The multiple geographies of MNE strategies

(1)

T H E M U L T I P L E G E O G R A P H I E S O F M N E S T R A T E G I E S M A S T E R T H E S I S International Management Dr. Johan Lindeque Erik Dirksen MSc Alessandro Melloni 11236167 23-06-2017 MSc. Business Studies Supervisor Second reader Student Student ID Submission Date

(2)

INDEX OF TABLES AND FIGURES

FIGURES

1 THE AAA STRATEGY TRIANGLE 19

2 MULTILEVEL SCALING OF FIRM'S STRATEGY MEASURED THROUGH THE AAA FRAMEWORK 22

2 OBSERVATIONAL FEEDBACK PROCESS 60

TABLES

1 SELECTION CRITERIA 29

2 METHOD OF ALLOCATION OF THE THREE TYPES OF SUBSIDIARIES 31

3 CATEGORIZATION OF SUBSIDIARIES 32

4 DESCRIPTION OF CASES AND FIRMS 33

5 CATEGORIZATION OF COUNTRIES 34

6 DESCRIPTION OF THE INDICATORS FOR THE CATEGORIZATION OF COUNTRIES 36

7 CATEGORIZATION OF CITIES 38

8 LEGOANALYTICAL OUTCOMES 42

9 MATTEL ANALYTICAL OUTCOMES 43

11 ADIDAS ANALYTICAL OUTCOMES 46

12 ASICS ANALYTICAL OUTCOMES 48

13 ZEISS ANALYTICAL OUTCOMES 51

14 AGGREGATE ANALYTICAL OUTCOMES 55

(4)

ABSTRACT

The conceptualization of distance and locations in the IB literature, traditionally bound to the country as primary reference point, has been recently broaden by contributions from the Economic Geography stream. New super-national and subnational levels of analysis are now arousing more interest among researchers. However, still little attention has been addressed to how a multilevel geographical analysis might affect the way MNE strategies are perceived. Building on the

Arbitrage-Aggregation-Adaptation framework by Ghemawat (2007), the strategies of six firms belonging to three different

industries have been examined across multiple geographical layers represented by industrial clusters, cities, countries and macro-regions. The study has been carried out through a deductive multiple case study. The results prove that a multilevel spatial perspective offers a considerably more nuanced view of the strategic phenomenon. Firms appear to pursue different themes across the geographical layers and to target more coherently specific types of locations. The findings also show that a specific spatial level is more easily associated with a limited number of strategies.

Keywords: geographic scale, AAA framework, macro-regions, industrial clusters, countries, global

(5)

ACKNOWLEDGEMENTS

I would sincerely like to thank my supervisor Dr. Johan Lindeque for his knowledgeable guidance and great understanding of the issues a person, and not just a student, might encounter during the work on the thesis. He has always been capable of placing the last piece of puzzle exactly where it was missing, and to encourage me to give my very best. It’s rare in the academic world. And I don’t say these sentences to conform to the norm, I do mean it.

I would like express all my gratitude also to my girlfriend Madeline, who helped me with great love and sympathy throughout these last months. I would never stop being surprised of how truly gentle she is.

I would then like to thank my brother, that helped me with precious suggestions at any time of the day. Finally, my sincere gratitude and affection go as well to my parents who allowed me to study abroad in Amsterdam. I am so proud of all the trust they always put in me and I hope these pages can pay them back, at least in part.

(6)

Statement of Originality

This document is written by student Alessandro Melloni 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

(7)

1. INTRODUCTION

Traditionally, the concept of location has been studied by economic geographers and international economists. The first analyzed the characteristics of spaces and traced borders or connections, while the second were mostly concerned with trade flows, national GDPs and balance models (Beugelsdijk, 2010). This literature has been mainly focused on countries and subnational regions, advancing the conceptualization of locations. On the other hand, the International Business (IB) research stream concentrates its research efforts on the organizations and the multiple ways they interact with the external world.

Such context leaded to two different academic communities, one focusing on countries and their potential resources to be exploited (e.g. inexpensive labor, governmental policies or technological development), and one focusing on firms and their internal resources and capabilities (Mudambi et al., 2016). In the last decades, scholars intensively tried to merge these two different streams. Approaches such as the renowned Eclectic Paradigm (Dunning, 1988) tried to take into consideration elements related to the specific location and how firms incorporate them into their boundaries. Later, the view that MNEs develop Location-Specific Advantages besides their unique Firm-Specific Advantages (Rugman & Verbeke, 1992) became later widely acknowledged. These locational advantages can partially contribute to the global development of the firm, and thus be incorporated as constituent features of the company.

More recently, the advanced studies on the concept of location of the Economic Geography converged with the IB school (Boschma et al., 2015), with still unclear and unexplored implications. Applying geographical concepts where the MNE does not represent the focal unit of analysis to an MNE-based architecture made some concepts overlap or collide (Mudambi et al., 2016) and generating multiple new research routes (Cano-Kollman et al.2016). In particular, alongside with the country (the traditional economic unit of analysis), super-national regions (Rugman & Verbeke, 2004), industrial districts and clusters (e.g., Porter, 2009), country clusters (Sethi, 1971) and sub-national aggregates, such as global cities (Sassen, 2004) rose and gained quickly popularity among scholars. The geographical economy is hence divided into several layers of locational extents which offer, as unit of analysis, different insights to the IB research. Zooming out or zooming in the focus introduces some variables and overlook others, explaining in different ways the strategic behavior of firms (Boschma & al., 2015).

(8)

However, these definitions stemming from the geographical economy literature do not always match perfectly with the conceptualizations of distance elaborated by IB researchers. In this regard, the concept of distance and proximity are multilateral and dynamic notions tailored on the firm (Nachum et al., 2008) while the purely geographical conceptualization of aggregates (macro-regions, countries, cities, etc.) is static. This means that geographic distance does not always correspond to economic distance, which can be holistically defined in terms of administrative, cultural, geographical and economic differences (Ghemawat, 2001). Consequently, two units geographically close might result to be economically more faraway than another unit from a different continent (Nachum et al., 2008).

Yet, relatively little work has been done to investigate how this multitude of geographical aggregates recently adopted by IB scholars relates to MNE strategies (Boschma et al., 2015) and, more specifically, how strategies are perceived when one or another of such geographical extents is used. Thus, this thesis will address the extent to which the MNEs strategies, defined on their location choices, might appear different when considering different geographical levels of analysis. For instance, regions or countries. In other terms, the overarching assumption to be tested is whether, by shifting the focus from a geographical layer to another, the analytical perception of the strategic dimensions varies.

