Simona Iammarino and Philip McCann
One of the crucial aspects of the current phase of economic globalization lies in the new modes of creating and diffusing new knowledge and technology. The central role played by contemporary multinational enterprises (MNEs) in such processes has been analyzed in a variety of scholarly perspectives, and emphasis has been placed on the metamorphosis of the MNE from mere “vehicle” of technical knowledge to “creator” of new technology (e.g., Cantwell 1994; Archibugi and Iammarino 2002; Ietto-Gillies, this volume, Chapter 6). MNEs are today the largest source of technology generation, transfer, and diffusion in the world. In comparison with all other firm types, the share of new technologies produced globally by MNEs is increasing; they are associated with a higher probability of entry into new and cutting-edge technological fields; they account for the great bulk of expenditure on knowledge-creating and skill-enhancing activities, and of trade in technology and technology-intensive products (UNCTAD 2009, 2013). MNE access to a broad variety of sources of new knowledge, both intra- and inter-firm, provides immense opportunities to acquire new competitive advantages for both the firm itself and all the actors involved in its networks.
On the other hand, such MNE networks of internal and external sources of knowledge and information have obvious geographical manifestations. Recent years have seen economic debates directing increasing attention to the phenomena of spatial concentration, and one result has been an increasing body of evidence on the spatially bounded character of knowledge creation and diffusion processes. Such a boundedness is a major explanation for the emergence and reinforcement of core-periphery forces and regional divergence, especially in the face of processes of economic integration and globalization (e.g., Fagerberg, Verspagen, and Caniëls 1997; Rodríguez-Pose 1998; Verspagen and Caniëls 2001; Rodríguez-Pose and Crescenzi 2008; Crescenzi and Rodríguez-Pose 2009; Paci and Usai 2009; Rodríguez-Pose and Fitjar 2013). Indeed, the spatial proximity of firms and people within individual regions has become progressively more important in scholarly thinking as a source of knowledge spillovers and agglomeration economies in general (see Lorenzen and Mudambi, this volume, Chapter 10). Dominant urban agglomerations appear to become increasingly specialized within the global economic system (e.g., Gordon 2002; Storper 2013), so that different localities are likely to be progressively more differentiated in terms of their competitive advantages.
This chapter argues that, in the current phase of economic globalization, geography and space are increasingly important for MNEs, and in turn MNEs are progressively more important for geography: the pivot on which this relationship turns is the creation, diffusion, and management of new knowledge and technology. The emergence of a new set of relationships between MNE organization and control of intra-firm and inter-firm innovation networks, and the role of city-regions as knowledge sources, have led to increasingly differentiated geographies across all parts of the world. The chapter is divided into five sections below. The first section revises some seminal contributions that have, ante litteram, addressed issues of MNEs, innovation and technology, and geographical hierarchies of places in the world economy. The second section highlights the organizational changes that have occurred in the contemporary multinational corporation and its internal and external networks, while the third section connects such changes with different spatial configurations through the lenses of knowledge and technological properties. The fourth section critically considers the role of cities, in particular the so-called global cities, in MNE innovation networks. The final section summarizes the main argument and offers some concluding remarks.
The major research questions in the study of MNEs have typically focused on determinants – why, where, and how firms become multinational – and effects – changes both in the sending and receiving economies involved in MNE operations, and in the MNE firm itself. Although one of the core questions regarding MNEs is precisely that of where their different operations take place, the explicit spatial dimension of MNEs has so far been largely neglected in the literature (Iammarino and McCann 2013). The increasing significance of specific geographies at various scales of analysis, from world regions to nations and, even more importantly, subnational regions, cities, or other typologies of spatially localized systems, has also rendered problematic the traditional scholarly separation between causes and impacts of multinational activities, as the valuation of MNE effects is inherently connected to the determinants of their existence (Ietto-Gillies 2005, 2012). Rather surprisingly, in spite of the ground-breaking insights provided long ago by scholars such as Stephen Hymer and Raymond Vernon, economics, international business, and economic geography theories have not explicitly and systematically identified the distinctive geography of MNEs activity, whether with respect to the process of becoming multinational, or with reference to the outcome of their investments (McCann and Mudambi 2004). We will therefore recall here the major intuitions of these two authors – nowadays perceived as classical in this field of study – that should have paved the way for a more careful observation of multinational corporations and space, which emerged only much more recently in scholarly thinking.
In Hymer’s first argument (published in 1976, though presented in 1960) the focus of the analysis is on the investor’s control afforded by a common ownership and organizational system, which ensures the exploitation of a unique set of firm-specific advantages. Although the initial analytical approach makes no serious attempt to explain the geographical spread of production, a relevant implication for location is found in Hymer’s pioneering recognition that foreign direct investment tends to be concentrated in certain industries across countries, rather than in specific countries across industries. Later on, drawing both on the historical account given by Alfred Chandler (1959, 1962) and on the “creative destruction” arguments of Josef Schumpeter (1934), Hymer formulates his “law of increasing firm size” (1970), identifying the evolutionary path of the modern corporation: from the competitive Marshallian firm, with a single function, industry, and ownership, to the large national oligopolistic corporation, vertically integrated with dispersed ownership, to the modern conglomerate enterprise, multidivisional with cross-country affiliates all coordinated by a centralized “corporate brain” (Hymer 1970: 442). Rehearsing Chandler and his co-author Fritz Redlich (1962), Hymer explains that during the evolutionary process of the firm toward the worldwide enterprise – the modern MNE – a pyramidal structure of three levels of business administration and decision-making could be detected. Level III, the lowest and widest, dealing with ordinary daily operations which are connected to production activities; the intermediate Level II, generated by the separation of the head office functions from the field offices of the firm, and controlling Level III as well as coordinating the management at that lower level; Level I, the top level of management, responsible for setting the goals and growth strategies – therefore research and innovation investments – of the whole firm. Thus, while in the case of the (uni)national corporation Level II and I are split off from Level III, in the modern MNE Level I and II become completely separated, with the top Level I acting as the “corporate brain” responsible for the firm’s overall strategy.
