This Research Bulletin has been published in Economic Geography, 87 (3), (2011), 267-308, under the title 'Sectoral Differentiation and Network Structure Within Contemporary Worldwide Corporate Networks'.
Please refer to the published version when quoting the paper.
A persistent paradox in the process of globalization is the focus of academics and policymakers on sub-national regions as the essential unit of economic activity. Indeed, economic geography has for a long time cultivated a self-conception of being the science of the meso-scale (Grabher 2006), which arguably has led to an “overterritorialized” view on regional development (Hess 2004). In this light, most studies fail to conceptualize regional development in an era of globalization (Dicken and Malmberg 2001). Instead, it is arguable that a strategic coupling of global production networks and regional assets should be pursued, in which activities are mediated across different geographical and organizational scales (Coe et al. 2004; Dicken et al. 2001b). Rather than the persistent focus on regions as locally embedded entities, they should be seen as “new islands of an archipelago economy” (Hein 2000), in which a process of transnational network embedding exists, creating interpersonal relationships of trust at different, interrelated geographical scales (Henderson et al. 2002; Hess 2004). Similarly, Gereffi et al. (1994) defined global commodity chains as interorganizational networks of products that link enterprises and states to each other within the world economy. Earlier, within a city-related context, Friedmann and Wolff (1982) developed a conceptualization of world cities as “command centers” regulating the “new international division of labor.” Common to both approaches is the emphasis on the importance of multi-scalar networks to the understanding regional development (Derudder and Witlox 2010). These initial approaches have led to various studies on cities and globalization (e.g., Sassen 1991; Amin and Thrift 1992; Castells 1996; Cohen 1981; Meijer 1993; Abbott 1997; Godfrey and Zhou 1999), but the number of empirical world city network studies remains limited due to scarcity of “relational” data (Smith and Timberlake 1995; Taylor, Walker and Catalano 2002). To date, only a handful of relational studies exist, e.g., on international banks (Meyer 1986), advanced producer firms (Taylor 2004), MNC governance (Rozenblat and Pumain 2006; Alderson and Beckfield 2004; Wall 2009a and 2009b), and corporate directorates (Carroll 2007). These are important studies because few studies have attempted to understand the significance of firm ownership for economic development in particular societies (Henderson et al. 2002), especially because these studies do not privilege one particular geographic scale (Coe et al. 2004).
Nonetheless, even in these studies, conceptual differences are evident (Derudder 2006), as is clearly exemplified in the critical debate between Taylor (2006) and Alderson and Beckfield (2006). The first concerns functional differences in the data being analyzed. Alderson and Beckfield (2004), taking the lead from Hymer (1972), argue that the key relationship linking cities into a world system is the multinational enterprise, regardless of the industrial sector observed. Alternatively, the GaWC research group (2004), taking Sassen's lead (1991), focuses on the advanced producer service sector, which they justify as representing “cutting-edge” global economic activity. This, they argue, is because producer service firms have become multinationals in their own right, creating an essential “interlocking” (Knoke and Kuklinski 1982; Taylor 2001) global network of offices; however, according to Alderson and Beckfield (2006, p. 902), “this is an important empirical question that should not be foreclosed” because, although producer services may lead the way in integrating cities into a global network, it is likely that other industrial sectors also create important connections between cities. Based on this argument, they recommend that future research compare multinational and producer service networks based on a “primary” dataset – a recommendation that we follow in this article. The second important difference in this debate concerns the inclusion of hierarchical differences in the analyzed corporate data. As acknowledged by Alderson and Beckfield, a key advantage of Taylor's (GaWC) method is its sensitivity to the relative importance of firms in cities (p. 902-3). Alderson and Beckfield address this issue by pointing out that future research should more sensitively measure hierarchical differences that exist within corporate organizations – another recommendation that we have heeded. Based on these recommendations, we question what the corporate network centralities and structures of cities will be when considering different levels of corporate ownership in terms of (1) the network generated by all industrial sectors, (2) the network derived from producer services, and (3) the degree of overlap between these systems. In this way, we empirically explore the three major dimensions of corporate networks, i.e., firms, industrial systems, and territories (Dicken and Malmberg 2001). Furthermore, because the network characteristics of hierarchy and heterarchy are fundamental to corporate systems, (e.g., Hymer 1972; Lin et al. 2001; Hedlund 1986; Grabher 1993; Dunning and Lundan 2008), these properties receive special attention in our analysis.
This study borrows techniques from the realm of social network analysis (e.g., Granovetter 1985; Irwin and Hughes 1992), although admittedly this is generally not appreciated in mainstream economic geography, due to its austere and formalistic style (Grabher 2006). To examine the three research questions formulated above, we collected a dataset similar to that of Alderson and Beckfield, which in our case includes the Fortune top 100 global headquarters (2005) and their subsidiaries; however, in our data, different corporate sub-levels are included. Additionally, the firms have been coded according to their industrial (SIC) codes and city locations, resulting in a dataset of 9,243 corporate ties in 2,259 unique cities. Social network analysis and GIS techniques have been applied to explore the interdependencies between firms at global and regional scales. The data and techniques will be explained in detail in the methodological section, followed by the analysis, results, and conclusions. In the next part of this article, we will discuss several conceptual and methodological differences within the world city literature.
The Worlds of Hymer, Friedmann, and Sassen
Hymer (1972) predicted that the structure of the world system would mirror the hierarchical organization found in multinational enterprises, i.e., the hierarchical division of labor between geographical regions would correspond to the vertical division of labor within the firm. He expected that there would be a diffusion of industrialization to developing countries, whereby intermediate corporate activities would be concentrated in middle- and low-range cities, while high-level activities would be most concentrated in a limited number of hub cities close to capital, markets, media, and government activities, e.g., New York, London, Paris, and Tokyo. Similarly, Friedmann's world city hypothesis (1986) includes several generalizations relevant to our research: First, the extent to which a city is functionally integrated into the world economy is decisive for its level of development. Second, core cities are used by global capital as “basing points” in the spatial organization of markets and production. Third, the economic linkages emerging from this system allow for the arrangement of world cities into an intricate spatial hierarchy. Based on these generalizations, Friedmann (1995) argued that world cities can be hierarchically ranked according to the economic strength that they command. Hence, cities that rank highest can be seen as the control centers of the global economy, followed at a lower level by cities that control supraregional economies and, at even lower levels, cities that articulate national and subnational economies.
Sassen's studies discuss the contemporary transformation of the world economic system into a “complex duality” of “spatially dispersed, yet globally integrated organization of economic activity” (2001, p. 3). Since the 1970s, the global dispersion of production has proliferated as corporations have increasingly sought lower wages, proximity to markets and resources, and ways to redistribute labor (Sassen 1991). This has led to the dispersion of production activities to semi-peripheral and peripheral cities of the world, accompanied by the strengthening of centralized control in core or “global” cities. In particular, these centralized functions are strongly represented by the advanced producer services, which enable the control of production worldwide. Unlike Hymer, but similar to Friedmann, Sassen stressed that the globalization of services would give rise to a world with a new class of service cities controlling an array of production-oriented cities, in which a “vast territory” would be increasingly excluded from the vital economic processes of the global economy (Sassen 1994, p. 4).
These theories are similar to the extent that they place competitive market mechanisms and city development within the wider context of globalization (Cerny 1991), by which a hierarchical division of world cities into core and peripheral regions is defined. Nonetheless, although these theories have existed for quite some time, most contemporary analysis of economic development remains “bedeviled” by analytical disjunction, resulting in separate studies at the macro, meso, and micro levels (Henderson et al. 2002). Instead, studies must focus on the complex interactions between territorialized relational networks and global production networks (Coe et al. 2004). To date, only a handful of studies have addressed the interdependency of different geographic scales (e.g., Amin 2002; Bair and Gereffi 2001, Taylor 2004; and Alderson and Beckfield 2004); however, many conceptual differences are evident within these studies. In the next section, we discuss the differences between two leading studies.