RQ: How extensive is the degree of aggregation, adaptation and arbitrage in MNE value creation

when (re)scaling the geographic levels of MNE strategies?

Additionally, other dimensions that might influence the relationship between strategies and geographical layers will be inquired. Namely, the industry (and the industry-related drivers) to which the MNEs belong and their countries of origin. These are indeed factors that might be capable of shaping the geography-strategy relation. In order to assess the strategic pattern of multinational firms, the recent AAA framework developed by Ghemawat (2007) will be used. This tool allows to study three well-defined strategic dimensions and to assess them across different geographical layers.

The research is based on a qualitative and multiple case study method with a deductive approach. The subsidiaries of six different MNEs belonging to three industries will be analyzed and different strategic motives will be identified according to the threefold subdivision of the AAA framework. The objective is shifting the observational point across different geographical layers and study if and to what extent these strategic fashions differ. The ultimate aim of this study is to offer a clearer picture on how IB scholars and economic geographers should approach and carefully study international strategies, and how managers should consider multiple conceptualizations of locations.

(9)

Structured around the research question, the paper will first consider in the literature review the role of locations and the four geographical sets here selected, as well as a brief examination of the conceptualization of distance. Later, the basic strategic models, their relationship with locales and the implication of using the AAA framework as analytic tool will be reviewed. Secondly, research statements will be outlined into a theoretical framework.

In the central section, the methodology and its criteria will be illustrated. The core of the thesis is constituted by the selection criteria, the within-case and cross-case analysis. The discussion of results, the conclusions and a few last additional notes will then close the paper.

(10)

2. LITERATURE REVIEW

This section will start off by spanning all the fundamental conceptualizations of the considered geographic dimensions, later utilized in the research stage. The second part will briefly introduce the the AAA strategic framework, while the last part will present more in detail the research gap and the working prepositions.

2.1 A Conceptualization of Distance

Discussing locations necessarily implies defining at first a notion of distance in an economic sense. In the IB stream this concept includes several acceptations related to multiple schools of thought, going from the Liability of Foreigness (Hymer, 1967; Kindleberger, 1969, Zaheer, 1995) to the Transaction Cost Theory (Rugman, 1980; Hennart, 1982), with the first focusing on the extent to which the MNE, ceteris paribus, incurs additional costs that local firms do not bear (Sethi & Judge, 2009); and with the second referring to the categories of expenses that burden those international companies which prefer market transactions to internalization. But this second view has risen a broad criticism, as distance does not always increase costs (e.g. IT industry), and it might raise advantages as well (Hutzschenreuter et al., 2014).

At the same time, psychic distance gained popularity too (e.g. Johanson & Vahlne, 1977). But later on, a different strand of literature joined the debate highlighting the importance of culture constructs as a matrix of distance, with several different approaches (Hofstede, 1984; Schwartz, 1994; House et al., 2004). Still, the cultural dimension did not appear to be able to represent the sole answer to the discussion on distance, as it is flawed by multiple methodological inadequacies (Shenkar, 2001).

By the end of the century, with the increasing interest by the research community in institutional linkages with the IB themes (Peng, 2002), a wealth of literature brought to the fore the institutional distance, with some attempts to restructure this concept through institutional perspectives into new explanatory frameworks (Xu & Shenkar, 2002; Berry et al., 2010).

But across all these representations of distance, the geographic factor has been overshadowed, while even the most recent works emphasize its effect on the strategic behavior of firms (Ragozzino, 2009).

In the first place, this happened due to the rising globalization phenomenon that significantly influenced the academic debate (Garrett, 2000). But the effect of the “globalization apocalypse” (Ghemawat, 2007), which was supposed to flatten and lead the world a complete global convergence (Friedman, 2005), has been later disputed and downsized by another group of academics (Eden &

(11)

Miller, 2004). These researchers argue that distance is still playing a major role in the political, economic and institutional relationships between countries and MNEs. Mainstream economists have traditionally approached this notion through gravity models (e.g. Losch, 1954), where distance increases as the reference point (e.g. a district) moves away from major centers (e.g. a global city) and where larger centers may influence more the extension of distance. Ghemawat (2003), contesting the gravity model, claims that distance deeply affects the MNE strategic decisions through four dimensions stemming from the long-standing IB tradition: geography, culture, economy and administration (CAGE). Such holistic working approach puts the company within a bilateral dialogue between countries (Hutzschenreuter et al., 2014), quantifying the distance and allowing comparisons between spaces, serving as a useful tool both to policy makers and managers in order to assess foreign direct investment decisions or regulative gaps (Ghemawat, 2003). The CAGE approach enables a sizable comparison not just between countries, but also between industries, which might develop different levels of sensitivity to the four components of distance (Ghemawat, 2007).

Due to its fourfold structure, the CAGE appears to overcome the over-simplification error that many holistic distance frameworks encounter when trying to reduce a multi-dimensional approach to a single output (Zaheer et al., 2012). Thanks to its complexity and solidity, this framework will offer a suitable background in the next reflection on strategies and locations.

However, the CAGE and all the aforementioned attempts to conceptualize distance in the IB stream limit their analysis to the country as only unit of space (Hutzschenreuter et al., 2014). More recently, IB scholars are reconsidering the geographic parameter with new insightful and multi-directional structures that overcome the limits of the static Euclidean approach (Beugelsijk & Mudambi, 2013), getting surprisingly closer to the gravity model. But the most important insights in this field come from the recent influx of notions from the economic geography, especially more geographical units of analysis to be used as reference points in distance measurements (Andersson et al., 2011; Boschma et al., 2016).

2.2 Four Types of Locations

The revamped interest for distance and geographical aspects can be attributed to the recent developments of the debate on globalization. Globalization is here meant as an international integration of goods, services and capital (Garrett, 2000), is a complex and vast phenomenon which affects the economic and social reality we live in. But new concepts such as semiglobalization (Ghemawat, 2003) disputed the extensive effect of this process. Analyzing the integration of both the international product markets (through observations on the trade flows, FDIs and price integrations)

(12)

and the factor markets (capital, labor and knowledge), Ghemawat (2003) proved that a perfect integration is still far from being achieved, since MNEs build the bulk of their advantages on exploiting differences within the markets, rather than similarities.

A new debate is now rising around global cities, macro-regions, sub-national regions and industrial cluster. These units are now triggering a new intense research activity and might convey deeper and more extensive insights into the IB panorama (Boschma et al., 2016). In the next paragraphs four old and new concepts of location will be briefly presented, together with their contextualization and relevance within the following analysis.