This hierarchy of corporate decision-making is further discussed in subsequent work, in which the “law of increasing firm size” is merged with the “law of uneven development” (see also Pitelis 1991, 2005). In particular, Hymer (1972) suggests a “correspondence principle,” that is, the existence of a direct relationship between the centralization of power and control within the MNE and the centralization of power and control in the world economy. The “correspondence principle” thereby highlights the correlation between corporate growth and evolution, on the one hand, and uneven development across different spatial locations, on the other. It is here, in the “correspondence principle,” that the critical connection between the MNE’s ownership advantages, strategic behavior, and specific geographies lies.
Hymer suggests that, by applying the location theory to the Chandler-Redlich model, the “spatial dimension of the corporate hierarchy” could be effectively analyzed (1972: 122). The pyramidal structure of corporate control centralization translates directly into a hierarchical structure of geographical locations. Some of these locations are heavily dependent on others, underlying the uneven spatial structure of economic development. Therefore, Level III activities will tend to be relatively evenly spread and distributed worldwide, according to the attractiveness of resources such as labor, markets, and raw materials; it is this phenomenon which diffuses industrialization across less advanced economies. In contrast, Level II activities will tend to be concentrated in large city-regions, as they require primarily white-collar labor, as well as sufficient communication and information systems. At this level, MNEs from different industries will locate in the same cities, thereby giving rise to a strong spatial and functional concentration at the subnational level. Level I activities will be even more geographically clustered, as they need to be close to sufficient supplies of highly specialized human capital and services, capital markets, media, and governments: the provision and exploitation of these high-profile service relationships require strong connectivity, both locally and with the rest of the world.
Hence, the highest-level functions of the leading MNEs will almost all be located in the world’s major global cities, which themselves will be “surrounded by regional subcapitals” (Hymer 1970: 446). Hymer argues that, eventually, the major global cities which will be the home to the core headquarter functions and high-level strategic planning of the world’s MNEs will be New York, London, Paris, Bonn, Tokyo, Moscow, and Beijing (1972: 124). From these highest levels, Hymer envisages that a geographical hierarchy will characterize the spread of MNE operations, with “intermediate” and “lower-level” activities distributed across lower-tier urban centers and regions. In his words: “geographical specialization will come to reflect the hierarchy of corporate decision making, and the occupational distribution of labor in a city or region will depend upon its function in the international economic system” (1972: 124).
Much of the later international business literature has explicitly and implicitly borrowed from Hymer’s analysis (see Dunning and Pitelis 2008), and his arguments have also been somewhat influential in some geography and location studies (e.g., Dicken 1977; Norton 1979, 1987). Curiously, however, in spite of a few studies (e.g., Young, Hood, and Peters 1994; Bailey and Driffield 2002; Pitelis 2002a, 2002b, 2005; Kottaridi 2005) explicitly acknowledging Hymer’s key importance in explaining the relationship between MNE activity and subnational regional development, the international business literature itself has largely overlooked the ante litteram relevance attributed by Hymer to the interplay between spatial (particularly urban), organizational, and industrial structures, and information and communication systems, all of which underlie the locational choice and the economic geography of MNEs.
To some extent as in Hymer’s thought, in Vernon’s product life-cycle (PLC) analysis the geographical location of international production follows a hierarchical pattern which is strictly linked to the ownership advantages that characterize each stage of the oligopolistic structure (Ietto-Gillies 2012). In the innovation-based oligopoly, the location of production is naturally placed in the economy where the innovation process firstly initiated. In this early stage, when technology is unstable, the demand for the new product is uncertain, technological capacity is vital, and location occurs in the metropolitan centers of advanced countries. This observation stems from the earlier work of Vernon on agglomeration economies in metropolitan regions such as New York (Vernon 1957, 1960). Critically, and in parallel with Hymer, in this work he stresses that, among the main causes underlying the growth of the core urban areas, a crucial role is played by the availability of a variety of business services, which previously were provided internally by the firm’s business functions (Vernon 1957: 17). In contrast, in mature oligopolies advantages based on economies of scale, transport, and marketing lead to a strong concentration of investment either in the home country of the MNE or in similarly advanced economies, spreading the location of the firm’s investments out of the initial agglomerations (Vernon 1966, 1974, 1979). The mature or standardized stage of the product cycle – characterized by stable technology, long production runs, strong price competition, and high intensity of unskilled labor – offers developing economies the opportunity to successfully enter the international competitive environment by attracting MNEs from advanced locations into their emerging local industries (see also Hirsch 1967).
The PLC model has also seen wide applications in regional and urban analysis, basically predicting that the concentration of early stages of product development will be in urban areas, where high-skilled labor, external economies, and top management are easily available, followed by subsequent shifts of standardized production to more rural and lower-cost regions (e.g., Norton and Rees 1979). However, in this literature the structural and technological aspects of Vernon’s arguments as they relate to MNEs are largely ignored. On the other hand, the same simplistic skilled–unskilled dichotomy of the international product cycle model is frequently applied also at the subnational scale, thereby attracting criticism that these approaches have mainly failed to explain the processes of historical change in industry spatial and organizational structures (e.g., Storper 1985; Taylor 1986, 1987; Mack and Schaeffer 1993).
