorptive capacity and potential for learning (Boschma, 2005). Thus, close cognitive proximity means that people share the same knowledge base, which enables effective communication and makes it possible to learn from each other. Shared cognition is the basis of so-called ‘epistemic communities’. These communities can be described as “research societies in which the members share a similar

educational and/or professional background and have thereby acquired a similar frame of reference with regard to scientific knowledge” (Moodysson & Jonsson, 2007: 120). Thus, being

part of an epistemic community makes it easier to establish and retain relations with other members of the community. Scholars, such as Nooteboom et al. (2007), have demonstrated that the formation of R&D alliances is indeed highly influenced by the cognitive proximity between the different partners.

Linking the concepts of cognitive proximity and epistemic communities to external knowledge transfer and knowledge sourcing, I argue that when one or more MNEs have located in a science park, others are likely to follow. As knowledge seeking MNEs believe they will benefit from a positive balance between knowledge outflows and inflows when working with other MNEs (Mariotti et al., 2010). One of the reasons for this belief may be that MNEs feel that other MNEs possess the same absorptive capacity and potential for learning. In other words: they may feel there is a high cognitive proximity between themselves and other MNEs and that this proximity will lead to a positive balance of knowledge transfers. Based on the aforementioned, I propose:

Proposition 1: Knowledge-intensive multinational enterprises will agglomerate when

there is a high cognitive proximity between them, in order to facilitate external knowledge transfer and knowledge sourcing activities.

(18)

17 2.3.2 Social proximity

Social proximity relates to the relationships between people of different organisations and the degree to which these relationships involve trust based on experience, friendship and/or kinship (Boschma, 2005). The more people trust each other, the higher the social proximity between them. Trust-based social relationships have been argued to foster the exchange of tacit knowledge, which is by nature difficult to acquire trough trade on the market (Maskell & Malmberg, 1999). Trust-based relationships exist, for example, in ‘old boys networks’ and between actors with a shared history (went to the same schools, worked at same place, etc.). The concept of social proximity is strongly related to the concept of embeddedness that was linked to knowledge sourcing in section 2.1. The difference between the two is their level of analysis. Embeddedness of a subsidiary is affected by the structure of the overall network of relations (Oerlemans & Meeus, 2005), while social proximity focuses on the personal relationship between actors.

The sharing of ideas and experiences increases social proximity between actors (based on trust) and thus enforces the benefits of agglomeration between actors. These benefits become a clear determinant for future FDI to the location (Bunnell & Coe, 2011; Mariotti et al., 2010). Trust-based relations may also be present before a MNE chooses where to locate. Pre-existing relations between a MNE and a firm/university/institute in a science park could therefore explain the propensity of the MNE to locate on the same science park. Thus:

Proposition 2(a): Pre-existing relations (social proximity) between a multinational enterprise and one or multiple actors in a science park explain the propensity of the multinational to locate in the science park in question.

Proposition 2(b): Social proximity between a multinational enterprise and science parks actors embeds the firm in the local network and this ensures knowledge sourcing processes.

2.3.3 Organizational proximity

Organizational practices are central to the issue of interactive learning (Boschma, 2005). While cognitive and social proximity are important in bringing organizations together and enabling interactive learning, knowledge creation depends on the capacity of actors to coordinate the exchange of complementary pieces of knowledge that are owned by a range of different actors between and within organizations. This becomes easier when the organizational context of the interacting parties is similar (Knoben & Oerlemans, 2006). Organizational proximity, defined as:

“the extent to which relations are shared in an organizational arrangement, either within or between organizations” (Boschma, 2005: 65), thus sets the stage for collaborating parties to

(19)

transfer and co-create knowledge. Organizational proximity can be studied on two levels: the structural level and the dyadic level. The former is about belonging to the same organization (Oerlemans & Meeus, 2005) while the latter is about the similarity of the organizational context in which actors from different organizations work (Wilkof et al., 1995). At both levels, one can look at the shared relations and place them on a continuous scale, based on the rate of autonomy and the degree of control that can be exerted in the organizational arrangements. According to Boschma (2005), organizational proximity is low when there are no ties between independent actors (e.g. ‘on-the-spot’ market) and high for actors that are part of the same hierarchical system (firm or network). At the dyadic level, one can also define organizational proximity by looking at the degree to which organizations use the same incentive mechanisms and routines (Metcalfe, 1994).