2.2.1 Regions

While the regional studies already had a large popularity in the economic geography stream (Fujita & Krugman, 2004), it was Ohmae (1985) who first introduced IB scholars to the division of the world into three economic regions (USA, EU and Japan), each of them sharing similar technological infrastructures, highly capitalized firms, moderate country growth, protectionist forces on the government and homogenous demand. This view was further developed by Rugman and Verbeke in 1993 and an intensive theoretical and statistical follow-up by the same authors and other scholars extended the resonance of this thesis until today (Rugman, Verbeke, 2007; Hoon Ho & Rugman, 2014; Hoon Ho & Li, 2015). The Ohmae’s original regions were later broadened to NAFTA, EU and Asia-Pacific (Rugman & Hoon Oh, 2008).

It appears to be difficult even for successful companies in their home region to overcome the “global impasse” (Ohmae, 1985) and then achieve good results in terms of sales in all the regions. Rugman and Verbeke (2004) argue that firms’ global expansion is hindered by products not equally accessible or attractive in every region (only a few superproducts manage to fully reach the three different demands) and the market position as well may require more knowledge about the local economy and consumption trends, hard to gain and transfer through the regions. This implies a low transferability of firm-specific advantages (FSAs) across the foreign divisions and MNE subsidiaries (Rugman & Verbeke, 2001). At the same time, it is noticeable a lack of sufficient location-bound FSAs (Rugman & Verbeke, 1992), as firms fail to incrementally develop the adequate knowledge base and relationship network needed for a profitable internationalization (Johanson & Vahlne, 2008).

On the other hand, a “triad power” is defined as a company with remarkable capacity to exploit opportunities in all the three markets with a similar penetration and no blind spots (Rugman & Verbeke, 2004).

(13)

The method followed by Rugman and Verbeke through the years did not vary, and it consists in a sale distribution among the domestic and host regions of the world top 500 firms. Their methodology has been criticized by Dunning et al. (2002), who proposed to review the work making use of macro-economic data, besides a division in six areas instead of three. Five limitations to the applied micro-level method were found by the authors themselves: possible sales difference within SBUs, the place of registration of sales which may be different from the real one, sales strategies that could alter the data validity, and differences from an industry to another regarding the minimum market share to be considered a relevant actor in the market. Nevertheless, several flaws can be identified in a macroeconomic perspective as well, but the centrality of firm makes the outcome more relevant to IB studies (Rugman & Verbeke, 2007).

The original regional study showed that the vast majority multinational enterprises tends to be home-region oriented, few bi-home-regional, while just a minority host-home-region oriented or really global. This brings us to another important outcome of both the Rugman’s and Verbeke’s (2004) and Ghemawat’s (2003) findings: while at the downstream end (sales and services supply) of the internationalization process firms tend to capitalize similarities and thus keep most of their business in the home region (regional multinationals), at the upstream end (gathering resources) they manage to exploit differences through scale and scope economies, realizing an international arbitrage. This, to simplify, means that MNEs preferably sell in their home region, a more familiar business environment, products manufactured abroad (or with resources acquired in foreign regions), generating an asymmetry in the MNE’s flows (Rugman & Verbeke, 2004). Besides the geographic extent, regions present different and unique features in terms of culture, law and policies, and economy. Therefore, they are coherent with the CAGE distance notion and, besides the critiques, regions represent a valid reference point suitable to the analysis of business strategies, and constitutes the largest geographic aggregate used in this thesis.

2.2.2 Countries

“Most types of economic activity that can be conducted either within or across borders are still quite localized by country” (Ghemawat, 2007, p. 11).

In the IB studies, the nation has always been the focal term of comparison (Andersson et al., 2011). The concept of business distance itself, regardless of the different conceptualizations, is traditionally conceived as distance between countries, rising often doubts of oversimplification (Hutzschenreuter et al., 2014). However, even though multiple other locations are now at the center of the academic discussion, the relevance of country as fundamental geography is unquestionable as it influence the

(14)

firms though a multitude of factors. Even when alternative definitions of distance are given, the national borders still represent a commonly accepted “hard” driver of distance (Beugelsijk & Mudambi, 2013). The country influences indeed the firm’s competitiveness (Rugman & Verbeke, 1992) and strategy (Wan & Hoskisson, 2003), in particular entry modes (Kogut & Singh, 1988). Most of the studies focus however on the countries’ impact on MNE performance and profitability (Makino et al., 2004; McGahan & Victer, 2010). Such recognized effects on multinational companies happen through a large array of methods that correspond to all the main four determinants of the GLOBE distance. For instance, large interest aroused more recently around institutional aspects of nations (e.g. W.R. Scott, 1995). Other traditional factors generally acknowledged by IB scholars are the culture of countries (e.g.Hofstede, 1980), the psychic distance (Johanson & Wiedersheim-Paul, 1975) and the economic characteristics. Countries represent the main reference points when assessing

international business distance and influences on multinational firms thanks also to the simplicity of

assessment. Countries are indeed a widely recognized unit upon which large amounts of data and deep theoretical foundations are already available (Hutzschenreuter et al., 2014). For these reasons, they represent a fundamental geography for the following analysis.

2.2.3 Clusters

While aggregating the country unit into regional sets sheds more light on the direction of the major global value chains, zooming in onto a national or subnational level gives insight on the micro strategic decisions of firms (Boschma et al., 2016). In this regard, building on the North’s (1990) view of institutions, Chan et al. (2010) prove that subnational regions are capable to extensively affect the strategies and performance of multinational firms. Subnational regions are a diffusely acknowledged extent also among economic geographers (e.g. Markusen, 1987).

The first subnational geography herein considered are the industrial or, more generally, business clusters. Industrial clusters have been subject of study by economic geographers (e.g. Losch, 1954) for a long time, and just by the end of the last century, Porter (1990) consistently drove the IB literature to dedicate to them. These are intended as “geographic concentrations of interconnected companies and institution in a particular field” (Porter, 1998:78).

Differently from macro-regions, countries and cities, clusters do not have a clear geographic shape and a precise governance system set by policy makers (Humphrey & Schmitz, 2002). Their identity is indeed more directly related to an economic scope, rather than a cultural, social and political dimension. Due to their ambiguous nature, many, since the seminal work on agglomeration economies by Marshsall (1890), have been the attempts to clearly identify an accurate and

(15)

comprehensive definition and phenomenal description. Gordon & McCann (2000) trace three microeconomic models: pure agglomeration, industrial-complex and social-network, in an attempt to explain the different motives that bring to the rise of clusters.