Parallel Paths in Network Analysis
In the 2006 debate between Taylor and Alderson and Beckfield, the conceptual differences between their respective studies are discussed. These studies are functionally dissimilar in that the former supports Sassen's advanced producer service approach, while the latter support Hymer's emphasis on multinationals. Alderson and Beckfield defend the idea that the multinational has taken advantage of new communication technologies to create today's global economic system (Alderson and Beckfield 2004). Alternatively, Taylor argues that advanced producer services are essential to the overall structure of the world city network (Taylor 2006) because these firms form a “cutting-edge” service network that “interlocks” with the networks of other industrial sectors, thereby facilitating global production (p. 885, 892).
Broadly following Godfrey and Zhou (1999), Alderson and Beckfield (2004) collected data on the global Fortune 500 firms (2000) and their direct “sharehold” governance relations to subsidiaries worldwide. From this data, a city x city matrix of 3,692 cities worldwide was produced. Alternatively, the GaWC method was to collect data on 100 advanced producer firms and their offices across 315 cities worldwide (Taylor et al. 2009). This “service values matrix” is particularly interesting because it hierarchically qualifies different strengths of corporate connections between cities. Furthermore, the studies differ in the types of analysis employed to measure the centrality and structure of the world city network. Alderson and Beckfield, coming from a sociological background, use a social network analysis toolkit to obtain several measures of network centrality: outdegree, indegree, and betweenness. These terms are discussed in the next section on methodology. Alternatively, the GaWC method of measuring centrality is called total interlock connectivity and involves summing the products of every firm's service value in a given city with the same firm's service value in all other cities. Because their data are classified into different corporate categories, the measure can be disaggregated to consider three levels of corporate linkages. In our own analysis, we apply a hierarchical classification similar to that of Taylor, to data similar to that of Alderson and Beckfield's.
The above discussed conceptual differences led to variations in results. In the Alderson and Beckfield results (Table 1), we first see that Tokyo, New York, Paris, and London top the list in terms of both outdegree (outgoing linkages) and indegree (incoming linkages). The fact that Singapore and Hong Kong are strong in terms of incoming linkages supports Godfrey and Zhou's (1999) conclusion that these cities play a vital role in the world economy due to their strong presence of subsidiaries. Taylor's interlocking model reveals three overlapping levels of network connectivity (Table 1), i.e., global, dominant, and subordinate. Again, London and New York top the list, and interestingly, Hong Kong ranks above Tokyo. Taylor finds it strange that the Alderson and Beckfield study undervalues cities like Hong Kong and Singapore (p. 889) in terms of outdegree.
Table 1: Comparison between ranking of Alderson and Beckfield, and Taylor et al.
In these results, similarities and differences are evident. Obviously, this has to do with the different types and sizes of datasets used, as well as the varying methods and techniques employed. In our analysis, we follow the two central recommendations raised in this debate: (1) to analyze both the multinational and advanced producer service networks based on a primary dataset and (2) to use the existing hierarchical structure of corporate holdings, found in the Directory of Corporate Affiliations, as a means to more sensitively measure the presence of various kinds of firms in cities.
The Hierarchical and Heterarchical Organization of Firms
By emphasizing the critical role of the position of actors in “webs of affiliation”, Georg Simmel (1890), laid the foundations for social network analysis. Because economic action, rather than being socially embedded, is fundamentally social (Polanyi 1944; Grabher 2006), it is arguable that social network analysis is similarly applicable to understanding firms. In this context, the firm can be considered as a constellation of network relations governed by social actors (Yeung 2005), whereby, rather than being perceived as a mechanistic production function, the firm is perceived as a contested site for discursive and material constructions at various organizational and spatial scales.. According to De Filippis (2001), networks encompass hierarchies of power; otherwise, they would not be networks in the first place. Furthermore, it is argued that there would be no incentive for stronger actors to remain in a network if they did not disproportionately gain from the benefits of network participation (Lin et al 2001; Christopherson and Clark 2007). So, considering that hierarchy exists within corporate networks, then it is questionable to what degree this is, and whether other structures of organization can also be found? Hymer (1972) saw the world system as mirroring the organizational structure of the multinationals of his day and assumed that this static, top-down hierarchy would simply be perpetuated at the start of the 21 st century; however, we do not appear to have moved far beyond Hymer's simplistic projection of the hierarchical-organizational structure of the large multinational onto geographic space (Dicken and Malmberg 2001), when in fact a firm's organizational architecture is far more complex than this (Dicken and Malmberg 2001). In fact, the late 20 th century witnessed the emergence of new organizational forms, such as strategic alliances and inter-firm networks that are essentially different from the hierarchical control of the firm's activities through vertical integration (Yeung 2005). Only in the more recent studies of, e.g., the GaWC research group do we see acknowledgement of more complex relationships through their incorporation of a five-level “service-value” analysis of differing firm strengths, which allows for an initial understanding of the more subtle differences in vertical and horizontal relationships between headquarters and regional units.
In a recent article, Taylor, Hoyler, and Verbruggen (2008) postulate the need for a new theory concerning the horizontal spatial structure linking non-local interactions. This theory would complement Christaller's (1933) “central place theory,” which is known for its conception of vertical spatial structures that link local scales of interaction. These authors call this “central flow theory,” which relates to the fact that the vertical structure of previous multinationals has changed considerably (Chandler 1962, 1990) with respect to the nature of the internal relationships between headquarters and subsidiaries and between subsidiaries themselves (Birkinshaw 1995, 1996; Ivarsson 1996; Taggart 1997). Hence, intra-corporate competition between various units of the firm serves as a vital mechanism by which to redefine spatial divisions of labor and time-space configurations (Schoenberger 1997, Phelps and Fuller 2000), encapsulating the complex interrelations between “flow economies” and “territorial economies” (Yeung 1998, p. 229). Thus, the dominant vertical form of corporate organization is disintegrating (Taylor and Asheim 2001) as new economic innovations lead to intricate networks in which vertical and horizontal connections intertwine (Koestler 1978, Knaap v.d 2007, Wall 2009a), creating a “heterarchical” system (Hedlund 1986) in which firms become enmeshed in loosely coupled networks of interdependence, reciprocity, and unequal power relations (Grabher 1993, 2006; Taylor 1995, 1996).
A heterarchy has three aspects that distinguish it from the hierarchical model of corporate organization. First, resources and managerial capabilities are dispersed throughout the organization, instead of being located only at the top. Second, lateral relationships exist between subsidiaries in terms of products, people, and knowledge flows. Third, activities are coordinated along multiple dimensions, typically geography; products; and functions (Hedlund 1993, 1994). Hence, the multinational has evolved from a comparatively simple set of unidimensionally and vertically controlled processes into a complex system of vertical and lateral intra- and inter-firm relationships (Maskell 2001; Dunning and Lundan 2008). Nonetheless, although it may be true that an organizational transformation towards a more integrated network is taking place (Bartlett and Ghoshal 1989), we should be careful not to neglect the primacy of the vertical, hierarchical dimension (Hedlund 1986; Koestler 1978). From this, it is arguable that distinct or essentialist categories of network structure do not exist, but instead overlap and interpenetrate each other to varying degrees. Networks form complex combinations of overlapping, juxtaposed and nested governance mechanisms (Grabher and Powell 2004). So, if hierarchy and heterarchy coexist, then what are their characteristic features? As indicated in the conceptual diagram (Figure 1), the topology of a hierarchical, vertically organized network would resemble the star-shaped structure on the left (Hannemann and Riddle 2005). Alternatively, a heterarchical, horizontally organized structure would resemble the “universal” network structure on the right, in which all actors are mutually connected (Todeva 2006).