Too much organizational proximity can be a reason for firms not to co-locate, since high organizational proximity is characterized by a lack of flexibility (Petruzzelli & Carbonara, 2007). Multinationals that are not located at the subnational location in question may still enjoy some of the benefits of agglomeration. By locating relatively nearby they are still able to access a highly skilled workforce and are also able to (if they like) work together with actors at the centre. It is assumed that connections between MNEs located outside the focal subnational location (science park) and firms, research institutes, and universities located in a science park are mainly sustained through organizational proximity (Petruzzelli & Carbonara, 2007). However, organizational proximity will likely not create interaction by itself, but will play a role in particular events and supply chains.

Proposition 3(a): Organizational proximity sustains interaction between MNEs outside and firms/institutions inside the science park.

Proposition 3(b): All actors perceive high organizational proximity to positively influence knowledge integration and knowledge sourcing processes.

2.3.4 Institutional proximity

Institutional proximity is associated with institutions at the macro-level (as opposed to the micro-level of social proximity). Boschma (2005) uses North (1991) who makes a distinction between informal institutions (taboos, customs, traditions, norms and values of conduct) and formal institutions (laws, property rights, constitutions). Both forms of institutions influence the way organizations coordinate their actions (Hall & Soskice, 2001). This, in turn, affects the level of knowledge transfer, interactive learning and therefore the level of innovation (Boschma, 2005). Institutional proximity is strongly interconnected with social and organizational proximity, since the formal and informal institutions are embodied in the norms (social) and routines (organizational) at the micro-level (Boschma, 2005).

(20)

19 An example of how informal institutional proximity affects interaction can be found in a study by Agrawal et al. (2008). They show that ‘co-ethnicity’ (being part of particular group for which membership raises the likelihood of sharing social capital), increases the probability of knowledge flows. Another example is Ponds et al. (2007); their study shows that two organizations with the same institutional background (e.g. two universities) are more likely to successfully collaborate than two organizations with different institutional backgrounds. Based on the above, I propose the final working propositions of this study:

Proposition 4(a): Because of their high institutional proximity universities and knowledge institutes want to be closer to/work together with other universities and knowledge institutes.

Proposition 4(b): Knowledge-seeking multinational enterprises want to work together with other knowledge-seeking multinationals because of their high institutional proximity.

2.4 Conclusion

Having established the conceptual foundation for a subnational perspective of knowledge intensive MNE location choice, the following chapter will deal with the methods this study uses to find out if the above working propositions are valid and to answer the main question.

(21)

3. Methodology

This chapter elaborates on the methodology used in this study. The first section of this chapter discusses the ontological and epistemological foundations on which this research is based (Brannick & Coghlan, 2007; Saunders et al., 2009). Next, the multiple-case research design adopted for this study will be discussed. This discussion includes the selection of cases and considers important quality criteria associated with the research design. The chapter will conclude with a description of the data collection and the data analysis.

3.1 Ontological and Epistemological foundations

Fleetwood (2005: 197) explains that “the way we think the world is (ontology) influences: what

we think can be known about it (epistemology) [and] how we think it can be investigated (methodology and research techniques)”. Thus, both the research philosophy adopted by the

researcher and the social science phenomenon to be investigated have important implications for the research design. This study adopts a subjectivist ontological view and an interpretivist (subjective) epistemological view, since the possibly different perceptions of social actors (managers) are the main input of the research. The subjective ontology argues that reality is a product of the human mind (Brannick & Coghlan, 2007) and the related interpretivist epistemology reflects this by arguing that insights will be lost if the complex world and the differences between humans is reduced to a series of law-like generalizations (Saunders et al., 2009). The contrast of this positioning to an objectivist ontology and (post)positivist epistemology implies different views on the most appropriate ways of obtaining knowledge and the nature of what is considered adequate knowledge (Morgan & Smircich, 1980). The adopted position is thus reflected in the qualitative multiple-case study design explained below.

3.2 Qualitative multiple case study research design

While quantitative research is most appropriate when the goal is to discover causal links between variables, qualitative research is useful when researchers aim to establish beliefs, perceptions and/or motivations related to a particular phenomenon (Van der Velde et al., 2004; Eisenhardt, 1989). In other words: qualitative research is useful when one seeks to answer ‘how’ and ‘why’ research questions (Yin, 2013). Since this study focuses on the perceptions of proximity and the motivation behind location choice, this study uses a qualitative research design. When using qualitative research design it is assumed that each individual is unique (Taylor, 2005). In order to study the uniqueness of the individual, qualitative researchers employ data collection methods such as interviews, observations and case studies (instead of employing random selection techniques as is the case in quantitative research) (Taylor, 2005).