According to the IB literature, the reason behind their existence is related to an increase of competitive advantage linked to the proximity effect (Porter, 1990). Setting their subsidiaries in a close network of similar firms, MNEs might benefit of shared and developed infrastructures, a large endowment of industry-specific resources, a highly specialized local labor market, policy advantages and inter-firm knowledge spillovers (Mariotti et al., 2010). Clusters improve also vertical integration through a reduction of arm-length transactions. This implies a cut to the economic and geographic distance between firms and their clients and suppliers, all concentrated in a limited space (Porter, 1998). Furthermore, not only companies belonging to the same business area or industry tend to agglomerate, but all sorts of enterprises that might engage forms of complementarity with locally set firms (Delgado et al., 2010). However, these agglomerations do not foster just coordination, but also competition, both within the cluster and towards external companies, and brings the clusters’ agents to uniform into a precise strategic and organizational structures (Porter, 2000). From the whole set of advantages, the modern microeconomic approach prefers the importance of competition and knowledge over the traditional Ricardian focus on endowments or the agglomeration theory (Porter, 1998). The recent stream has indeed vastly highlighted the role of intangible assets, and especially knowledge, in the development of business clusters (Fleming & Marx, 2006; Bathelt et al., 2004; Audretsch & Feldman, 2004), even though such spillover effect might prevent agglomeration in case the knowledge outflows/inflows and comparative advantages gaps of local companies and MNEs are perceived as significant (Mariotti et al, 2010). This is, knowledge and information externalities spillovers can reduce the the uncertainty due to the increased distance and grow competitive advantages, but outflows towards competitors may exceed the benefits of cluster’s network. Thereby, clusters address all the dimensions of the CAGE distance, acting on the knowledge and social networking (Gordon & McCann, 2000), the economy, the space and the institutional aspects via cluster-specific policies undertaken by regional or national governments (Martin & Sunley, 2003).

To what concerns this analysis, the cluster is definitely a complete and intriguing unit of analysis, diffusely recognized and pregnant of strategic implications.

2.2.4 Cities

The last geographic unit adopted in this study is the city. While industrial clusters have specifically economic characteristics, aspects that enclosed their relevance mostly within IB boundaries, cities are

(16)

a geographical unit familiar to many disciplines like social sciences, microeconomics and anthropology (Sassen, 2002). Such subject aroused great resonance especially among the economic geographers (A. Scott, 2002). One of the fathers of this discipline, Losch (1954), connected cities with business clusters, arguing that industrial agglomerates might generate cities or fill the distance between already existing ones. To what pertains the economics field, most of the contemporary literature is based on the seminal monocentric urban model developed by Alonso (1964), later expanded into a non-monocentric structure by Fujita & Ogawa (1982) and more recently further improved by the New Economic Geography school. These microeconomic models aim fundamentally to find a general equilibrium both of money flows and offer/supply, overlooking the role of strategies and companies as discrete independent agents (Mudambi et al., 2016). Moreover, tools commonly used in this stream, such as the core-periphery model, are based on little empirical research, weakening their validity and applicability by policy makers (Fujita & Krugman, 2004). Nonetheless, the city as a unit of analysis is a rather novel concept to IB scholars, and several structures directly stem from this previous work. One of the cornerstones here is the first attempt by Hymer (1972) to categorize the different types of cities into three tiers, according to increasing levels of global connectedness, knowledge out(in)flows, skilled workforce abundance and market size. Later, Friedmann (1986) divided cities into primary and secondary classes, keeping the interconnectedness as major criterion.

However, in the IB literature the city as a unit of analysis gained a consistent popularity just in the last decades, since the preferred metric has always been the country-state or, more recently, the regions (Goerzen et al., 2014). The “global city” is not solely a vast and densely populated metropolis (demographic view), but more specifically an urban center characterized by (a) advanced producer services (law, finance, etc.), (b) a cosmopolitan environment (which fosters a managerial, cultural and working exchange) and (c) international connectedness (Goerzen et al, 2014). This functional tradition shares its roots with the agglomeration approach implemented by economic geographers who suggest a model of agglomeration, meant as result of centripetal and centrifugal forces. The first are represented by linkages, thick markets and knowledge spillovers (Fujita & Krugman, 2004). The second by immobile factors, high land rents, congestion and other diseconomies that could deter companies from installing subsidiaries into a global city. These two contrasting pressures dynamically shape the borders and size of the urban centers and all the other agglomerations.

Nevertheless, both the IB and the economic approaches identify similar dimensions which depict the city as a network of relationships where the material and knowledge flows (direct or indirect, as externalities) play a central role (Beaverstock et al., 2000). Considering it as a space, the city is where

(17)

the centripetal forces widely exceeds the centrifugal pressure, which implies that a foreign company can more easily overcome the uncertainty, discrimination and the other detrimental effects of distance (Goerzen et al, 2014). Differently from the purely economic geographic approach, the IB adds the specific strategy of the firms into the frame. In this regard, it has been argued that MNEs co-evolve with the urban environment, and firm’s specific idiosyncrasies determine the locational choice about setting the subsidiary within or outside the global city Goerzen et al., (2014). Cities hence act on all the four dimensions defined by Ghemawat (2003) in his conceptualization of CAGE distance. Therefore, urban centers are the lowest layer of analysis to be included in this research and, like the other three geographies mentioned, can serve to explain the strategic motives of firms in terms of location choice.

2.3 Strategic Framework

In this paragraph, few mainstream strategic frameworks and the extent to which they embed the locational variable will be briefly reviewed. The way such theories contributed to the studies of MNE behaviors helps understand the choice of a relatively secondary analytical tool. The rationale behind the choice of the AAA model as designated structure will be then presented, together with the illustration of its salient traits. This Ghemawat’s framework has the advantage to thoroughly balance into a whole the effects of incomplete globalization, of organizational complexity and the tradeoff between standardization and local adaptation.

In the IB literature, after a two decades’ shift from the country-level the firm-level of analysis, Johanson’s and Vahlne’s (1977) Uppsala model made headway in bringing the locational matter back up. This staged view assumes that firms internationalize incrementally, both strategically (from simple export to setting a subsidiary) and geographically (from psychically close to faraway countries), due to the effects of distance. Even though the locational choice has been later proved to be discrete and not built upon the previous path (Benito & Gripsrud, 1992), and the spatial unit was limited to the country, this work had the the merit to reintegrate the locational variable into the strategic function.

Rugman (1981) distinguishes the firm-specific advantage (FSAs) from the country-specific advantage (CSAs). This view hence supports the existence of a locational source of competitive advantage, and in a later review (Rugman & Verbeke, 1992) CSAs are divided among typical advantages stemming from the home country and from the host country. The capability of ventures to leverage and internalize these locational benefits gives rise to the distinction between non-location

(18)

bound FSAs, transferable from the subsidiary to the parent firm and vice versa, and location-bound FSAs, a set of competitive advantages that can be leveraged just locally.