Among the criticisms of the PLC type of MNE behavior, two in particular seem to be most relevant here. First, the models tend to display a simplified and deterministic view of the innovation process, which is assumed to be a linear and uniform sequence from invention to a marketable product to standardization, and where skills and innovative capabilities are seen as effortlessly transferable across geographical and institutional contexts (Cantwell 1995). Second, the locational issues arising in each of the stages of the PLC model following the introduction of the new product have been disregarded. In many PLC models the spatial determinants of foreign and local firms’ behavior are reduced mainly to labor cost differences, thereby largely ignoring the evolutionary transformation of both the firm’s internal relationship between power, ownership, and control, and the external connections with the economic environment. This “ambiguous enterprise context” (Taylor 1987: 82) does not allow us to distinguish between global and local competition; nor does it shed light on the intra-firm versus inter-firm relationships which are crucial to identify the specific spatial dimension of both industry and technological cycles.
Yet, although the PLC model has been given a space-specific characterization by authors who have followed Vernon’s own extensive work on location and urban and regional economics (e.g., Vernon 1957, 1959, 1960, 1966, 1991), such aspects have been surprisingly overlooked in the international trade and business literature, as well as in the economic geography of the MNEs. In fact, in spite of Vernon’s path-breaking analysis of the causes of urban concentration and geographical spread of multiplant activities which takes place across boundaries of product lines, the subsequent Vernon-inspired literature still tended to treat MNE location behavior using the highly unspecific geographical dichotomy between skilled capital-abundant advanced (national) economies and unskilled labor-abundant developing countries. Vernon’s emphasis on the spatial shifts involved in industry development cycles, and the changing relevance of external economies during such cycles, should have called for a more explicit incorporation of locational issues in the explanation of MNEs’ operations (Iammarino and McCann 2013).
In spite of the criticisms attracted over time, the work of both Hymer and Vernon has been considered seminal in the analysis of MNEs for two reasons. First, they both introduce market imperfections and dynamic elements into the analysis, such as changes in demand structures, innovation and technological advantages, and information lags. Second, they both recognize the importance of ownership comparative advantages – particularly those based on superior knowledge – in explaining multinational behavior. As discussed above, both Hymer’s and Vernon’s pioneer works have strong micro-foundations, and as such provide crucial connections between the structure and behavior of MNEs and the subnational spatial organization of their economic and innovative activities.
In sharp contrast with the visions offered by Hymer’s and Vernon’s theories, since the 1990s and the acceleration of economic globalization the “flat world” hypothesis – the well-known claim that the world has become in some economic and social sense mostly flat due to the disappearance of barriers and transaction costs across space – has gained support in both scholarly and political circles (e.g., O’Brien 1992; Ohmae 1995; Cairncross 1997; Friedman 2005). Such a claim predicates that advances in technology and business organization have enabled actors anywhere in the world, be they consumers, investors, entrepreneurs, or firms, to link up and do business with others anywhere with an ease and accessibility that obviates the advantages that once accrued to those who were attached to favored localities, regions, or nations (Guy 2009; Iammarino and McCann 2013). The “flat world” thesis implies that economic activities can be done just as well in India as America, by small firms as by large firms, because we live in the age of the network organization, and these networks are global.
The hypothesis of flatness – harshly criticized in the last decade or so, but far more in academic circles than in policy ones – rests on two big assertions, one of which is geographical and the other one organizational. Many have claimed that distance is becoming unimportant, and many others have argued that big MNEs are dinosaurs while networked small and medium-sized enterprises (SMEs) are the mammals and flowering plants that will take their place (Iammarino and McCann 2013). However, there is ample evidence that the removal of barriers to international trade, investment, and communication can actually make geographical proximity more, rather than less, important. Many economic activities benefit – still, if not more – from co-location, whether for reasons of simple logistics, shared inputs, or benefits of face-to-face contact (Arita and McCann 2000). When barriers to cross-border or long-distance trade are reduced, some agglomerations – local, regional, or even national – may find that their comparative advantages are strengthened, producing fast economic growth. The integrated world becomes a more differentiated one, and proximity to places and cities such as Silicon Valley, the City of London, Hong Kong, or Guangzhou becomes more important than ever. Rather than becoming flatter, in many aspects globalization is nowadays shaping the world more unevenly, more curved (McCann 2008), or more spiky (e.g., Florida 2005; Rodríguez-Pose and Crescenzi 2008; Rodríguez-Pose and Fitjar 2013). In formal economics these changes are represented by the increasing returns models of the new trade theory and the new economic geography (see, for all, Krugman 1991; Fujita, Krugman and Venables 1999).
Turning to the organizational claim, it is true that networks have grown, as the reduction of transaction costs has made it possible for firms to outsource many functions. Yet, innovation and flexibility can thrive under administrative guidance just as mass production did: rumors of the death of the large corporation have been greatly exaggerated (Piore 1994). The picture of flat global networks is problematic for the simple fact that flatness and global reach are often in conflict. The global reach of production and innovation networks is to a substantial extent an administrative accomplishment: distance has been shrunk not only by improved technology, but also by corporate organization. Managerial hierarchies – or, according to the recent trends in corporate structure, heterarchies – and standardized procedures are, among their other functions, distance-spanning devices. Conversely, the most robust and innovative flat networks involve disproportionately local actors. One role played by large MNEs is to link such local clusters and networks to the larger world by acting as “gatekeepers” – extending them beyond national boundaries (e.g., Amin and Thrift 1992; Enright 2000; Giuliani 2011). This is an increasingly important role because the reductions of transactions, communication, and transportation costs, where they are evident, are not accessible everywhere and to any firm, while they are relatively readily available to large MNEs.