Figure 1: Simplified network diagrams to explain hierarchy and heterarchy.
As is the case in the studies of Taylor and Alderson and Beckfield, we pursue an intercity network produced by firms' location decisions. Thus, the firms are the origin of the network, not the cities themselves (Pred 1977). The data compiled for this study are similar to those of Godfrey and Zhou (1999) and Alderson and Beckfield (2004) as they concern the Fortune Global 100 multinationals (2005). These 100 headquarters accounted for 27% of OECD revenue in 2005, indicating the economic importance of these firms. Next, the subsidiaries of these headquarters and their different levels of share ownership were extracted from the LexisNexis Directory of Corporate Affiliations (2005). This database covers more than 180 000 of the most important headquarters in the world and their respective subsidiaries (see www.lexisnexis.com/dca), organized at different levels of corporate ownership, but also including information on their industrial sectors and city locations. However, because about 20% of the data is incomplete, the missing information had to be obtained using various internet search engines. First, it can be seen that there are five levels of ownership (Figure 2a). At the first level, headquarters A owns shares in primary subsidiaries B, C, D, and E. At the second level, subsidiaries C and E in turn govern sub-subsidiaries F – J. Firm I, at the third level, owns shares in firms K and L, and so forth.
After organizing the firms into the five governance levels, the standard industrial classification (SIC) code for each firm was identified. In this way, we could analyze the entire production network versus the producer service network based on a single dataset. Next, the city location of each firm was identified, resulting in an adjacency matrix containing 9,243 corporate holdings in 2,245 unique cities. In the diagram (Figure 2a), subsidiary levels 3, 4, and 5 were condensed into one level because these levels hold low counts of shares (roughly 5% each of the total), and the individual ties are not strong. The next diagram (Figure 2b) shows how the corporate organization in Figure 2a translates into a city x city network of corporate relations. Firms in Figure 2a are located in any of cities 1 - 5. For instance, firms B, F, I, and N are situated in city 2. These same relations are shown in the included matrix and intercity network diagram (Figure 2a). Based on this, the four centrality measures of outdegree, indegree, diagonal, and betweenness can be explained. In the matrix, moving from left to right, we see outdegree relations (corporate holdings) stemming from firms in each of the five cities to other firms in these same cities. This is simply a measure of the amount of ownership that firms in a particular city have of firms in other cities. This can be interpreted as the “power” of certain cities over others (Alderson and Beckfield 2004). In this case, city 3 has the most influence over other cities, and it ranks first in terms of outdegree centrality. Furthermore, its strongest holdings are in cities 1 and 2. Inversely, by moving from top to bottom, we can determine the degree to which firms in particular cities are owned by firms in other cities. This is a measure of the prestige of a city (Alderson and Beckfield 2004). In this sense, a city is prestigious because governing firms in other cities are dependent on a certain number of its firms. In this case, we see that cities depend most on city 2, making it rank first in terms of indegree centrality. In network analysis, this type of data is called “directional” because we can measure the direction of corporate ownership between cities. In Figure 2b, the arrow directions clearly show that hierarchical “vertical” relations can exist (e.g., from city 3 to city 2) in addition to reciprocated “horizontal” relations.
Figure 2a: Levels of corporate ownership
Figure 2b: Corporate ownership between city nodes
Furthermore, a city can possess more than one tie to another city. For instance, city 3 has two ties to city 1. In this sense, the analysis allows for “valued” ties. The third measure is called the diagonal and is simply a measure of corporate “self-ties” within a city. In the matrix, only city 3 has a self-tie (between firms C and G), which can be identified along the diagonal running from the top corner of the matrix to the bottom corner (hence the name “diagonal”). Based on these techniques, the extent to which a city is dependent on intercity and intra-city ties can be derived; however, betweenness is unlike the other centrality measures because it measures the strategic position of a city in relation to other cities. To do this correctly, the matrix is “dichotomized” into unvalued relations of either 0 or 1. Of interest here is simply whether cities are linked or not, not the strength of the relationship. In Figure 2b, cities 1 and 2 each have four unvalued ties to other cities and therefore rank highest in terms of betweenness, followed in descending order by cities 3 (including its self-tie), 4, and 5. As discussed by Alderson and Beckfield (2006), “betweenness” measures the brokerage or strategic intermediacy of a city within a system of cities. Due to space constraints, we have excluded the mathematical formulations of these measures. Those interested should see Alderson and Beckfield (2004), p. 822-825; Irwin and Hughes (1992); and Hannemann and Riddle (2005). To determine how much the centrality of multinational networks overlaps that of producer services, the different scores were correlated.
The defined centrality measures are “nodal” measures of the network concerning the aggregations of ties; however, the structure of a network concerns the tie strengths between nodes. For instance, in Figure 2b, the tie between cities 1 and 3 is strongest; however, the structure also includes the overall configuration of the network into core and periphery zones. In the diagram, city 4, with the weakest combined outdegree and indegree, can be seen as the most peripheral city of this network. Although it is weakest in terms of the combined five levels of corporate ownership, it is equally as important as cities 2, 3, and 5 when measured at only the first level of ownership (between headquarters and first-order subsidiaries). To determine network structures, the network analysis software Ucinet/Netdraw (Borgatti, Everett and Freeman 2002) was used. The software reveals the essential nodes, linkages and core, and periphery zones of the data. Because this analysis disregards geographic location, the geographic information systems (GIS) software MapInfo was used to map the spatial distribution of the network. Lastly, to measure how well centrality and structure of the entire multinational network corresponds to that of the producer services, a correlation was carried out between the nodal and linkage strengths of these two networks.
Centrality within the Network of all Industrial Sectors
The results in Tables 2a – 2d refer to the first research question. The top table concerns the combined five levels of corporate holdings (Figure 2a). The grey part includes the purely intercity values of outdegree and indegree which represent only ties “between” cities, and also the intra-city diagonal values (self-ties) which are inter-corporate ties that occur “within” the city. However, because it is arguable that self-ties should be included in outdegree and indegree scores, so as to get a total measure of city connectivity - we have accordingly provided a separate white table with these scores.
Table 2a: showing differing centrality scores at each level of corporate governance.
Table 2b: showing differing centrality scores at each level of corporate governance.
Table 2c: showing differing centrality scores at each level of corporate governance.
Table 2d: showing differing centrality scores at each level of corporate governance.
In the “outdegree” column, we see that New York has the largest number of shares in firms located in other cities, followed by Paris, Tokyo, and London. However, as will be shown, Paris's strength lies at lower levels of the corporate hierarchy. In Taylor's “global network connectivity” results (Table 1), Tokyo ranks similarly to our results. Of the 2,259 unique cities in our dataset, only 17% have outdegree scores; and all are from developed countries, underlining Hymer's point (1972) that corporate activity is concentrated in a limited number of cities close to capital markets. In this context, New York, Paris, London, and Tokyo together hold 25% of corporate ownership of firms in other cities. In the lower group of cities, we see a strong match with Alderson and Beckfield's study, i.e., Zurich, Dusseldorf, Munich, and Amsterdam. Furthermore, the presence of specialized cities like Palo Alto and New Brunswick confirms Alderson and Beckfield's (2004) observation that headquarter functions are not necessarily concentrated in world cities and are instead often located in the city of the firm's origin.