This complex structure is then translated into four types of international strategies, with the “transnational solution” as the ultimate strategic organization (Rugman & Verbeke, 1992). A transnational MNE combines a high degree of responsiveness and adaptation to the local environment, while exploiting both home and host CSAs (dual use), with an extended set of internal FSAs that allows as well high levels of standardization (Rugman & Verbeke, 1992).

In his “eclectic paradigm”, Dunning (1980, 1988) proposed a holistic view of the firm hinged on three dimensions: ownership (O), location (L) and internalization (I). The L component has though been largely overlooked compared to the other two (Dunning, 1998), and gained just recently a fair interest (Andersson et al., 2013; Beugelsdijk et al., 2016).

The globalization recalled the IB scholars’ attention to the spatial matters in the last decade of the previous century. Alongside the over mentioned Rugman’s and Verbeke’s (1992) update to the FSAs model, Dunning (1998) found that the globalization, together with the emergence of “alliance capitalism” and the intellectual capital revolution, were overturning the business landscape. Four motives drive modern MNEs, seeking natural resources, markets, efficiency or strategic assets (Dunning, 1998). Countries are more and more seen as a complex environment to interact with, instead of simple resources recipients. Even the Uppsala model is updated in order to answer to the new interconnectedness of business texture (Johanson & Vahlne, 2005).

However, as discussed before, such overwhelming force of the globalization phenomenon has been demurred by a second group of scholars who advocates the persistence of geographical influences in the MNE strategic design (Rugman, 2010; Nachum & Zaheer, 2005). For long time, companies tried to enforced a single global strategy assumed to be superior, and they dealt with local complexity just through adjustments (Ghemawat, 2003). But the concept of local is now striking back and furthering the traditional dichotomy between global integration, typically associated with standardization, and local responsiveness which, on the other hand, hinges upon a high level of differentiation (Ghemawat, 2007).

The transnational model of firm theorized by Rugman and Verbeke (1992) and Bartlett and Ghoshal (1999) exemplifies an organizational structure capable to embrace both these strategies. As result, a transnational company is more than just a mere addition of the two drivers, but a solution able to

(19)

shape a multifocal engine which exploits all the local differences and similarities of spaces, the unique set of FSAs and the opportunity of scale economies (Rugman & Verbeke, 1992).

This picture can be explained in the light of a new dichotomy, this time between differences and similarities (Ghemawat, 2007) rather than standardization and adaptation. In fact, an MNE can be seen as an organization exploiting either one of these two sources of competitive advantages. It is fundamental to understand that spatial units and the distance between them are crucial in this strategic framework. Considering differences between places, a firm can either adjust to them through adaptation or overcome them via aggregation (Ghemawat, 2007). But Ghemawat (2003) introduces arbitrage, a third “forgotten strategy”, adds a third possible route to the traditional dichotomy. This can be achieved by exploiting the differences. A threefold Adaptation-Aggregation-Arbitrage (AAA) strategic division is thereby traced with the intent to encompass all the international strategies and

how they can be implemented.

While global firms generally aim to develop a competitive advantage by deploying only one of these three strategic dimensions, in a few cases two of them can be embarked - usually a primary and a secondary one, but a purely triple-A strategy is realistically impossible to be achieved (Ghemawat, 2007).

Figure 1 - the AAA strategy triangle (Source: author’s adaptation on Ghemawat, 2007)

Still according to Ghemawat (2003), adaptation can be further divided into five sub-categories:

variation strategies attain to chances in products and services, but also in the business and managerial

settings (policies, codes, etc.); focus strategies refer to specific products, geographies and vertical-stages strategies in the global value chain, especially market segments that can reduce the effect of regional and country-related differences; narrowing the scope, the externalization strategies aim

(20)

conversely to expand the scope through alliances, franchising, user adaptation and networking; design strategies serve to reduce the cost of variation: flexibility, platforms and modularity are three typical components of this process, helping the company to increase productivity and cutting expenses; finally, innovation strategies give firms the chance to enhance the beneficial effect of adaptation keeping the company’s offer up to date and following demand’s needs promptly.

The second A refers to the aggregation and to the traditional motives of similarities exploiting processes. In this case, the firm can improve efficiency and productivities flattering the curve of costs per unit of time. This might overcome possible weaknesses in the adaptation strategic field and it is represented by the two classic strategies of economies of scale and economies of scope. The first one allows decreasing costs on increasing amounts of product, while the second permits the firm to use a given set of resources or capabilities in different contexts. It is important to highlight how, for this force, geographic concentration is extremely relevant, since it gives the chance to concentrate the production chain as well, creating at the same time a homogeneous demand.

The last of the three As is the arbitrage, which belongs to the differences exploiting route. This effect intensively uses geographical distance and semiglobalization into a global strategy (Ghemawat, 2007). Alongside the traditional cost arbitrage (acquiring resources where they are the cheapest), defined as economic, Ghemawat (2007) also considers administrative and geographic arbitrage. The first pertains to institutional, political and legal differences. For example, a country might be particularly appealing thanks to its tax policies. The second, the geographic arbitrage, has been subjected to a drastic decrease in the last decades due to lower transportation costs, but it still has a major role in sectors such as energy, where the possibilities of moving resources across geographic dispersed regions are limited.

Further explanation of these three strategic themes will be given in the methodology section. However, while the drivers of the AAA strategies are multiple and complex, as Ghemawat (2007) lays the foundations of his framework upon the holistic “CAGE” distance, this study will primarily focus on the way these strategies relate to locations choices (rather than, e.g., firm-specific features like the diversification of product portfolios).

Ultimately, the AAA framework is a tool that puts the burden on locational effects integrated in a comprehensive and structured view of international strategies. In addition, making use of a rather novel tool can shed new light on MNE behavior across spaces and display implications until now undetected. On top of this, compared to other strategic views, the triangle and the multiple

(21)

sub-categories appear to be clearly measurable and comparable, offering the researcher an extensive instrument of analysis.

2.4 Scaling spaces and strategies

The previous paragraphs illustrated the conceptualization of distance endorsed in this study and the four geographical layers on which the MNE strategies will be analyzed in the next chapters. This will be carried out through the AAA framework, an analytical architecture capable to link geographies and strategies, divided into three distinct themes. Building on such theoretical foundation, this section will present four working propositions that embrace geographies and strategies.