On the other hand, the stable relationship between ownership and control, which has long been understood as being problematic when looked at in terms of the divide between investor and executives, has been disturbed both along the supply chain and within the corporation (Ietto-Gillies 2005). In outsourcing and offshoring strategies, ownership changes but control of the value chain activities is largely retained through various means of pressure on the suppliers and their competitive bidding (e.g., narrow transfers of technology, strict product specifications, tight supplying schedules, etc.) (UNCTAD 2013). Conversely, in vertical integration strategies ownership is not altered, but the distribution of control within the MNE can vary greatly, with different degrees of autonomy of the affiliates and subsidiaries that can lead to intra-firm competition and even to various degrees of restraint in the powers of the central headquarters of the MNE (e.g., Birkinshaw and Hood 1998, 2000; Birkinshaw, Hood, and Young 2005).
Such organizational changes have had huge implications for the location choices and innovation advantages of MNEs, which are increasingly dependent on the balance between technological competencies and capabilities within and outside the firm, or on the integration of various sources of knowledge which are internal and external to the firm. Also, different geographies have emerged in relation to different types of multinational hierarchical versus heterarchical structural forms and internal organization, such as integration versus externalization through outsourcing and offshoring strategies, the unbundling of headquarters and core functions such as R&D, or centralized/decentralized control of strategic assets within the firm (Desai 2009).
The identification of the MNE’s spatial location, and its characteristics with respect to geography, industry structure, knowledge, and technology flows, is a result of complex interactions between firm(s), industry, organizational, and knowledge characteristics. The simple host–home country dichotomy has long become insufficient. This is particularly so not only with respect to the economically advanced parts of the world, but also increasingly in all the main cities and urban regions, where “host” and “home” may actually overlap to a great extent: the locations which are host to a major presence of MNEs are also those which are most connected by outward linkages. On the other hand, the location of corporate headquarters of large MNEs nowadays has little geographical connection with the location base for specific business units and operations: a new set of relationships between the location of knowledge assets, the role of city-regions as knowledge sources, and the spatial behavior of MNEs has emerged.
Both Hymer and Vernon depart from the prevailing neoclassical tradition in that they treat innovation as an endogenous phenomenon conditioned on structural issues – MNE innovation-based ownership advantages – rather than as an exogenous phenomenon dependent on the efficiency of market clearing. Hymer points out that multinational behavior implies the existence of international flows of groups of activities and resources, including technology, capabilities, skills, and entrepreneurship. An important qualification he made, also preempting Krugman (1991), is that while growth in the hinterland or periphery means growth in the centre, the reverse does not necessarily hold.
On the other hand, the two major criticisms of the PLC-based explanations of international production mentioned above have important spatial implications for our argument here. First, on the basis of the evolution of the contemporary world economy, ownership advantages are to be attributed to firms, rather than countries, thus making the geographical origin of MNEs much less predetermined by the national level. This point, which was seemingly captured in Vernon’s original framework, has been highlighted by both regional economics (e.g., Taylor 1986, 1987) and international business (e.g., Cantwell 1995) scholars who attribute the limitations of the PLC model to an inadequate conceptualization of both the firm and technological progress. Second, observation suggests that agglomeration forces have attracted MNE activities – even high value-added ones – to particular locations in both advanced and emerging economies, thus making the geographical destination of MNEs far less dependent on purely cost-based and intrinsically static hierarchical patterns. Indeed, this latter observation seems to point to the opposite side of the globalization argument: skills formation and knowledge accumulation processes are ever more dependent on sources that are external to any single firm (however large and multinational it may be) and are highly spatially situated (Storper 1997, 2013). 1
In the global economic race of the twenty-first century MNEs compete primarily on the basis of their internalized knowledge and technological assets as well as of how these are employed to engage with the intangible assets of the geographical localities in which they operate. In line with Hymer’s and Vernon’s predictions, one of the dominant features of the current wave of globalization is the emergence of a new set of relationships between MNE knowledge assets and the role of particular city-regions as knowledge sources. Although the emergence of these relationships has been taking place gradually over the last decades, in the most recent years their nature and ordering have become more apparent, as their manifested trends appear to be accelerating.
In terms of the geography of MNEs, the reality is presumably not far from that envisaged by Hymer, in his emphasis on a locational hierarchy based on ownership and control, and by Vernon (1991) in his later vision of multi-product and multi-technology industry dynamics; and possibly even more diverse and in rapid evolution than what is suggested by the “world cities” (Friedmann 1986) or “global cities” (Sassen 2001, 2002) hypotheses in urban studies, 2 applied also in the international business literature (e.g., Dunning and Norman 1983; Nachum and Wymbs 2005; Goerzen, Asmussen, and Nielsen 2013). Importantly, both these strands of urban research stress that it is not simply a matter of locating MNE facilities in the largest cities or urban regions: other factors are just as important for economic growth as is urban scale, and different spatial configurations, regions, cities, or localized production systems, which are open repositories of knowledge creation and exchanges, will offer advantages for different firms, whether MNEs or not.