Apart from the division between quantitative and qualitative research designs, there are two different approaches to conducting research: the deductive and inductive approach

(22)

21 (Saunders et al., 2009). Whilst the deductive approach is geared towards testing theory, the inductive approach is concerned with the generation of new theory emerging from the data (Saunders et al., 2009). This study is set up using a deductive bottom-up theorising approach (Shepherd & Sutcliffe, 2011), as propositions concerning proximity, location choice and knowledge processes have been formulated based on academic literature, and I seek to understand how actors perceive the importance of proximity to location choice in order to refine the conceptualization in an incremental theory building process (Ridder et al., 2012). The deduction of theory will be done using a multiple case study research design. In doing so, this research draws on techniques used in the inductive approach as described by Eisenhardt (1989), as these support the multiple case study design and are equally applicable to the deductive approach (Hyde, 2000).

3.2.1 Multiple-case Study Design

The case study is a research strategy that focuses on understanding the dynamics present within single settings (Eisenhardt, 1989). Since this study looks at three different science parks (three different single settings), this is a multiple case study. In order to answer the research question, I will speak to MNE managers, science park managers and managers of scientific institutions at all three parks. Yin (2013) typifies this specific multiple case study approach with embedded units of analysis as a Type 4 study (see figure 1).

Figure 1: Multiple case study with embedded units of analysis

Amsterdam Science Park

•University •Two MNEs

•Two Knowledge Institutes •Science Park Management

Leiden Bio Science Park

Utrecht Science Park

(23)

Using a multiple case study requires a ‘replication logic’ (Zucker, 2009). The ‘logic’ underlying the use of multiple case studies is: each case must be selected so that it either (a) predicts similar results (a literal replication) or (b) produces similar or contrasting results but for predictable reasons (a theoretical replication) (Yin, 2013). This study uses a theoretical replication logic since the three cases (science parks) differ in size, composition and specializations (see appendix 8.1) and these factors will most likely influence the perceived importance of the different dimensions of proximity in each science park. Furthermore, this study is a parallel multiple case study, since the cases are being studied concurrently (Thomas & Myer, 2015). The aim of this study is to add to and refine the literature on subnational location choice. Thus, the study is exploratory in nature.

3.2.2 Quality criteria

The strength of a multiple case study approach is that it allows for an in-depth description and analysis of decision-making processes and/or events (Yin, 2013). In addition, using multiple cases strengthens the theoretical implications of the findings by triangulation of the evidence in several observations (Eisenhardt, 1989). However, case studies do have weaknesses. For example: the reproduction of evidence is often unsuccessful and the method has been criticized for lacking generalizability (Yin, 2013). However, case study findings are generalizable to theory (Ruddin, 2006). Furthermore, the intensive use of empirical evidence in multiple case studies may lead to too much complexity and/or too narrow and idiosyncratic theory as researchers are unable to assess what the most important relationships between observations are (Eisenhardt, 1989). To enhance the credibility of the multiple case study approach, four quality criteria must be considered. These are: construct validity, internal validity, external validity and reliability (Yin, 2013; Gibbert et al., 2008).

Construct validity is “the degree to which a test measures what it claims, or purports, to be

measuring” (Cronbach & Meehl, 1955). Thus, construct validity is concerned with the

measurement of the concepts of research (Yin, 2013). Triangulation has been advocated to improve the construct validity of a study by combining methods. This can mean using several kinds of methods or data (Patton, 2001). However, I do not have other sources (next to the interviews) at my disposal and the study may be influenced by my subjectivity. To account for these weaknesses I will test for construct validity using two other methods as described by Yin (2013), I will have a draft case study report reviewed by key respondents and I will repeat the same chain of evidence at the collection of data within and across cases. This allows readers to reconstruct the path from initial research question to final conclusions and limits the use of ‘subjective’ judgements, thus increasing the construct validity (Gibbert et al., 2008).