“Most management problems involve multilevel phenomena, yet most management research uses a single level of analysis. A micro or a macro lens alone yields incomplete understanding at either level” (Hitt et al., 2007, p. 1).

All the four aggregates here considered denote, to different extents, the capacity to explain the business reality or, more specifically, the strategic patterns of firms. At the same time, none of them can exhaustively serve as the sole managerial and academic landmark, since all of the units can offer some insights the others cannot (Boschma et al, 2015). Yet, there is still a large void in the literature about how MNEs relate to different spatial dimensions, as the majority of works are based exclusively on the country level of analysis (Andersson et al., 2013). In other words, a comprehensive alignment between the different geographical aggregates and the strategies implemented by multinational firms is still lacking. As all the four geographies present different ways to relate to the organizational complexity (Boschma et al, 2015), when analyzing the MNE strategic behavior on different spatial layer through the AAA triangle, the observer might perceive the same strategies differently. This can be indeed connected to the fact that, changing the reference points, the observation of a single phenomenon can be altered. Rephrasing such concept differently:

WP 1 Rescaling the geographical level of analysis, the same MNE strategy is expected to change in

terms of the most salient strategic dimensions

As discussed, the AAA strategy that a firm can pursue is generally a single one at once, while in fewer cases a minor one can be undertaken by the firm as secondary theme. So the most salient dimension is generally just one A per geographical level, and occasionally two.

(22)

Figure 2 - multilevel scaling of firm's strategy measured through the AAA framework (source: author).

Other possible insights derivable from this investigative approach concern the organizational complexity of firms. In detail, MNEs might ground their strategies on a specific spatial level rather than others. For instance, UK supermarket companies have been proven to address more coherently the subnational layer and, more specifically, urban centers or even specific districts of cities (Clarkson et al., 1996). Like in this example, industry-specific motives can play a decisive role in the selection of the focal geographical level. Another significant influence on the MNE management and thereby location choices might be represented by the specific domestic country of the firm (Rugman & Verbeke, 1992). Thereby, a second working statement is formulated as follows:

WP 2 The strategies of similar types of firms are expected to be more coherent at a specific geographical level of analysis

The use of the generic work “type” can include different methods of grouping similar companies. This research will indeed consider the multilevel location choices of firms belonging to the same industries or to industries driven by the same factors and value chain’s features. But firms can belong to the same type also in terms of domestic country. So firms can be grouped according to their industry of nationality to seek more insights.

Following akin premises and argumentations, such types of MNEs should follow similar strategic routes on each geographical level. In other words, firms belonging either to the same (or an affine) industry, or sharing the same national origins, are expected to deploy reciprocally identical AAA strategies on each spatial layer of the geographical hierarchy here considered. However, this does not

(23)

imply that a certain type of companies follows only one AAA strategy across all the levels. A firm can indeed potentially pursue a different strategy per each level, but other companies of the same type (thus either same industry or same nationality) are expected to do the same. Therefore:

WP 3 Similar types of firms are expected to pursue the same strategies at every geographical level of analysis

Finally, based on the assumption that a given geographical reference point has just a limited explicative power (Hutzschenreuter et al., 2014), it is interesting to investigate whether, on a specific geographical level, companies appear to pursue just a limited number of strategies (in terms of the AAA framework, one or two strategies). Thus, the last statements can be phrased as follows:

WP 4 Every geographical level of analysis is associated with specific salient MNE strategic dimensions

Hence, some MNE strategies might not be observed on given spatial levels, regardless of the firm type. More information about the different working propositions will be provided in the discussion chapter.

(24)

3. RESEARCH METHOD

In order to investigate possible variations of strategies across the four geographic layers selected, a multiple case study will work as an appropriate tool of analysis. Before motivating this choice, a note about the philosophical approaches that underlie it must be done.

3.1 Ontological foundations

Ontology embraces the studies and concepts linked with the notion of existence and its manifold aspects, broadly reality can be perceived as either an objective or subjective construct (Brannick and Goghlan, 2007). By relying on secondary data collected without the direct intervention and influence of the researcher on the matter of analysis, the present study adopts an ontologically realistic approach (Saunders & Lewis, 2012). This is, the researcher does not influence the observed reality. Built on this premise, it is here investigated whether just by rescaling multiple levels of analysis - in this case geographic levels - and therefore multiple points of view, the reality changes (as an output of the subject) or holds unvaried (as an objective truth). An objective data collection and methodology might then not be sufficient to achieve outcomes not influenced by the observer. The inquirers and their mind can indeed modify reality by the choices undertaken (e.g. variables, perspective, spatial levels selected) even without personally altering it. This means that objectivism and subjectivism can, to some extent, coexist (Morgan and Smircich, 1980).

The reflection on ontology must be completed by a remark on the Postpositivist epistemology, the research approach here undertaken. Gephart (2004) defines the principles of the Postpositivist school as a strictly realistic philosophy where the final goal is discovering truth through an objectively scientific method. On the other hand, Interpretative Research is a different tradition rooted into relativism, aimed at the investigation of complicated meanings and understandings (Gephart, 2004). While the analytical tradition adopted in this research will follow positivist and realistic constructs, it will be questioned whether this approach can actually lead to objective outcomes (“the cat in the box is dead or alive”), or these are inseparably shaped by the subject-object linkages (“the cat in the box is both dead and alive”)1.

3.2 Research design: multiple case study

By departing the common ground of statistical criteria, qualitative methods hide several pitfalls that made them frequently questioned tools among Social Science scholars (Yin, 2013). These flaws are

(25)

unanimously rooted into a lack of rigor and generalizability (Gibbert & Ruigrok, 2010). Other common critiques, in particularly upon case studies, concern the excessive structural complexity linked to the great use of empirical evidence (Eisenhart, 1989) that has created the distinction between quantitative as “hard” research against qualitative as “soft” research (Yin, 1984); the reliance on arbitrary considerations rather than logical (Flyvbjerg, 2006); lack of a common vocabulary (Eisenhardt & Graebner, 2007); and remarkable time consumption (Baxter & Jack, 2008). For these reasons, academics like Yin (2013), Tracy (2010) and Eisenhardt (1989, 2007) invested a large research effort in defining qualitative standards and designs commonly trusted and accepted. This work allows the great variety of qualitative methodologies to conform to a commonly acknowledged approach. The quality criteria will be further explained in the next section.