Our top two cities in terms of indegree exactly match Alderson and Beckfield's list, but we identify Singapore and Hong Kong as the second and third most prestigious cities. This is similar to Godfrey and Zhou's (1999) and Taylor's results (Table 1) showing that Singapore and Hong Kong, but also capital cities in developing economies like Mexico City, Buenos Aires, Jakarta, and Bangkok (see Table 2a), hold important subsidiary functions. The indegree list holds a richer variety of cities from developed as well as developing nations, supporting Hymer's prediction of a strong diffusion of industrialization to developing countries. It is also interesting that cities once conceptualized as being “peripheral,” like Bangkok, Mexico City, and Sao Paulo, as discussed by Sassen (1994), have indeed emerged as highly ranked world cites. Furthermore, in terms of diagonal centrality, only 16% of interactions proved to be intra-urban (self-ties), supporting the view that important cities derive their status from what flows between them rather than what remains fixed within them (Amin and Graham 1999; Allen 1999; Castells 1996). Thus, firm networks are indeed capable of exercising control and power over the firm “at a distance” (Yeung 2005 p. 316).
Tables 2b, 2c, and 2d are sub-networks of Table 2a. Table 2b concerns headquarters to first-order subsidiaries (first level). Table 2c represents first subsidiary to second subsidiary holdings (second level), and Table 2d represents the remaining linkages (third, fourth, and fifth levels). The first level accounts for 44% of all connections, followed by the second level with 30% and the remaining levels with 26%. From this, it is clear that corporate governance is strongest towards the top of the system, possibly because higher-end functions need more control at this level than the more production-oriented functions downstream. At the first level (Table 2b), the strongest outdegree cities are New York, Zurich, and Munich. Furthermore, London and Paris play weaker roles at the first level than in aggregate (Table 2a). Tokyo's outdegree (see Table 2a) drops significantly from the 3 rd to the 24 th position within these higher economic functions; however, as shown in Table 2c and 2d, Tokyo's outdegree rises to the top position at more production-oriented levels.
It is also clear that unlike in Taylor's results, Hong Kong and Singapore do not play strong roles in terms of outdegree at any corporate level, as shown in Alderson and Beckfield's results. This difference possibly stems from the fact that Taylor is measuring producer service data; however, even in specific producer service data (shown later), Singapore and Hong Kong do not play significant roles in terms of outdegree centrality. In terms of indegree scores, at the first level (Table 2b), there are only minor divergences from the top cities in Table 2a. The more interesting point is that both Singapore and Hong Kong drop significantly in the rankings. This indicates that their strengths as subsidiary cities are weaker at this level of higher-end functions.
As seen in Table 2c, the positions of Hong Kong and Singapore rise significantly within lower subsidiary levels. In terms of self-ties, New York, London, Paris, and Tokyo dominate the upper echelons of the list. These global cities, besides being strongly linked to other cities, are equally strong in terms of their intra-city ties. This is confirmed later in the correlational analysis, which shows a strong coherence between a city's global and local economic functions. Lastly, observing betweenness (Table 4 - left), we see that New York, Paris, Tokyo, Dusseldorf, and London prove most strategic for corporate interactions taking place amongst the majority of the world's cities. In this context, these cities are the prime mediators or brokers of global corporate ownership relations.
Table 4: Betweeness centrality. Source: Wall and v.d. Knaap, 2009.
Structure within the Network of all Industrial Sectors
In the network of all levels combined (Figure 3a), two patterns emerge in terms of network topology and geography. Topologically, a hybrid structure between hierarchical and heterarchical interdependence is observed. The star-shaped structures, in which central cities (e.g., New York, Paris, and London) exercise corporate governance over others clearly exemplify Hymer's (1972) hierarchical, vertically organized interaction. The triangulated structures connecting cities (e.g., the London, Hong Kong, and Singapore triad) represent heterarchical interactions between cities (Hedlund 1986). Furthermore, Singapore and Hong Kong are not only part of London's network, but also share positions in other networks. For instance, Singapore forms part of Tokyo's and Zurich's system. It is also interesting that New York and Tokyo are essentially connected to cities within their proximate regions.
Figure 3a: All industries - all headquarter to subsidiaries (levels 1 - 5).
Figure 3b: All industries - headquarters to first subsidiaries (level 1).
Figure 3c: All industries - first subsidiaries to second subsidiaries (level 2).
Figure 3d: All industries - second subsidiaries to remaining subsidiaries (level 3 - 5).
In the sub-network (Figure 3b) that represents corporate level 1 (Figure 2a), we see the primary linkage between New York and London, confirming Taylor's statement that this city pair is the “global cities dyad par excellence” (Taylor et al. 2009). Furthermore, unlike New York, London is essentially connected to transnational cities. Notably, Paris plays a moderate role at this level of interaction. Furthermore, as evident in the outdegree and indegree scores, Singapore and Hong Kong do not play important roles at the top of the corporate system. Zurich, Munich, Dusseldorf, and Milan form important European centers, and Tokyo essentially connects to Taipei and Bangkok.
At level 2 (Figure 3b), Paris, Tokyo, London, and New York prove to be important central cities, revealing both hierarchical “hub-and-spoke” structures as well as heterarchical, triangulated structures. Whereas Paris was insignificant at level 1, it is the primary city at level 2, with its strongest links to Brussels, New York, and London. Interestingly, Hong Kong and Singapore play important roles at this level. At level 3, the last network of this series (Figure 3d), the importance of Tokyo, New York, London, and Paris are clear, but heterarchical characteristics are weak. Tokyo proves to be strongly connected to London and an array of proximate Japanese and Asian cities. It is clear at this level that Tokyo performs a strong national and regional function within Southeast Asia. In this light, the role of the Japanese integrated trading companies called “Sogo Shosha 1” (Young 1979, Edington and Haga 1998) is less transnational and instead more nationally oriented (Hill and Kim 2001). This raises the question of whether or not Tokyo can be considered a world city (see also Friedmann 2001; Sassen 2001; Hill and Kim 2000).
Figure 4: Geographic information system map of the all industries multinational network.
The corporate network between cities was mapped using GIS (Figure 4). The white dots depict the presence of firms within cities and are scaled according to the sum of each firm's indegree and outdegree. The black lines illustrate the aggregate corporate shares between cities of at least five links (holdings), while the grey lines show ties of fewer than five links. The supra-regional East/West triad (Friedmann 1986; Carroll 2007) between North America, Europe, and Pacific Asia is clearly evident. For instance, 70% of Europe's connections are with the rest of the world, and only 30% take place within the European region. In the case of North America, 65% of its connections are global. These results verify that there is indeed a strong coupling of regions to global production networks (Coe et al. 2004). In addition, it is found that approximately 30% of economic ties are localized because they depend on interdependencies of geographical agglomeration achieved through territorial embeddedness (e.g.; Storper and Salais 1997; Whitley 1999, Feldman 2000; Hall and Soskice 2001). Conversely, the fact that roughly 70% of ties operate “at a distance”, verifies the conceptions that warn of a too strong local and regional embeddedness (see Coe et al. 2004; Dicken et al. 2001b; Henderson et al. 2002, Bathelt et al. 2002). Furthermore, the southern hemisphere linkages are mainly to Commonwealth countries (see Alderson and Beckfield 2004, p. 835) and South America. Africa is primarily bound through Johannesburg, Abidjan, Lagos, and Cairo, but the relative share of connectivity to this continent is very sparse (1% of the total). The grey lines indicate frequent but weak inter-regional activity. Also, the prominence of Trans-Atlantic interaction is clear.
Table 5: Outdegree and indegree shares of four major cities to various nations.