By reinterpreting a classification of spatial typologies based on transaction costs (Gordon and McCann 2000; McCann, Arita, and Gordon 2002; McCann and Sheppard 2003) in terms of technology, knowledge, and innovation features (Table 14.1), we have elsewhere shown (Iammarino and McCann 2006, 2010, 2013) that it is possible to uncover more subtle and detailed aspects of firm location behavior. In particular, in the case of MNEs, any simply assumed correlations between localized knowledge spillovers, spatial concentration of activity, and the advantages of a particular location for firm investment, do not necessarily hold. For example, in the case of the first spatial configuration of our technology-based classification reported in Table 14.1 – pure agglomeration, which corresponds by and large to the spatial model of global city – the knowledge base is primarily explicit and codified, available to any local actor and organization, and generated outside firms’ boundaries, being largely created in other private and public organizations. Variety and promiscuity are distinctive features of cities, particularly global ones, in the sense of there being an absence of long-term loyalty between agents. The combination of different streams of knowledge therefore occurs across a broad range of sectors (Jacobs-type externalities), and individual and organizational innovation linkages or relations are unpredictable due to the low degree of knowledge cumulativeness. In other words, while the overall links between innovation and cities are undoubtedly very strong (e.g., Acs 2002), the arguments relating MNEs to knowledge spillovers in such contexts are rather more complex: the knowledge objectives and the innovation performance of MNEs are not necessarily centered on cities (Simmie and Sennet 1999).
Table 14.1 Technological classification of spatial types.
Social network | ||||
Pure agglomeration | Industrial complex | Competence-based SN | Trust-led SN | |
Nature of technical knowledge | Codified, explicit and mobile, transmitted by way of information | Specific, systemic, routinized, R&D-intensive; based on non-transferable experience | Tacit, generic, non-systemic, sticky and leaky; transmitted within cognitive networks | Codified, sticky, mature; transmitted within geographically localized networks |
Technological trajectory | Oriented to processes, problem-solving, organizational innovation | Oriented to complex products, cost-cutting | Oriented to radically new products | Oriented to processes, customer-driven |
Dynamics | Stochastic | Strategic | Disruptive, uncertain | Incremental |
Sources of innovation | External to the firm, multi-sectoral | Internal to the firm and the sector, institutionalized R&D in incumbents, relevance of basic science | External to the firm and the sector, relevance of applied science and cooperative R&D | External to the firm, internal to the sector, relevance of machinery acquisition |
Appropriability of innovation returns (and market structure) | Low (perfect or monopolistic competition); based on standards, copyright, design | High (oligopolistic competition); based on patents, secrecy | Low (low entry barriers); based on patents, copyright | Low (collaborative monopolistic competition); based on trademarks |
Technological opportunities | Variable | Medium/low | Very high, uncertain | Low |
Cumulativeness | Low | High | Low | High |
Knowledge base | Broad, diversified | Narrow, targeted, specialized | Broad, research-based | Narrow, specialized along the filière |
Modes of governance | Market | Corporate hierarchies | Relational and cognitive networks | Social and historical networks |
Examples of industrial specialization | Finance, banking, insurance, business services, retailing | Steel, chemicals, automotive, pharmaceuticals, machine tools, medical instruments, ICT hardware | High-tech, general purpose technologies | Customized traditional goods, textiles, footwear, furniture, tourism |
Example of cluster | Global cities; Silicon Valley (California) since the 1990s | Silicon Glen (Scottish Electronics Industry) | Silicon Fen (Cambridge, UK) | Italian industrial districts in made in Italy industries |
Pavitt classification | Information Intensive, Supplier Dominated Firms | Production Intensive Firms (Scale Intensive and Specialized Suppliers) | Science-Based Firms | Supplier Dominated Firms |
In contrast, the second spatial type – the industrial complex – is often characterized by regions or city regions with a highly specialized industrial structure. For modes of innovation which are primarily based on major R&D and capital investment and long development lead times, the industrial complex may provide the most suitable environment for technology creation and experimentation. Where evident, the composite and advanced local knowledge-production basis is also likely to display a strong concentration at the spatial level. The industrial complex structure is also often a suitable structure for the development of multilateral networks of dissimilar but complementary relations between MNEs and local actors (Patrucco 2003). Yet, in the industries in which the host region is technologically strong, the major presence of indigenous firms tends to deter MNEs in the same industry from carrying out considerable innovation activities in the primary technological fields related to the local industry strengths (Cooke 2001). As such, when MNEs choose to invest in these types of regions it generally implies that the focus of the local relations is mostly related to supply chain issues, rather than technology or knowledge sourcing. However, MNEs operating in other industries will still be attracted there for innovation and knowledge purposes in those technologies whose lines of development represent diversification from the primary technologies of their own industry (Cantwell and Kosmopoulou 2002).
There is also a fundamental difference in the particular modes of innovation governance between the two social network spatial types reported in Table 14.1 (Simmie et al. 2004; von Tunzelmann 2003, 2009). In the competence-based social network model, high technological opportunities come primarily from sources outside the firm and the industry, such as university academic research. In this kind of technological environment the type of knowledge tends to be both generic and non-systemic, with high rates of market entry and exit, a strong degree of volatility of market shares, and low levels of market concentration. In such an environment, the tacit and sticky nature of knowledge requires geographical proximity, particularly in the early stages of the industry or cluster life-cycle. The openness of the local system, the relatively “leaky” character of new knowledge and consequently the high potential for spillovers, and the related emergence of new rules, standards, blueprints, and verification procedures, all point to the importance of external sources of technical knowledge. Innovations therefore mainly come from knowledge that does not have a routinized nature, with new firm start-ups playing an important role, and SMEs accounting for a large share of innovative activity. There is normally either no role or at best only a limited presence of large MNEs in situ in the first stages of regional development. In later stages, however, where both industry and spatial patterns have reached a certain stability, such locations may become highly attractive to MNEs. In the trust-led social network, instead, knowledge is largely codified and mature and develops along stable trajectories which are mainly oriented toward process innovation. Knowledge is transmitted essentially by way of informal personal contacts which are strongly embedded in a common cultural matrix, comprised of elements of social and political lobbying, and intense backward and forward linkages, sometimes intensified further by the presence of MNEs. As such, the structure of vertical and horizontal linkages between MNEs and local firms may be crucial in determining the nature and extent of local spillovers. Even in the same industry, regions may be characterized either by highly vertically integrated firms, or by stage production with significant subcontracting linkages. The trust relationships and the role of reputation will differ considerably in such different spatial and industrial settings (Guy 2009).