Multiple case study is a qualitative method that earned a significant popularity among business researchers in the last two decades (Eisenhardt & Graebner, 2007). It is not simply used when quantitative procedures are not viable (Gephart, 2004), on the contrary it can better capture the relationships of causality in real-world contexts and complexities (Bryman, 2009) and represents the most suitable tool for strategy analysis (Grant, 2016). This is motivated by the fact that this method enables an in-depth understanding of the observed reality, when intricately intertwined with the contextual effects that influence it (Yin, 1981; 2013). It also guarantees the researcher more flexibility and attention to the processes (Yin, 2013) and in such research pursuit, qualitative data (interviews, free observations, documents, photos, etc.) and quantitative data (numerical sources) can be combined (Eisenhardt, 1989). The qualitative nature of the approach lies indeed in the method of analysis and not merely in the kind of data or sources involved. Multiple case studies can be an effective method both for building theory (Eisenhardt, 1989) and for testing it (Yin, 1998). Testing is indeed the main focus chosen by this thesis which hinges upon the deductive examination of mainstream approaches based on a single geographical analytical scale. Namely, the choice of the country-level as preferred, and often only, terrain of investigation. At the same time, as in this case, it is not uncommon in case studies that theory building and testing create a continuous back-and-forth dialogue (Eisenhardt, 1989). Since the goal of this inquiry is to evaluate and compare strategic choices across multiple geographies, a multiple case study enables an accurate evaluation of strategic patterns (Yin, 2013) and differences between cases (Baxter & Jack, 2008). And significantly, case study methodologies are adopted as well by Ghemawat when analyzing firms’ strategy with the AAA framework (2007).

To summarize, the multiple case study is the preferred method of this research, since it adequately supports the investigation and comparison of different cases, allowing to identify strategic fashions departing from a multitude of theoretical bodies, like the Economic Geography, the International

(26)

Business and the AAA Strategic Framework. This is accomplished by relying on both explanatory (the detailed distribution of MNEs’ subsidiaries across the globe) and exploratory (the strategic behavior of firms across different geographic aggregates) objectives (Yin, 1998).

3.3 Quality criteria

In order to overcome the aforementioned criticism about qualitative methods and multiple case studies, Yin (1984) delineates two pillars that can assure a fair level of rigor and generalizability: validity and reliability. The first is further divided into construct, internal and external validity. These concepts have been widely recognized by international researchers and constitute the current paradigm for most of the qualitative method-based studies (Eisenhardt & Graebner, 2007).

On the words of Gibbert & Ruigrok (2010, p. 712), construct validity is “the extent to which a study investigates what it claims to investigate”. This can be achieved through a triangulation of methodologies and data sources (Yin, 2013; Patton, 2002). The first is here ensured by the adoption of a variety of procedures like value-based categorizations, strategic scorecards, combinations of multiple economic indices, classification of cities, and so on. The second one is guaranteed by the reliance on both qualitative (e.g. strategic scorecards, observations, company portfolios) and quantitative data (e.g. GDP, economic indices, company balance sheets, etc.) from multiple sources (e.g. global firm database, company websites, newspaper articles). Another method to achieve the construct validity is establishing a chain of evidence, and this will be carried out in the discussion and analysis section of this thesis, where the reader will be guided through linkages of detailed observations and grounded syllogisms.

So while the first type of validity is mostly related to data collection, the internal validity is built around the data analysis stage (Yin, 2009). The aim of this requisite consists in fact into establishing a clear causal relationship between conditions and conclusions (Gibbert & Ruigrok, 2010). To assure that, the next paragraphs will offer a detailed explanation of the logical transitions, supporting the

interweaved flow of assumptions, investigations and inferences, during both the explorative and explanatory phases (Yin, 1994). A variety of structured schemes and indicators has been designed purposely for this inquiry, so without following any predefined route built by other researchers. Consequently, the internal validities plays even more a crucial role in ensuring the rigor of the study.

Finally, by conferring external validity to their work, researches adopting qualitative procedures can achieve a good level of generalizability of their results (Yin, 2009), although internal and construct validity are argued to be more relevant, since they represent the corner stones of the entire analytical

(27)

building (Gibbert & Ruigrok, 2010). Generalizing from the chosen cases to the global is particularly critical in multiple case studies as it is not supported by statistical sampling (Yin, 1994). However, the gap between analytical and statistical generalization can be filled by a solid replication logic of the selected cases. Indeed, there is potentially no limit in terms of quality of generalization, especially when significant cross-case insights are identified (Eisenhardt, 1989). Fundamental to this regard, is to elucidate a continuous and reciprocal feedback between data and theory (Eisenhardt, 1989). The theoretical triangulation can contribute to strengthen internal and external validity (Denzin, 1978). In this work it is obtained by combining theories stemming from the Economic Geography (e.g. the conceptualization of cities and clusters) and the IB (the definition of distance, the AAA strategic framework, the triad regions, etc.). The replication logic herein adopted will be thoughtfully explained in the next paragraph.

Reliability, the second pillar defined by Yin (1984), is the extent to which similar results can be achieved through a replication of the research methodology (Silverman, 2005). Therefore, the reader should not be dependent on the unique researcher’s analytical actions. Transparency throughout all the phases of the research with descriptions of protocols and rationales behind the choices will accompany the reader in the next paragraphs, in the attempt to reduce the random error and minimize inferential voids (Gibbert & Ruigrok, 2010). It represents a particularly delicate task, given the original methodology developed and the working propositions examined in these pages.

3.4 Case selection

In general terms, the overarching scope of this investigation is not to define patterns or truths valid for all the international enterprises in the world, but rather to understand if, to some extent, by changing the focal geographical term of analysis, our perception of a strategy pursued by an MNE varies. Therefore, even from a relatively small sample it is possible to infer an approach useful to test a truth (rather than a truth itself) valid for all the population. Namely, the usage of multiple, instead of just one, geographic scales to analyze strategic phenomena.

Given this necessary remark, the case selection has been conducted according to both the two directives defined by Yin (2013): literal and theoretical replication. Six multinational corporations belonging to three industries have been selected. The prerequisites are a large scale in terms of global presence, brand awareness and financial turnover, alongside a leading position in the specific industry within the home country or even into the global economy. This criterion guarantees a consistent number of subsidiaries, which here constitute the basic unit of analysis - or, better, their location under a strategic point of view. Further, this model of firm mirrors the picture of a strategically

(28)

“global” company painted by Ghemawat (2007). Manufacturing firms have been preferred over services firms, as services have a more labile linkage between physical location and delivery and production processes.

The second stage of the selection process is designed around the scope of investigation. On purpose, the working propositions refers generically to types of firms. To differentiate international firms into categories, this study used indeed two classification systems (so two types of MNEs): the reference industry and the geographic origin of the firms. The scope is indeed to test whether firms belonging either the same country or to the same (or similar) industry present interrelated strategic patterns across geographies. So two major dimensions are taken into account to examine possible connections: the industry and the country of origin. Dividing firms into these “types” can in fact elicit more insights.