Although the world map gives a clear view of the entire corporate system, it is not clear how many of a city's ties are intra-national versus transnational. Therefore, we more closely examine New York, Paris, London, and Tokyo (Table 5). The top two tables concern the “all industrial sectors” outdegree and indegree scores. New York is essentially connected to U.S. cities (59% outdegree and 74% indegree to U.S. cities), and its transnational ties are sparse. Although New York's world city-ness is questionable, the immense scale of the U.S. should be taken into account, where the size of its regional network roughly equals that of all European countries combined (see Figure 4). The second most intra-national city is Tokyo (47% outdegree and 63% indegree to Japan), emphasizing the nationally oriented character of Japanese Shogo Shoshas (Hill and Kim 2001). Paris is the third most nationally oriented city and has the highest number of unique city connections (87%), making it the most integrated city of the four. London is the most “global” of the cities because it is primarily connected to Canada (17%) and other countries.
Table 3a: Differing centrality scores at each level of corporate governance.
Table 3b: Differing centrality scores at each level of corporate governance.
Table 3c: Differing centrality scores at each level of corporate governance.
Table 3d: Differing centrality scores at each level of corporate governance.
Centrality within the Network of the Advanced Producer Service Sector
Table 3a concerns the combined five levels of producer service corporate holdings. In the outdegree column, New York, Paris, and London head the list in terms of outdegree, followed by a secondary set of cites: Zurich, Frankfurt, The Hague, and Amsterdam. Tokyo plays a more marginal role within producer services than in total industries (Table 2a). Hence, Tokyo is arguably a production-oriented city. The absence of Hong Kong and the insignificance of Singapore (30th) show that these cities do not have a strong presence of headquarters. Nonetheless, in terms of indegree (subsidiaries), these cities rank third and fourth, respectively, and are governed specifically by headquarters in London (Figure 5a). The highest ranked cities in terms of outdegree are located in developed countries, whereas the highest ranked cities in terms of indegree are located in both developing and developed nations. Concerning the diagonal, 18% of the service links are self-ties. Hence, in this “cutting-edge” network (Taylor 2006), intercity connectivity predominates (Amin and Graham 1999).
We now investigate sub-levels of corporate holdings (Tables 3b – 3d). Level 1 (Table 3b) claims 45% of all service connectivity, indicating that the majority of corporate governance exists at this top level. At lower levels, the connectivities weaken: Level 2 (Table 3c) claims 34% of all service connectivity, and levels 3 – 5 (Table 3d) claim 21%. In Table 3b, it can be seen that New York, London, The Hague, and Paris rank highest. Note that in terms of outdegree, New York is three times stronger than London and the other cities. This differs from Taylor's findings (Table 1), wherein London ranks above New York in terms of “global network connectivity.” Also striking is Tokyo's lack of prominence in terms of outdegree at this level. This differs from Taylor's findings, in which Tokyo is ranked fourth in connectivity. Furthermore, unlike in Taylor's findings, Hong Kong and Singapore are not prominent in our results. In terms of indegree scores (Table 3b), New York, London, Paris, and Toronto are leading cities. Together with Singapore and Mexico City, Tokyo and Hong Kong are significant in terms of producer service subsidiaries.
Figure 5a: Advanced producer services - all headquarter to subsidiaries (levels 1 - 5). Source: Wall and v.d. Knaap, 2009.
Figure 5b: Advanced producer services - headquarters to first subsidiaries (level 1). Source: Wall and v.d. Knaap, 2009.
Figure 5c: Advanced producer services - first subsidiaries to second subsidiaries (level 2). Source: Wall and v.d. Knaap, 2009.
Figure 5d: Advanced producer services - secondt subsidiaries to remaining subsidiaries (level 3 - 5).
At a lower level (Table 3c), Tokyo ranks high. Interestingly, Paris is the primary city and is at least 1.5 times stronger than London at this level. Hong Kong rises to first place in terms of indegree scores, followed by London, Singapore, and New York. Furthermore, Paris and Tokyo are less prestigious at this level. Interestingly, a sub-Saharan African city (Johannesburg) plays a moderate role. At the lowest level (Table 3d), the usual cities continue to top the outdegree list. On the indegree side, cities like Jakarta, Bangkok, and Shanghai play important roles, holding sub-regional subsidiary offices in this sector. Betweenness centrality (Table 4 - right) reveals that New York, Paris, London, Houston, and Toronto are the most strategic “brokers” within the total producer service system. Furthermore, compared to the “all industries” results, Singapore holds an important strategic position.
Structure within the Network of the Advanced Producer Service Sector
We can now discuss the structural characteristics of the advanced producer service sector (Figures 5a – 5d). In Figure 5a (combined levels), we again see hierarchical and heterarchical structures, evident in the star-shaped and triangulated structures of New York, Paris, London, and Zurich. Hong Kong and Singapore play central roles within this network. Although New York is strongly connected to London and Paris, Paris and London are weakly connected. At the first level of economic activities (Figure 5b), New York claims the strongest position. The strongest dyad is between New York and London and is strongly connected to Frankfurt, Paris, and Boston. Heterarchical structures dissipate as we drop to lower corporate levels (Figure 5c). The hierarchical star-shaped arrays of Paris, New York, and London are evident. At this level, the only exceptional link is between London and Hong Kong. In Figure 5d, all heterarchy has dissolved, and only fragmented hierarchies prevail. The only noteworthy link is that between London and Singapore. Hence, across all three levels, London plays an essential role, making it the most important city within the producer services system, as in Taylor's results (Table 1).
Figure 6: Geographic information system map of the producer service industries network.
In the GIS map (Figure 6), we again see the triad of North America-Europe-Pacific Asia. Indeed, although an “archipelago economy” appears to exist (Hein 2000), it certainly is not uniform, but instead is comprised of three major archipelagos - connected almost entirely to each other. This is evident in the strong trans-Atlantic connectivity (Carroll 2007), and the weak participation of cities and regions in the Southern Hemisphere, illustrating Sassen's statement that a “vast territory” is increasingly excluded from the “major economic processes that fuel economic growth in the new global economy” (Sassen 1994). This exclusion is related to the fact that these regions have very different state-labor configurations, which arguably influence how these regions are articulated into global production networks (Coe et al. 2004). For regions to develop, they will need to readjust these mechanisms to facilitate coupling to global production processes. It is also clear that Singapore and Hong Kong play important roles in Pacific Asia (Godfrey and Zhou 1999). From this map, it is clear that Tokyo is not strongly connected to western cities, but instead to Japanese and other Asian cities, as observed by Hill and Kim (2001). In this sense, Hong Kong and Singapore are more akin to world cites than Tokyo is. Similar to the previous results on intra-national and transnational relations (Table 5), we see that New York is primarily connected to producer services within the U.S. (65%). Again, Tokyo is the second most intra-nationally connected city, followed by London and Paris; however, Paris is primarily connected to other countries, making it the most transnational of the four in terms of outdegree ties. Similar evidence can be found in the indegree results.
Table 6: Multiple correlations between outdegree, indegree, diagonal and betweenness (both networks). Source: Wall and v.d. Knaap, 2009.
Table 7: Correlation between linkage strengths (both networks). Source: Wall and v.d. Knaap, 2009.