Some empirical analyses have shown that inter-industry knowledge spillovers are likely to become more intense in centers of technological excellence where spillovers seem to operate mainly through exchanges in and around core technological systems (Crozet, Mayer, and Mucchielli 2004). This is the case where spillovers are primarily rooted in “general purpose technologies” such as, for instance, background engineering, mechanical methods, new materials, electronics, and ICT, and the additional spillovers which these create between actors in quite separate alternative fields of specialization. These types of centers of excellence exhibit inter-industry technology spillovers, and are most likely to be classified either as pure agglomerations or competence-based social networks. However, as we have seen, these two different spatial types tend to offer different possibilities with regard to the role of MNEs, and the contribution that they make to local spillovers. Yet, it is these two typologies which tend to experience a faster process of convergence between old and new technologies, and potentially therefore also a greater degree of competitiveness. 3
In addition to the characteristics of knowledge based on industry and technological regimes – on which our simple taxonomical exercise is based – there is new substantial evidence of increasing spatial agglomeration at the functional level, particularly visible in service industries which are far more affected by functional than by sectoral features (e.g., Defever 2006; Crescenzi, Pietrobelli, and Rabellotti 2013; Goerzen et al. 2013; Ascani, Crescenzi, and Iammarino 2014). This may provide an additional angle to look in dynamic terms at the taxonomy described above.
During the twentieth century there were qualitative changes in the role of cities in the industrialized world which favored the competitive advantages associated with cities being centers of knowledge; these changes are now occurring increasingly in developing parts of the world (see Florida and Mellander, this volume, Chapter 15). The clues as to why particular cities and regions are highly productive lie in the types of innovation system that characterize them. The vast literature on innovation systems, with its strong emphasis on the links between the actors, organizations, and institutions involved in knowledge creation and diffusion – business firms, universities and research institutes, industry associations, government agencies, etc. – has discussed the main determinants of system variety at various levels of geography (see Iammarino and McCann 2013 for a review).
In order to generate the required returns to their knowledge investments, firms – and the other actors and institutions within the urban or regional system – must capture markets which extend well beyond the borders of their own country. Traditionally these returns were generated by exports, but the increasing engagement facilitated by international investment has become far more important. International flows, and particularly flows of knowledge, are increasingly understood as being bi-directional or multi-directional, with concepts such as “openness” and “connectivity” coming to replace terms such as “inward” and “outward” flows.
While clear trends toward global regionalism, rather than simply globalization, are emerging, the integration processes taking place between larger groups of both rich and also emerging economies offer greater rewards to MNEs than ever before from exploiting the possibilities for better affiliate-specific location matching, as well as coordination between dispersed subsidiaries. At the same time, the higher demands for timeliness, the greater requirements for higher frequency transactions, and the increased preferences for customization and variety, all tend to raise the distance costs associated with knowledge-related transactions (McCann 2007) and thereby the opportunity costs of sub-optimal MNE affiliate locations.
Large cities and city-regions provide growing opportunities for a wide range of sharing, sorting, and matching mechanisms and institutions, which increase the overall efficiency of the local activities. Although there is still clearly a very important role for large cities in the industrialized world in terms of driving productivity and innovativeness (Rosenthal and Strange 2004), when we consider the new context of globalization and the potential advantages of different spatial types for MNEs, the relationship between city size and productivity – the main, and often only, indicator available for measuring economic growth and attractiveness – becomes much less straightforward.
Interestingly, the majority of the world’s highest productivity cities are not what the OECD (2006) classifies as “mega-cities” of over 7 million inhabitants. For OECD cities of over 1.25 million inhabitants, there is only a very weak cross-sectional link between city per capita productivity and population, which if anything is slightly negative (OECD 2006). However, the productivity advantages of very large cities appear to be relatively more important for lower-income and transition economies within the OECD than for the most advanced ones (OECD 2006). More generally, while estimates of the relationship between city size and productivity within individual countries make some sense, estimations across countries are far more complex because of the spatial, structural, and institutional heterogeneity at the national level. As well as an ∩-shaped relationship between city size and productivity, across the OECD it is generally the case that the higher-income cities are actually outgrowing lower-income cities, irrespective of population scale (OECD 2006). This suggests that other characteristics of cities are also just as important for economic performance and attractiveness as scale, and in particular the crucial role that cities or urban regions can play as centers of knowledge (e.g., Storper 2013), creativity (e.g., Florida 2005), and innovation (e.g., Acs 2002).