The industrial aspects are particularly influent in this study and are subdivided into other two criteria. The first refers to the way the firm creates value. This is not connected to firm-specific, or even subsidiary-specific (Rugman & Verbeke, 2001), advantages that are by definition difficult to be imitated and substituted (Barney, 1991), but rather to standard processes shared by all the firms of a specific industry. These industry-specific processes are capable of significantly influencing the value creation of MNEs (McGahan & Porter, 1997), meant as financial performance, profitability and strategic development and scale growth.

Connected to the first selection criterion, the second one assesses the complexity of the industry, which ultimately shapes the value creation internal and external dynamic. The notion of industrial complexity does not refer to a “competitive” complexity as commonly acknowledged in the organizational and strategic schools (Dess & Robinsons, 1984), but to the production characteristics of extensive economic chains (Hausmann et al., 2014). The latter notion is synthetized by the Economic Complexity Index developed by Harvard University which offers sizeable landmarks upon the economic complexity of industries and countries (Atlas of Economic Complexity, 2017).

Around this rationale, four companies belonging to two similar industries have been selected according to the theoretical replication logic. These four cases should be able to build sufficiently strong insights about industries characterized by relatively low complexity and with the production phase mainly driven by unskilled labor. So on the one hand, Mattel and LEGO, first and second global leaders in the toy industry, and on the other hand Adidas and Asics, global MNEs active in the sportswear and footwear industries, have been selected. On top of this, a couple of polar types (Eisenhardt & Graebner, 2007) belonging to a highly complex industry have been chosen to pair to

(29)

the other four cases. These are Carl Zeiss and Canon, multinational leading firms specialized in the production of imaging and optical manufacturing and sale.

So while the first two criteria follow the type of industry as leading dimension, the other selection criteria hinge upon the home country of the firms. The geographic origin of firms is a more immediate concept traditionally central in the IB stream (Porter, 1994). The influence of home countries has been argued to impact on MNE strategy in a great array of ways, spanning, just to mention a few, cultural (Hofstede, 1980; Scwartz, 1994; etc.), psychic (Johanson & Vahlne source) and performance (Brouthers, 1998; Makino et al., 2004) forces. In detail, Japan, Germany and the USA are historically countries characterized by strong and diverse managerial cultures, deeply influent on the national MNEs (Witt & Redding, 2009; Yates, 1993).

Table 1 - Selection criteria - colored areas (source: author)

Choosing both the geographical origin and the industry as selection criteria for the types of firms offers the investigation more points of view and strengthens the inferential chain of the analytical phase. Thereby in this case the methodological triangulation contributes to the construct validity (Denzin, 1978). Notably, selecting just one single replication criterion could have harmed the construct and external validity, as dimensions substantially influent on the outcomes would have been overlooked, with a loss of detail (Siggelkow, 2007). The replication logic and selection criteria logic

Cases

Industry Geography

Industry Major Cost

Driver Complexity

Home Region

Home Country

1.1 LEGO Entertainment Labor Low Europe Denmark

1.2 Mattel Entertainment Labor Low Americas USA

2.1 Adidas Sportswear;

Footwear Labor Low Europe Germany

2.2 Asics Sportswear;

Footwear Labor Low Asia-Pacific Japan

3.1 Zeiss Imaging _Electronics Technology; _Knowledge High Europe Germany

(30)

for multiple case studies is particularly complex and has to be conducted with great attention to the specific context (Eisenhardt & Graebner, 2007).

One last consideration has to be done about the relevance of triangulation at this stage to ensure the research validity: such twofold replication method can grant more solid and rich cross-cases insights. For instance, the selection of industries all driven by the same factors could bias the generalizability of the outcomes, since the industrial-specific influences on the company strategy would be overlooked. Similarly, selecting MNEs from one single country would diminish the explicative potential of the research. As last remark, the geographical replication logic aims to outline possible differences within similar industries due to diverse geographical origins (so within the toy/entertainment and the sportswear/footwear industries), and potential similarities across differently featured industries (sportswear/footwear and imaging/optic electronics).

To summarize, the selection criteria can be presented as follows:

1. Overarching selection criterion à large MNEs, with a great number of globally dispersed subsidiaries.

2. Industry

a. Within-case literal replication à two firms belonging to the same industry

b. Cross-case literal replication à two unskilled labor-driven industries with low complexity

c. Cross-case theoretical replication à one technology and skilled labor-driven industry (against the 2 labor-driven industries)

3. Geography

a. Cross-case theoretical replication à at least one firm from each of the three major economic regions.

b. Cross-case literal replication à two couples of cross-industry firms from the same country (Japan and Germany).

(31)

3.5 Data collection and analytical architecture

The methodology designed to support the working proposition is articulated into three stages: firm’s subsidiaries have to be divided into categories according to their geographic location and contribution to the firm’s value production; the different geographic levels have to be classified into categories according to the primary ways they are expected to influence MNE’s locational choices; a strategic scorecard has to be outlined in order to attribute to the locational patterns one or more AAA strategies. These schemes have to be thereupon populated with data on all subsidiaries of the six firms and on indicators involved in the categorization of geographical aggregates.

MNE Subsidiaries

Geographical

Level Reference Group Categ. Criteria

SU PPO R T MA N U F A C T U R E SA L E S à Regions AF (Africa) Home / Host AM (Americas) AP (Asia-Pacific) EU (Europe) à Country Upstream à Method 1 or 2 LOW MEDIUM HIGH Downstream à Method A or B LOW MEDIUM HIGH à Subnational City Tier 1 Tier 2 Tier 3 Specialized Industrial Cluster Yes / No

Table 2 – Method of allocation of the three types of subsidiaries into multilevel geographies, divided into categories according to their attractiveness for certain types of subsidiaries (source: author)

The ultimate objective of the methodology is to systematically associate the location choice of firms with a specific AAA strategy. To make this possible, it is necessary to discern the complexity of the value-creation processes of firms and the extent to which certain locations might be more or less suitable places for these value-value creation activities. So firstly, some specific methods of value creations have been identified. Secondly, the attractiveness of different locations to specific phases of the value-creation has been assessed. Therefore, locations are differentiated not just by their geographical layer (namely, regional, national, subnational), but also by their suitability to certain types of MNE activities.

Thus, the first phase concerns the classification of subsidiaries according to the method they contribute to a firm’s value. They might alternatively represent retailing or distribution terminals,