The Degree of Overlap between the All Sectors and Producer Service Networks
First, in Table 6, a strong correlation coefficient (0.84) is found between the outdegree centrality scores of the two systems. This means that cities harboring all sector headquarters will also have high counts of producer service headquarters. In terms of indegree (subsidiaries), the coefficient is modest (0.66) due to the higher variance of indegree cities within the networks. A high correlation is also found between the diagonal centralities of the two systems (0.78). This indicates a strong coherence between producer firms and the services on which they depend. The coefficient of betweenness (0.69) indicates that cities that are strategically positioned within the all-sector system are equally likely to be strategically placed within the service sector system. In linkage correlation (Table 7), the coefficients are separated into two parts. The top results, “matching structure,” concern a correlation where “all sector” dyads (city pairs) are used only if they are also found within the “producer service” network. This analysis includes 2,196 unique dyads. The bottom results concern a correlation across the entire network, involving 5,863 unique dyads. In the “matching structure” results, we see a coefficient of 0.73. This means that the strengths of the dyadic linkages in the two networks are highly coherent; however, when the rest of the network, i.e., the part excluding producer services, is included in the calculation, the coefficient is weak (0.33). On one hand, these results confirm Taylor's argument that producer services “interlock” strongly with the economic system. On the other hand, it is evident that this “avant-garde” interlocking (Taylor 2006, p. 892) only holds for linkages at the top of the system. The low coefficient in the last result indicates that a system of production linkages exists at lower levels that have no correspondence with the producer service network.
Our study has contributed to understanding the functional and organizational interdependencies between firms at local, regional, and global territorial scales (Yeung 2005; Coe et al. 2004; Dicken and Malmberg 2001). By showing that 84% of the multinational network occurs between cities and not within them, and that approximately 70% of European and North American ties extend beyond their respective supra-regions, we empirically support the claim that cities have become dissociated from their local geographies as their positions in worldwide networks have grown (Friedmann 1986, Knox and Taylor 1995, Sassen 2001). We show that indeed it is no longer meaningful to speak of regional and local economies without considering them within a new type of globally distributed system (Storper 2009). Furthermore, the difference in corporate ownership directionality (outdegree and indegree) substantiates Friedmann's (1995) postulation that world cities can be organized into a hierarchy according to the economic relations that they command. For example, although Hong Kong and Singapore are shown to be “important” cities (strong in terms of their number of subsidiaries), they are not strong in terms of headquarters.
Furthermore, because only 17% of cities hold outdegree centrality and New York, London, Tokyo, and Paris combined claim 25% of all outgoing connections, the premise that high-level economic activities are concentrated in a limited number of cities (e.g., Hymer 1972, Sassen 1991) is empirically confirmed. In addition, the GIS maps show that a “complex duality” of “spatially dispersed, yet globally integrated organization of economic activity” exists in the modern world (Sassen 2001, p. 3) and is clearly evident in the dense economic agglomerations within and between North America, Europe, and Pacific Asia. Furthermore, it is evident that a “vast territory” exists that is excluded from the vital economic processes of the global economy (Sassen 1994). For example, although 10% of the world's population resides in sub-Saharan Africa, this region claims only 1% of multinational corporate connectivity. In this context, the unevenly distributed economic system of the world becomes clearly observable (Harvey 2006).
Our study has also contributed specifically to world city network research, especially concerning the different approaches of Taylor and of Alderson and Beckfield, by exploring their approaches based on a primary dataset and by integrating several of their concepts and methods. This was done, first, by consistently observing the “all industrial sectors” and “advanced producer sector” networks, and second, by including three separate levels of corporate ownership. The former technique proved useful to test for functional differences between previous studies and ours, while the latter has been essential for understanding variations within intercity corporate organizations. It has been shown here that the two sectorally different networks correlate strongly, meaning that, indeed, producer services “interlock” strongly with the overall economic system (Taylor 2006, p. 892). Nonetheless, as shown here, this interlocking is only strong at the apex of the system. The weak correlation across the entire system supports Alderson and Beckfield's assertion that although the producer-services sector could lead the way in binding cities into a global network, it is likely that the industrial sector also constructs other forms of connections amongst cities (Alderson and Beckfield 2006).
One of the major similarities of the two types of networks is that New York, London, Tokyo, and Paris generally top the lists at all three corporate levels of analysis. This signifies an important trait of world cities: their ability to articulate between global, regional, and local networks, irrespective of the corporate level observed. This is different from Hymer's and Friedmann's strictly vertical definitions, where primary cities articulate at a global scale, secondary cities at the supra-regional scale, and tertiary cities at the periphery. This means that, today, linkages between cities are not passed down in a clear-cut “tree-like” hierarchy, but instead reciprocated, horizontal interactions do exist between cities at all levels; however, as shown in the separate network diagrams, the interaction between cities is neither purely hierarchical nor purely heterarchical, but a hybrid between the two exists, confirming the coexistence of these different organizational principles (Hedlund 1986, Koestler 1978, Grabher and Powell 2004, 2006).
The fact that New York, Paris, London, and Tokyo have hierarchical “hub-and-spoke” structures in most diagrams reveals the headquarters control functions of these cities over subsidiary cities. Indeed, networks do reveal the tendency to create hubs and “aristocrats” (Watts 1999). Furthermore, although lateral relations are found between different categories of cities, it is evident from our results that the strongest evidence of “horizontalization” is found between the top-ranked cities and that this is most prolific at the top level of corporate ownership, gradually declining in importance as we drop down to lower levels. Hence, heterarchy is most evident at the top level of corporate governance. This is not surprising as this level concerns high-end, complex activities between headquarters and first-order subsidiaries. Because the majority of individual ties are neither reciprocated nor triangulated, we can say that in terms of the “variety of connections”, the system is still essentially hierarchic. But these diverse ties are generally weak! Alternatively, although the number of reciprocated and triangulated ties is limited to only a handful of cities, these ties are disproportionately stronger. Hence, in terms of “strength of connections”, the system is essentially heterarchic.
It was also shown that cities with high outdegree are all located in developed countries. In contrast, in terms of subsidiaries, the highest ranked cities are from both developed and developing nations, e.g., New York, Paris, Mexico City, Buenos Aires, Singapore, Jakarta, and Bangkok. This clearly shows how shifts in competitive advantage in the global marketplace have increased the global reach of economic activity, driven by competitive market mechanisms, technological change, and space-time compression (Cerny 1991, Sassen 1991, Castells 1996).
Another important aspect of Alderson and Beckfield's study and ours is that specialized cities, like Vevey, Palo Alto, Irving, Ludwigshafen, or New Brunswick, also appear among the top ranks. It is therefore apparent that important functions are not necessarily concentrated in world cities, but instead are often located in the city where the firms originated (Alderson and Beckfield 2004). In this respect, it is arguable that both approaches gives a more complete view of intercity corporate relations because cities are not predefined, but instead include all cities that exhibit headquarters or subsidiary ties. This explains why our study includes 2,259 unique cities. Comparing our top ten results to those of Alderson and Beckfield, an 80% match can be observed; however, the rankings themselves differ slightly. Both studies identify New York, Paris, Tokyo, and London as the leading outdegree cities. The difference is that we identify New York as the most connected city in terms of both outdegree and indegree, while their study recognizes Tokyo as the most connected in outdegree and the fourth most connected in indegree.
In our three-level study of the network comprising all industrial sectors, we found that Tokyo's outdegree at the top level is very weak; however, Tokyo's strength in outdegree lies at the second and third levels of our analysis, and, as evident in the structural studies, Tokyo is primarily connected to Japanese and Pacific Asian cities. Therefore, our results support the argument that Tokyo might not be a global city (Hill and Kim 2001). Furthermore, because Tokyo plays a more marginal role in the producer services network than in the all industrial sectors network, this city is more production- than service-oriented.