Interdisciplinary seminal studies on cities and economic growth (e.g., Hall 1966; Jacobs 1969; Hymer 1972; Friedmann 1986; Sassen 1991; Henderson, Kuncoro, and Turner 1995) have shown that economic growth at the international scale is being increasingly dominated by networks of particular major urban centers and regions (e.g., Borja et al. 1997; Sassen 2001; Scott 2001; Scott et al. 2001; Derudder et al. 2003; Button et al. 2006). As mentioned above, there is also abundant evidence which examines the role played by these global cities as the principal location bases for the largest MNEs (e.g., Sassen 1994, 2002; Knox and Taylor 1995; Enright 2000; Taylor 2004; Bel and Fageda 2008; Verbeke, Li, and Goerzen 2009). The analysis of global cities suggests that, in the current phase of globalization, the links among such global cores are a key determinant of the city-region’s performance.
The global urban centers are locations which not only exhibit significant agglomeration advantages, but which also primarily interact with other similar globally-oriented cities in other countries, rather than with smaller urban centers and regions within their national boundaries or even within the same macro-region. In sectors such as financial services (COL 2009; MasterCard 2008) there is clear evidence that global markets are increasingly dominated by power networks of a few global urban centers such as London, Paris, Tokyo, Sydney, and New York (Sassen 2002). The increasing relative dominance of these global cities appears to be associated with strong transnational regulatory institutions and high density of information technology assets (e.g., Sassen 2002; Derudder et al. 2003; Button et al. 2006; Taylor et al. 2011); their importance as major nodes (Limtanakool, Schwanen, and Dijst 2007) within international networks is also reinforced by their role as hubs within the global air (Burghouwt 2005), rail, and marine transportation systems (Leinbach and Capineri 2007). These observations are also supported by the findings of Ni and Kresl (2010) who found that the most important element in the competitiveness rankings for global cities is connectivity, rather than urban size or structure.
There are a few points that it is worth stressing about the correspondence between global cities and MNE networks. First, as shown by the urban literature, the definition of “global cities” is not related to the city size, but rather to its degree of “connectivity,” “openness,” and “accessibility.” As yet, and despite the emergence of global hubs in some emerging economies, most of the world’s largest cities located in developing countries still do not exhibit the same information, financial, transportation, and management bi-directional flows – together with comparable local institutional settings – that the established global cities in the world exhibit.
Second, the emphasis on the role played by the presence of MNEs and their innovation networks in making cities “global” has probably been overstated, largely neglecting the crucial evolution of the relationship between corporate power and control. The centralization of political and economic power is certainly concentrated in the few cities of Hymer’s Level I (1972): changes over less than a decade (2006–2014) in the location of the headquarters of Fortune Global 500 corporations show a substantial mobility in the global cities hosting more than five of such companies (McKinsey Global Institute 2013).
However, as argued above, MNE intra-firm and inter-firm networks display very different geographies according to organizational structure, degree of integration versus externalization, functional coordination, balance between control and autonomy of company’s units. Such complex networks – and the variety of spatial typologies that can arise by combining one or more of those types stylized in our taxonomy above – seem to offer a rather different picture of the geography of the MNE than that offered on the basis of the location of headquarters of large corporations. Indeed, some recent evidence from the empirical literature on the economic geography of MNEs indicates that, in the cross-borders co-location of the different stages of the value chain of MNE affiliates in the context of the European Union, MNE headquarters do not display any pull effect over the location of any other MNE function (Defever 2006; Ascani et al. 2014). Goerzen et al. (2013) have shown that competence-exploiting and competence-creating (Cantwell and Mudambi 2005) activities of MNEs follow very different spatial patterns: while the first tend to agglomerate in global cities, the latter, far more valuable for local economic development, tend instead to concentrate in metropolitan (or smaller) core regions.
These observations cast further doubt on the identification of the connection between global cities as the main (or sole) geographical structure of MNE innovation and production networks. It is interesting to note that, among the top 25 cities listed by the McKinsey Global Institute in 2013 as locations of the world’s largest MNE headquarters (Table 14.2), five correspond to the top 25 cities of emerging economies (Table 14.3), and others in the latter list – from Mumbai to Bangkok, Johannesburg, and Buenos Aires – are very likely to enter the world top 25 in the near future. At the same time, though, the ranking of emerging markets’ cities as locations of foreign subsidiaries of large MNEs shows rather different positions, with eight cities in the top 25 for MNE large subsidiaries not included in the ranking of locations for headquarters.
Table 14.2 Top 25 cities ranked by number of large companies’ global HQs 2010.
Source: Adapted from McKinsey Global Institute (2013).
Rank | City | Large companies’ global HQs | Total revenue ($bn) | Average revenue per company ($bn) |
1 | Tokyo | 613 | 5,231 | 8.5 |
2 | New York | 217 | 1,964 | 9.0 |
3 | London | 193 | 1,924 | 10.0 |
4 | Osaka | 174 | 1,028 | 5.9 |
5 | Paris | 168 | 2,785 | 16.6 |
6 | Beijing | 116 | 2,503 | 21.6 |
7 | Moscow | 115 | 709 | 6.2 |
8 | Seoul | 114 | 1,150 | 10.1 |
9 | Rhein-Ruhr (DE) | 107 | 1,220 | 11.4 |
10 | Chicago | 105 | 695 | 6.6 |
11 | Hong Kong | 96 | 468 | 4.9 |
12 | Taipei | 90 | 472 | 5.2 |
13 | Los Angeles | 82 | 422 | 5.1 |
14 | Zurich | 79 | 770 | 9.8 |
15 | Sydney | 75 | 466 | 6.2 |
16 | Stockholm | 74 | 360 | 4.9 |
17 | Houston | 74 | 661 | 8.9 |
18 | Nagoya | 70 | 481 | 6.9 |
19 | Randstad (NL) | 67 | 1,516 | 22.6 |
20 | Singapore | 64 | 343 | 5.4 |
21 | Dallas | 63 | 804 | 12.8 |
22 | Washington DC | 62 | 655 | 10.6 |
23 | Toronto | 61 | 436 | 7.2 |
24 | Munich | 61 | 581 | 9.5 |
25 | Melbourne | 58 | 289 | 5.0 |
Table 14.3 Top 25 emerging countries’ cities ranked by number of large companies’ global HQs and number of large foreign subsidiaries 2010.