The data should still be viewed from different angles to find similarities and differences between cases and thus increase the internal validity (Yin, 2013). Internal validity, also called

(24)

23 logical validity (e.g. by Cook & Campbell, 1979), refers to the causal relationships between variables and results (Gibbert et al., 2008). The issue is whether the researcher uses powerful and compelling logical reasoning and plausible causal argument, which are able to defend the research conclusions (Gibbert et al., 2008). Yin (2013) proposes ‘pattern matching’ as a measure (among less applicable measures for a case study) to enhance the data analysis and thus the internal validity. Pattern matching deals with the comparison of empirically observed patterns with either theoretically predicted patterns or ones that are established in previous studies and in different contexts (Denzin & Lincoln, 1994). In this study, patterns are theoretically predicted and identified within and then across cases. In addition, the nature of this study allows for incremental theory development in case the data points to or creates gaps in the literature that need to be filled (Ridder et al., 2012).

The external validity (the extent to which the findings of a study are generalizable) is always problematic with multiple case studies (Yin, 2013). The reason for this is that the sample size is small. However, this does not mean that case studies are devoid of generalization. Eisenhardt (1989) argues that case studies can be used to develop theory. Not using statistical generalization, but analytical generalization to go from empirical observations to theory (Yin, 2013). Cross-case analysis may provide a good basis for analytical generalization (Eisenhardt, 1989; Gibbert et al., 2008).

Lastly, the reliability of a study is determined by the level of transparency and the possibility of replicating the study (and finding the same results) (Gibbert et al., 2008). To increase the reliability of this study an interview protocol has been constructed to ensure the consistency of (at least) the initial questions. In addition, all information pertaining the cases (notes, documents, recordings) will be organized in a case study database and the procedure of collecting and analysing the data is clearly specified in this chapter to allow other researchers to evaluate and replicate the study.

3.2.3 Selection of cases

Because this study approaches three different groups of people (science park managers, managers of MNEs located in the science parks, and managers of knowledge institutes located in the science parks), the respondents are selected using different sampling rationales. Table 1 provides an overview of the respondents. Science park management is relatively small in all three parks. I will therefore speak with only one science park manager at each park. The science park managers were asked to select two MNEs and two knowledge institutes they belief are most important to the park (see for a overview of the three science parks appendix 1). This selection method makes cross case analysis easier.

(25)

Table 1: Overview of respondents Science

Park:

Organization: Description: Job role respondent:

Amsterdam MNE 1 R&D department of a manufacturing firm Company Secretary MNE 2 Data centre and colocation services Vice-President Knowledge

institute 1

Building a bridge between research and advanced ICT

Relations managers Knowledge

institute 2

Fundamental subatomic physics research Institute manager University Beta faculty of the University of Amsterdam Director Market

Development Science park Central and joint organization of all partners of

the park

Director Leiden MNE 3 Pharmaceutical manufacturing and research CEO

MNE 4 Pharmaceutical manufacturing Director Facilities, Maintenance & Engineering Knowledge

institute 3

Research and education in Bio-Pharmaceutical Sciences

Institute manager Knowledge

institute 4

Enables business and government to apply knowledge

Research manager University Leiden University Medical Centre Director of Research Science park The foundation responsible for enhancing the

network and developing the cluster

Park manager Utrecht MNE 5 Veterinary pharmaceutical company Research and

Development Officer MNE 6 R&D department of a food company Site Director

Knowledge institute 5

Research focused on developmental and stem cell biology

Managing Director University University of Utrecht Policy officer

valorisation Science park 1 Organization responsible for the development

of the park

Director

Science park 2 Business Development Manager

I subsequently contacted managers of these firms and institutes since it can be assumed that managers have the most insights regarding the location choice of their organisation and their ties with other organisations. The choice to interview two MNE managers and two knowledge institute managers in each science park has been made based on the need for a broad dataset (in

(26)

25 order to produce useful results), while also taking into account the limited time available. In addition, I will speak with a university manager in charge of external relations and/or knowledge transfer.

The MNEs and knowledge institutes in Amsterdam and Leiden that had been mentioned by the science park mangers were willing to participate in the study. One firm and one knowledge institute that had been mentioned by science park management in Utrecht did not have time to take part in this study. I therefore interviewed a manager from another MNE (also mentioned by science park management). I did not contact another knowledge institute.