New York's importance as a global city is also questioned because it has been demonstrated in both analyses to be connected primarily to U.S. cities; however, in the producer service study (first level), it was shown that New York, in terms of total outdegree, is roughly three times stronger than London. Although London is not the strongest city in terms of overall outdegree and indegree, it is the most connected to cities in other countries. In this context, London is the most global city in the world. Furthermore, it forms the strongest dyadic link within the system with New York, in both the producer service network and the all sector industries network. Also, London was shown to be the prime connector to Singapore and Hong Kong. Similar to Taylor's, Alderson and Beckfield's (2004), and Godfrey and Zhou's findings (1999), Hong Kong and Singapore emerged as important subsidiary type cities. This is evident in both the all sectors network and the producer services network. Their strength is that they house subsidiaries that are owned to a large degree by headquarters in other cities. They are therefore important cities that are “sought out by other cities” (Alderson and Beckfield 2004, 824); however, we discovered that Hong Kong and Singapore are far less important at the first level of the corporate hierarchy than at the lower two levels.
Lastly, we recommend that future research focus on (1) how to measure the spatio-temporal interdependency between global city network formation and local urban development, and (2) how to more accurately weigh different headquarters, subsidiaries, affiliates, and branch plants to get more sophisticated measures of the centrality, structure, and hierarchy between cities.
Abbott, Carl. 1997. The international city hypothesis: an approach to the recent history of U.S. cities. Journal of Urban History 24:28–52.
Alderson, A. S.; and Beckfield, J. 2004. Power and position in the world city system. American Journal of Sociology 109:811-851.
Alderson, A. S.; and Beckfield, J. 2006. Whither the parallel paths? The future of scholarship on the world city system. American Journal of Sociology 112:895-904.
Allen, J. 1999. Cities of power and influence: settled formations. In Unsettling Cities. Ed. J. Allen, D. Massey and M. Pryke. 1999 pp. 181-281. New York: Routledge.
Amin, A. 2002 Spatialities of globalization. Environment and Planning A 34 385–99
Amin, A., and Graham, S. 1999. Cities of connection and disconnection. In Unsettling Cities. Ed. J. Allen, D. Massey and M. Pryke. 1999 pp. 738. New York: Routledge.
Amin, A.; and Thrift, N. 1992. Neo-Marshallian nodes in global networks. International Journal of Urban & Regional Research 16:571–87.
Bair, J. and Gereffi, G. 2001 Local clusters in global chains: the causes and consequences of export dynamism in Torreon's blue jeans industry. World Development 29 1885–1903
Bartlett, C.A;. and Ghoshal, S. 1986. Tap your subsidiaries for global reach. Harvard business review, November-December: 87-94.
Bathelt, H., Malmberg, A. and Maskell, P. 2002. Clusters and knowledge: local buzz, global pipelines and the process of knowledge creation. Danish Research Unit for Industrial Dynamics (DRUID), Working paper 2002 – 12, Copenhagen.
Birkinshaw, J. M. 1996. How multinational subsidiary mandates are gained and lost. Journal of International Business Studies 27:467–96.
Birkinshaw, J. M.; and Morrison, A. J. 1995. Configurations of strategy and structure in subsidiaries of multinational corporations. Journal of International Business Studies. 26:729 753
Borgatti, S.P., Everett M.G., and Freeman L.C. 2002. UCINET 6 for Windows: social network analysis software. Analytic Technologies. Needham, MA.
Carroll, W.K. 2007. Global Cities in the Global Corporate Network. Environment and Planning A 39:2297 – 2323.
Castells, M. 1996. The Rise of the Network Society. Blackwell Publishers, Oxford.
Cerny, P.G. 1991. The limits of deregulation: transnational interpenetrations and policy change. European Journal of Political Research 19: 173 –196.
Chandler, A.D. 1962, Strategy and structure: chapters in the history of American industrial enterprise. Camridge, MA: MIT Press.
Chandler, A.D. 1990. Scale and scope: the dynamics of industrial capitalism. Cambridge, MA: Harvard/Belknap.
Christaller, W. 1933. Die Zentralen Orte in Suddeutschland. Jena: Gustav Fischer.
Christopherson, S.; and Clark, J. 2007. Power in firm networks: what it means for regional innovation systems. Regional Studies 41:1223-1236.
Feldman, M.P. 2000. Location and innovation: the new economic geography of innovation, spillovers, and agglomeration. In In The Oxford handbook of economic geography 373-395 ed. Clark, G.L., M.A. Feldman, and M.S. Gertler. Oxford: Oxford University Press.
Coe, N. M.; Hess, M.; Yeung, H.W-C.; Dicken, P. and Henderson, J. 2004. ‘Globalizing' regional development: a global production networks perspective. Transactions Institute of British Geographers 29 468-84.
Cohen, R. B. 1981. The new international division of labour, multinational corporations, and urban hierarchy. In Urbanization and Urban Planning in Capitalist Society 287–315 ed. M. Dear and A. Scott. New York: Methuen.
De Filippis, J. 2001. The myth of social capital in community development. Housing Policy Debate 12:781–806.
Derudder, B. 2006. On conceptual confusion in empirical analyses of a transnational urban network. Urban Studies 43:2027-2046.
Derudder, B. and Witlox, F. 2010. World cities and global commodity chains: an introduction. Global Networks 10: 1 – 11.
Dicken, P.; and Malmberg, A. 2001a. Firms in territories: a relational perspective. Economic Geography 77:345-363.
Dicken, P.; Kelly, P. F., Olds, K. and Yeung, H. W-C. 2001b. Chains and networks, territories and scales: towards a relational framework for analysing the global economy. Global Networks 1: 89-112.
Dunning, J. H.; and Lundan, S. M. 2008. Multinational enterprises and the global economy. Cheltenham: Edward Elgar.
Edgington, D.W.; and Haga, H.1998. Japanese service sector multinationals and the hierarchy of Pacific Rim cities. Asia Pacific Viewpoint 39:16-178.
Friedmann, J. 2001. World cities revisited: a comment, Urban Studies, 38:2535 – 2536.
Friedmann, J., and Wolff, G. 1982. World City Formation: An agenda for research and action. International Journal of Urban and Regional Research 6:309-344.
Friedmann, J.; 1986. The world city hypothesis. Development and Change 17, pp. 69-84.
Friedmann, J.; 1995. Where we stand: a decade of world city research. In World Cities in a World-system. ed. P.L. Knox and P.J. Taylor 21-47. Cambridge: Cambridge University Press.
Gereffi, G., M. Korzeniewicz and R. P. Korzeniewicz (1994) ‘Introduction: global commodity chains' in G. Gereffi and M. Korzeniewicz (eds) Commodity chains and global capitalism, Westport, Conn: Praeger, 1–14.
Godfrey, B. J.; and Zhou, Y. 1999. Ranking world cities: multinational corporations and the global urban hierarchy. Urban Geography 20:268-281.
Grabher, G. 1993. The weakness of strong ties: the lock-in of regional development in the Ruhr area. In The embedded firm: on the socioeconomics of interfirm relations. Ed. Grabher G. pp. 255–278. London: Routledge.
Grabher, G. 2006. Trading routes, bypasses, and risky intersections: mapping the travels ‘networks' between economic sociology and economic geography. Progress in Human Geography 30: 163-189.
Grabher, G. and Powell, W.W. 2004. Exploring the webs of economic life. In Grabher, G. and Powell, W.W., ed. Networks. Cheltenham: Edward Elgar 1-36.
Granovetter, M. 1985. Economic action and economic structure: the problem of embeddedness. American Journal of Sociology 91: 481 – 510.
Hall, P. and Soskice, D. 2001. Varieties of capitalism. The institutional foundations of comparative advantage. Oxford: Oxford University Press.
Hannemann, R., and Riddle, M. 2005. Introduction to Social Network Methods. Riverside: University of California.