Source: Adapted from McKinsey Global Institute (2013).
Rank | City | Global HQs | Rank | City | Large foreign subsidiaries |
1 | Beijing | 116 | 1 | Singapore | 118 |
2 | Moscow | 115 | 2 | São Paulo | 58 |
3 | Hong Kong | 96 | 3 | Mexico City | 42 |
4 | Taipei | 90 | 4 | Moscow | 33 |
5 | Singapore | 64 | 5 | Buenos Aires | 31 |
6 | Mumbai | 57 | 6 | Hong Kong | 27 |
7 | Shanghai | 54 | 7 | Kuala Lumpur | 26 |
8 | São Paulo | 49 | 8 | Bangkok | 23 |
9 | Bangkok | 45 | 9 | Prague | 22 |
10 | Tel Aviv-Jaffa | 43 | 10 | Warsaw | 22 |
11 | Mexico City | 40 | 11 | Istanbul | 21 |
12 | Johannesburg | 37 | 12 | Budapest | 18 |
13 | Istanbul | 36 | 13 | Santiago | 17 |
14 | Kuala Lumpur | 35 | 14 | Beijing | 16 |
15 | Buenos Aires | 31 | 15 | Taipei | 14 |
16 | Santiago | 30 | 16 | Jakarta | 13 |
17 | Shenzhen | 27 | 17 | Rio de Janeiro | 13 |
18 | Delhi | 27 | 18 | Bucharest | 10 |
19 | Jakarta | 25 | 19 | Lima | 10 |
20 | Guangzhou | 22 | 20 | Bogota | 9 |
21 | Hangzhou | 22 | 21 | Caracas | 8 |
22 | Rio de Janeiro | 20 | 22 | Shanghai | 7 |
23 | Riyadh | 19 | 23 | Manila | 7 |
24 | Warsaw | 18 | 24 | Johannesburg | 6 |
25 | Hanoi | 17 | 25 | Curitiba | 6 |
Institutional and technological environments have changed radically since the late 1980s, and multinational corporations, the supposedly inflexible dinosaurs, have changed accordingly, to a much larger extent and rather more quickly than many other types of firm. The modes of international investment, the organization and management of intra-firm vertical and horizontal relationships for production and knowledge generation, the types of affiliation linkages, the diversification and distribution of functions, the integration of subsidiary objectives into the overall goals and strategy of the MNE, have all gone through substantial and rapid changes. These new organizational modes have occurred within MNEs, and are also seen in terms of their external relationships. Increasingly complex and systemic forms of integration of international operations have involved external firms, often SMEs, which are connected through contractual relations to the global production and innovation networks led by large corporations: many of these SMEs have actually become MNEs themselves, in advanced as in emerging economies. Also, the growing degree of complexity in MNE forms and organization has made clear that the boundaries and definitions of multi-product and multi-technology MNEs with respect to individual industrial sectors and technological fields are increasingly blurred (Iammarino and McCann 2013).
While Hymer’s intuition of the top Level I acting as the “corporate brain” responsible for the firm’s overall strategy (Hymer 1970: 442) has proved to be an essential feature of the contemporary multi-divisional multi-locational conglomerate enterprise, what we may observe in more recent times is actually a further splitting off of Level I functions into power and control. The “correspondence principle” suggested by Hymer (1972) has seemingly evolved in the establishment of a direct relationship between the centralization of political and financial power within the MNE and in a few truly global cities in the world economy. At the same time, in response to the organizational and technological requirements of global value chains and knowledge networks management, control has increasingly been decentralized within the corporation and across geography, involving a wider range of spatial locations. In other words, while power (political and institutional linkages, finance, lobbying, alliances, etc.) has certainly tended to agglomerate in global cities, control over MNE functions and operations – including strategic ones such as the generation of new technology and innovation – is gradually delocalized in second, third, and even lower tier cities and regions. The globalization of innovation (Archibugi and Iammarino 2002) is thus characterized by both geographical dispersion and concentration, arguably targeting a range of city-regions that extend beyond global cities.
In the light of both Hymer’s and Vernon’s legacy, the location behavior of MNEs cannot be understood without recourse to detailed and bundled considerations of organization, technology and innovation, institutional context, and, most importantly of all, firms’ ability to access and exploit internal, local, and global knowledge sources. Indeed, Hymer’s “correspondence principle” is still a crucial connection between the centralization of power and decentralization of control within the modern corporation, and the increasingly differentiated spatial hierarchy in contemporary economic systems. On the other hand, Vernon’s crucial observation of the strong relationship between industry life-cycles and spatial shifts helps explain different geographical orders of multi-product and multi-technology MNEs, and the increasing specialization of places.
MNE internal (intra-firm) networks have become relatively flatter over time, with increasing decentralization of control, decision-making, and functions; at the same time, MNE external (inter-firm) networks have spurred spikier geographies and uneven regional development, depending on the variation across urban and regional innovative and institutional capabilities to cash in on the presence of global “gatekeepers” to build new localized absolute and comparative advantages. Such changes could arguably alter the connectivity of cities and regions around the world: the configuration of geographical networks calls definitely for further research efforts, both theoretical and empirical.
18.118.163.250