3.3 Data collection

The qualitative data needed for this study have been collected using semi-structured interviews. Interviewing people using a semi-structured setup means that the interview is conducted using a set of questions and the interviewer has a good sense of what topics need to be covered (Fylan, 2005). The semi-structured interview, therefore, ensures flexibility (Saunders et al., 2009). Semi-structured interviews differ from structured interviews in that they are not conducted using a predetermined list of questions, which are covered in the same order for each respondent (resembling a questionnaire administrated verbally). Semi-structured interviews also differ from unstructured interviews, since the semi-structured approach has no set boundaries concerning the topics that should be covered (Fylan, 2005), creating the potential for unexpected insights that could contribute to developing the conceptual foundation of the study.

The interviews will need to uncover how the managers understand proximity and how they believe it matters to location choice. It is highly likely that the managers do not think in terms of proximity, therefore the interviews will cover themes linked to proximity, such as: working relations, results from working together, and attraction of knowledge. Table 2 (below) shows how the interview questions covering these themes are linked to the working propositions.2_{The questions seen in table 2 are only the essential questions of the MNE} interview set (the knowledge institute and science park management sets can be found in appendix 8.2). The rest of the interview will be filled with warm-up questions, probing questions and follow-up questions (Berg, 2009). The interviews will take about one hour. This time is needed to cover all the themes and to gain rapport (which is harder to achieve with less time).

(27)

3.4 Data analysis

All interviews are recorded (with permission), transcribed and subsequently uploaded into the software program NVivo. This software program makes it possible to systematically allocate content to predefined categories, which allows for a structured and thorough analysis (Payne, 2004). The allocation of content is done through a process called coding: assigning a word or short phrase (codes) that symbolically assigns a summative and/or essence-capturing attribute to specific parts of the written text of the interviews (Saldana, 2012). When following the process of deducting theory, as is done in this study, categories and codes can be developed

Table 2: Matrix interview questions Questions: WP 1 WP 2a WP 2b WP 3a WP 3b WP 4a WP 4b

Can you tell me a little bit about if (and if so, how) the presence of other MNEs has been a determining factor in the location choice? x Do you think that knowledge internationals tend to assimilate in the

same science parks? x

If you think about the multinationals and knowledge institutes located in this science park, what is your impression of their similarity regarding their (educational and/or professional) background, capabilities and knowledge sourcing activities>

x

Do you feel MNEs agglomerate in order to gain knowledge? x In your view, how do social relationships (based on trust, friendship

and/or kinship) influence the location choice of firms? x x How is the (social) interaction between firms and institutions? Does

this interaction start knowledge transfers? Is this, in your view, the intention of firms and institutions?

x x How would you typify the ties between firms/institutions with each

other? Are these loose (on the spot markets) or strong (part of a network)? Do firms have a lot of autonomy and control in their relations?

x x

Do you consider firms around here similar in how they work? x x In what way do strong organizational arrangements with other

firms/institutes influence the behaviour of your organisation? Does it make you inflexible or does it help you to reach out to others outside the park?

x x

Do firms in this science park interact with both other firms and

knowledge institutes on the same level? x x Do you feel there is a difference in norms/value/codes of conduct

(28)

27 before data is collected (Bourque, 2004). However, as stated in section 3.2 this research also draws on techniques used in the inductive approach as described by Eisenhardt (1989), as these support the multiple case study design and are equally applicable to the deductive approach (Hyde, 2000). In line with this partly inductive approach, some codes are generated after examining the collected data (Bourque, 2004) (see table 3 for the used codes).

Table 3: Coding scheme

Code: Sub code: Definition: Working

propositions

Cognitive Proximity

- Possessing the same knowledge base, being part of the same epistemic community

Proposition 1 Institutional

Proximity

Norms and Values

taboos, customs, traditions, norms and values of conduct

Propositions 4a, 4b Rules and

Regulations

laws, property rights, constitutions Propositions 4a, 4b Organizational

Proximity

Network Organizational proximity is high for organizations that are part of the same network

Propositions 3a, 3b Stand-alone Organizational proximity is low when there are no

permanent ties between actors

Propositions 3a, 3b Organizational

arrangements

The extent to which organizational arrangements are similar among organizations

Propositions 3a, 3b Social

Proximity

- Social proximity relates to the relationships between people of different organisations and the degree to which these relationships involve trust based on experience, friendship and/or kinship

Propositions 2a, 2b

Geographic Proximity

- The physical distance between organizations - Sourcing - The process by which managers identify and gain

access to relevant knowledge that is being created in the environment Propositions 1, 2b, 3b External Knowledge Transfer

- External transfer refers to those processes in which firms share their knowledge and technology across firm boundaries Propositions 1, 2b, 3a, 3b, 4a, 4b. Practical Reasons

- Practical reasons for locating somewhere - Small country - Respondents indicate the Netherlands is only a small

country

(29)

The codes that have been developed before the data collection are foundational (thematic) and come from the theory (Ryan & Bernard, 2003). Each code has been assigned a short description in the software program (see table 3). Recording the analytical thinking used to devise codes in this fashion, allows for a more transparent study (Shenton, 2004).