Harvey, D. 2006 Spaces of global capitalism: a theory of uneven geographical development. Verso.
Hedlund, G. 1986. The hypermodern MNC-A heterarchy. Human Resource Management 25:9-35.
Hedlund, G. 1993. Assumptions on hierarchy: an application to the multinational corporation. In Organization theory and the multinational corporation. Ed. S. Ghoshal and D.E. Westney. London: Macmillan.
Hedlund, G. 1994. A model of knowledge management and the N-form corporation. Strategic Management Journal 15: 73-90.
Hein, W. Die Ö konomie des Archipels und das versunkere Land. E + Z 41:304-307.
Henderson, J.; Dicken, P.; Hess, M.; Coe, N. and Yeung, H. W-C. 2002. Global production networks and the analysis of economic development. Review of International Political Economy 9:3 436-464.
Hess, M. 2004. ‘Spatial' relationships? Towards a reconceptualization of embeddedness. Progress in Human Geography 28: 165-186.
Hill, R.C.; and Kim, J.W. 2000. Global cities and developmental states: New York, Tokyo and Seoul. Urban Studies 37:2167-2195.
Hill, R.C.; and Kim, J.W. 2001. Reply to Friedman and Sassen. Urban Studies. 38:2541-2542.
Hymer, S. 1972. The Multinational Corporation and the Law of Uneven Development. In Economics and World Order. Ed. Bhagwati, J.N. New York: Macmillan.
Irwin, M. D.; and Hughes, H. L. 1992. Centrality and the structure of urban interaction: measures, concepts, and applications. Social Forces 71.
Ivarsson, I. 1996. Integrated international production: A survey of foreign transnational corporations in Sweden. Department of Geography, University of Göteborg, Series B:90.
Knaap, G.A van der. 2007. Urban Network Development under Conditions of Uncertainty. In Cities in Globalization. Ed. P.J. Taylor, B. Derudder, P. Saey, F. Witlox, F. pp. 890-107. London: Routledge.
Knoke, D; and Kuklinski J. H. 1982. Network Analysis. Beverly Hills, CA: Sage
Koestler, A. 1978. Janus – a summing up. New York: Random House.
Lin, N.; Cook, K. S.; and Burt R. S. 2001. Social Capital: Theory and Research. New York: Aldine de Gruyter.
Maskell, P. 2001. The firm in economic geography. Economic Geography 77:329-344.
Meijer, M. 1993. Growth and decline of European cities: changing positions of cities in Europe. Urban Studies 30:981–90.
Meyer, D. R. 1986. The world system of cities: relations between international financial metropolises and South American cities. Social Forces 64:553–81.
Phelps, N. A.; and Fuller, C. 2000. Multinationals, intracorporate competition, and regional development. Economic Geography 76:224-243.
Polanyi, K. 1944. The great transformation. The political and economic origins of our time. Boston. Beacon Press.
Pred, A. 1977. City systems in advanced economies: past growth, present processes and future development options. London: Hutchinson.
Rozenblat, C., and Pumain, D. 2006. Firm linkages, innovation and the evolution of urban systems. In Cities in Globalization. Ed. P.J. Taylor, B. Derudder, P. Saey, and F. Witlox. London: Routledge.
Sassen, S. 1991. The Global City: New York, London, Tokyo. Princeton, N.J: Princeton University Press.
Sassen, S. 2001b. Global cities and developmental states: how to derail what could be an interesting debate: a response to Hill and Kim. Urban Studies 38:2537 – 2540.
Sassen, S. 2001a. The Global City: New York, London, Tokyo 2d ed. Princeton, N.J.: Princeton University Press.
Sassen, Saskia. 1994. Cities in a World Economy. Thousand Oaks, California: Pine Forge Press.
Schoenberger, E. 1997. The cultural crisis of the firm. Oxford: Blackwell.
Simmel, G. 1890. Ü ber soziale Differenzierung. Soziologische und psychologische Untersuchungen. Leipzig: Duncker and Humblot.
Smith, D.A., and Timberlake, M. 1995. World cities: a political economy/global network approach. Research in Urban Sociology 3:181-207.
Storper, M. 2009. Roepke Lecture in economic geography Regional Context and Global Trade. Economic Geography 85:1 1-21
Storper, M., and R. Salais. 1997. Worlds of production: The action frameworks of the economy. Cambridge, MA: Harvard University Press.
Taggart, J. H. 1997. Autonomy and procedural justice: a framework for evaluating subsidiary strategy. Journal of International Business Studies 28:51–76.
Taylor, P. J., Catalano, G.; and Walker, D. R. F. 2002. Measurement of the world city network. Urban Studies 39:2367–2376.
Taylor, M. 1995. The business enterprise, power and patterns of geographical industrialisation. In The industrial enterprise and its environment: spatial perspectives. Ed. S. Conti, E. Malecki, and P. Oinas 99–122. Aldershot, U.K.: Avebury.
Taylor, M. 1996. Industrialisation, enterprise power, and environmental change: An exploration of concepts. Environment and Planning A 28:1035–51.
Taylor, M.; and Asheim, B. 2001. The concept of the firm in economic geography. Economic Geography 77:315-328.
Taylor, P. J. 2001. Specification of the world city network. Geographical Analysis 33:181-94
Taylor, P. J. 2006. Parallel paths to understanding global intercity relations. American Journal of Sociology 112:881 94.
Taylor, P. J., Ni, P., Derudder, B., Hoyler, M., Huang, J., Fengyong, L., Pain, K., Witlox, F., Yang, X., Bassens, D., and Shen, W. 2009. The way we were: command-and-control centres in the global space economy on the eve of the 2008 geo-economic transition. Environment and Planning A 41:7-12.
Taylor, P. J.; Derudder, B.; Hoyler, M.; Huang, J.; Lu, F.; Pain, K.; Witlox, F.; Yang, X.; and Shen, W. 2009. Measuring the world city network: new developments and results. GaWC Research Bulletin 300. University of Loughborough.
Taylor, P.J. 2004. World City Network: A Global Urban Analysis. London: Routledge.
Taylor, P.J., Hoyler, M., Verbruggen, R. 2008. External urban relational process: introducing central flow theory to complement central place theory. GaWC Research Bulletin 261. University of Loughborough.
Todeva, E. 2006. Business networks: strategy and structure. New York: Routledge.
Wall, R.S. 2009a. Netscape: cities and global corporate networks. Rotterdam: Haveka.
Wall, R.S. 2009b. The relative importance of Randstad cities within comparative worldwide corporate networks. Journal of Economic and Social Geography 100: 250-258.
Watts, D.J. 1999. Small worlds. Princeton, NJ: Princeton University Press.
Whitley, R. 1999. Divergent capitalisms: the social structuring and change of business systems. Oxford: Oxford University Press.
Yeung, H. W.-C 1998. Capital, state and space: contesting the borderless world. Transactions of the Institute of British Geographers 23:291–309.
Yeung, H. W.-C 2005. The firm as social networks: an organizational perspective. Growth and Change 36:307-328.
Young, A.K. 1979. The Sogo Shosha: Japan's Multinational Trading Company, Boulder, Westview Press.
Sanchez, R. (ed.). 1999. Beyond the boundaries of the firm: integrating theories of the firm and theories of markets. Oxford: Basil Blackwell.
** Bert van der Knaap, Erasmus School of Economics, Department of Applied Economics and ERIM, Erasmus University Rotterdam.
1. Sogo shosha refers to a specific type of Japanese firm: the general trading company, which is a complex agglomeration of diversified firms.
Note: This Research Bulletin has been published in Economic Geography, 87 (3), (2011), 267-308