All interviews are conducted in Dutch and subsequently entirely transcribed in Dutch, since, as Vygotsky (1987: 236) states: ”every word that people use in telling their stories is a

microcosm of their consciousness”. The importance of language as stated by Vygotsky makes it

difficult to accurately translate the useful parts of the transcribed interviews into English. Inevitably, a part of the meaning and richness is lost in translation (Halai, 2007). All qualitative researchers need to show sensitivity (or reflexivity) towards the ways in which they are part of the social world they study (Alvesson, 2003). Making the translation of an interview part of those reflexive deliberations ensures a limited negative impact on the reliability of the study (Piekkari & Welch, 2006).

The data that has been gathered and analysed using the methods as discussed in this chapter will be presented in the following chapter.

(30)

29

4. Results

This chapter provides an overview and a discussion of the results retrieved from the interview data. The chapter consists of three sections. The first section discusses each of the three cases (science parks) and their embedded units of analysis (interviewees) separately in within-case analyses. The second section provides a cross-case analysis, discussing the similarities and differences across the science parks in order to find patterns that may support the working propositions. The third and final section discusses in what way the findings from the cross-case analysis are in line with the theory-based working propositions of this study.

4.1 Within-case analysis

4.1.1 Case 1: Amsterdam Science Park

What becomes clear from looking at the interviews conducted in the Amsterdam Science Park is that knowledge intensive multinational firms, as described by the knowledge-based view of the firm, seem to be looking first and foremost for interesting/promising knowledge sources. According to the manager of MNE 1 “the science park is highly attractive for us because it bundles

a lot of different Beta knowledge”. When firms are interested in knowledge that is being

developed in the Amsterdam Science Park, they are likely to try to collaborate with the university, a knowledge institute or a researcher. However, this does not necessarily mean that these multinationals will locate in the park. According to the manager of the university in charge of market development “only about 10 per cent of the firms we work with are located here”. Adding that R&D departments of MNEs are global departments that are “only interested in where

they can get important knowledge”. The manager of knowledge institute 1 adds: “firms will move here for knowledge, if they decide to move at all”. This is because, while knowledge is important,

other factors also have to be taken into account when making a location choice decision. Examples of such factors mentioned by the respondents in Amsterdam are: the proximity to an airport, the hometown of current employees, amount of rent, available space, the look and feel of the location, and the facilities. The manager of the university thus concludes that “it remains to

be seen what the added value of the local knowledge is [for location choice decisions], proximity has lost its importance now people are able to communicate in so many ways”. So while the

clustering of knowledge increases the perceived location-specific advantages (LSAs) of the science park, information and communication technologies and the high (temporary) mobility of individuals decreases the importance of co-location for knowledge-intensive firms.

However, while geographic proximity does not seem to be too important for most MNEs looking for innovative knowledge, some firms have placed a small number of people in the science park in order to be close to the action. The manager of MNE 1 (which has recently

The influence of proximity on Multinational Enterprise location choice : a multiple case study in Dutch science parks

Enterprise Location Choice

A Multiple Case Study in Dutch Science Parks

Abstract

Statement of Originality

This document is written by student Niels le Duc who declares to take full responsibility

for the contents of this document.

I declare that the text and the work presented in this document is original and that no

sources other than those mentioned in the text and its references have been used in creating

it.

The Faculty of Economics and Business is responsible solely for the supervision of

completion of the work, not for the contents.

Acknowledgements

Table of Contents

Index of Tables and Figures

1. Introduction

2. Conceptual foundation

2.1 Knowledge-based view of the MNE

2.2 Knowledge-based view of location choice

2.3 Proximity and MNE subsidiary knowledge flows

2.4 Conclusion

3. Methodology

3.1 Ontological and Epistemological foundations

3.2 Qualitative multiple case study research design

3.3 Data collection

3.4 Data analysis

4. Results

4.1 Within-case analysis