This Research Bulletin has been published in Urban Studies, 42 (9), (2005), 1593-1608.
Please refer to the published version when quoting the paper.
INTRODUCTION: RETURN TO LEADING WORLD CITIES
In the globalization literature important cities have come to be known by two different names: world cities and global cities. For some the two terms are interchangeable; the classic case is Anthony King's (1990) book entitled Global Cities wherein all chapter headings refer to 'world cities'. Saskia Sassen (1991), on the other hand, in her Global City, makes a case for using this new term to distinguish contemporary leading cities from past 'world cities': it is global cities that are unique to our age. However, one problem of this latter usage has been to imply that globalization is the preserve of just the major cities across the world. Hence, Marcuse and van Kempen (1998) coined the term 'globalising cities' to make the point that globalization processes are to be found operating across many more cities than just a few 'global cities'. Taking the latter point on board I prefer to use the phrase 'cities in globalization' and in empirical studies of large numbers of such cities I use a model I call the 'world city network' (Taylor, 2004). This eschewing of Sassen's widely used concept does not necessarily mean that I reject her argument for the existence of special leading cities that we may identify as 'global'. The world city network most certainly is not a collection of largely equal cities; it exhibits strong hierarchical tendencies in its formation. Therefore, I have reserved the right to return to this matter and explore the existence of 'global cities' in the sense of especially important cities in the world city network (Taylor, 2004, p.42).
I have used the term 'leading world cities' in my title to indicate an initial neutral stance in the question of global cities. An important criticism of the impact of Sassen's (1991) study was that it encouraged research on just a few leading cities at the expense of many others. At GaWC 1 we responded to this situation by carrying out studies involving large numbers of cities, with data covering more than 300 cities (Taylor et al., 2002a), and most analyses are of over 100 cities (Taylor et al.,2002b), with some over 200 (Derudder et al., 2003). In this paper I return to focussing upon leading cities but this does not mean that such large-scale perspectives are being abandoned. Quite the contrary, in fact: leading cities can only be identified and understood in the context of their relations with myriad other cities across the world. That is to say, if London is indeed to be interpreted as a 'global city' it will be because many other cities, through their dependences and inter-dependences with London, make London special. This paper provides empirical evaluations of cities within the world city network to find nodes that appear as notably important.
The paper builds upon previous papers that have evaluated cities in terms of economic relations (Taylor et al., 2002a and b); the distinctiveness of this paper is that the evaluation extends beyond economic globalization. Data are derived and analysed for cultural, political and social globalization processes to define several networks of cities in globalization. This is important because 'cities as economic centres' is only part of their contribution to contemporary global practices. As Allen, Massey and Pryke (1999, p.5) have pointed out: "There is not simply one network of cities 'out there', but rather various networks with different paths of development .". This is also, of course, in keeping with Castells' (1996, p.224) concept of the middle strata of his space of flows in which Sassen's economic flows through global cities is but one of many types of networks. Thus 'multiple networks' are a key feature of this paper, bringing together results from different global realms of social activity into a single argument.
The paper divides into five sections. It begins with an outline of the interlocking network model upon which all analyses are based. The model defines the data needs and these are addressed in the second section. These data are then analysed in two distinctive ways to evaluate the importance of cities in worldwide networks. In the third section, cities are measured in terms of their nodal characteristics within networks. Two measures are involved: computing the size of the city as a node in the network, which is the quantity of network agency that is found within a city; and computing how these network agencies link the city to other cities, which is the network connectivity of the city. These measures represent the site and situation properties of a city within a network as previously defined (Taylor, 2001, p.184-7). In the fourth section, cities are measured as 'articulators' of particular processes within the network; this identifies cities as critical nodes within the operation of subsections of a world network. The idea of articulation is taken from Friedmann (1995); principal components analyses are employed to measure this process as previously deployed (Taylor et al., 2004; Taylor, 2004a). In the final substantive section, results from both analyses are brought together to create a taxonomy of cities to provide a composite picture of leading world cities in the original spirit of Peter Hall's (1966) functionally comprehensive concept.
THE MODEL: INTERLOCKING NETWORKS OF CITIES
This model has been outlined in detail elsewhere (Taylor 2001, 2004a) and here I provide a basic outline followed by a simple example of the first two measures, nodal size and interlock connectivity.
The model defines a process of world city network formation. The agents of this formation are institutions that use cities as networks in the everyday pursuit of their goals. Among the agents used below are financial and business service firms, non-governmental organizations, and media conglomerates. The world city network is constituted as the aggregate of agents' office networks (offices are where the practice of network formation is on-going). The model defines an 'interlocking network' because the agents 'interlock' cities through their activities (Knote and Kuklinski, 1982). For instance, global law firms in the practice of providing services for corporate customers will use offices in different cities through which flows of information, knowledge, instruction, ideas, plans, strategies, personnel, etc. will flow. The world city network is constituted as an aggregation of myriad such intra-firm connections.
Given a set of n institutions with offices across m cities, wherein the activity of institution i in city j is vij, there are two basic ways in which the importance of cities within the network can be measured. They are stated formally as follows.
1. Size of a node as measure of city site. The simplest measure of importance is the total activity within a city, which defines the size of the node in the network. Thus
where Sa is the network size of city a. This is a site measure because size incorporates information only on city a and no other city.
2. Connectivity of a node as measure of city situation. Interlock connectivity is the main measure of importance of a node in this model. It assumes that the amount of activity of an institution in a city generates a proportionate amount of flows from that city to other cities. Thus flows between pairs of cities can be estimated as proportionate to the products of their activities within a given institution's office network. Summing all these products for all institutions in a city across every other city in the network defines interlock connectivity:
(where a does not equal j) (2)
where Ca is the interlock connectivity of city a. This connectivity of city is a situation measure because it relates city a to all other cities within the network.
The data requirements for such a model consist of an 'activity value matrix' that arrays cities against institutions. In each cell, a score indicates the activity in a city by an institution pursuing its usual business. Such a set of data has been created in Table 1 to illustrate the two measures defined above. In this simple matrix there are 6 institutions that have offices across a network of 5 cities - I have invented the latter to represent five different strata of cities found in world city network research; thus Nylon represents a top 'global' city, Hong Sing a very important Pacific Asian city, Amsfurt a very important western European city, Boslanta an important US city, and Manbrum a leading English city outside London. Activity values range from 0 (the institution has no presence in that city) to 5 (indicating the institution's HQ city). Thus each row shows the mix of institutional activity in a city and each column indicates the geography of activity in an institution's office network.
The final two columns of Table 1 show nodal size and interlock connectivity measures. They broadly reflect the city strata they are designed to represent: thus Nylon is by far the most important city on both measures and Manbrum the least important. But the two measures do not simply replicate each other. Notice that the ranking is different for Hong Sing and Amsfurt on the two measures: the latter is the larger node but the former is the more connected. This illustrates the difference between the two measures for although Amsfurt has two headquarters to none in Hong Sing, for the other four firms Hong Sing has more overall activity (especially in the largest institution, 1), which generates more total connectivity. Hence, although closely correlated, the two measures are picking up different features of nodes within the network: one focuses on site attributes, the other on situation relations.
DATA FOR MULTIPLE NETWORKS
The interlocking model was initially developed to measure the world city network using advanced producer service firms following Sassen's (1991) identification of the latter as key agents of global city formation. This research produced site and situation measures from a 315 cities x 100 global service firms matrix wherein 'service values' ranging from 0 to 5 indicated the quantity and quality of service provided by a firm in a city (Taylor et al., 2002a; Taylor, 2004a). Results from this work are reported in this paper alongside new results from 7 other data sets similarly constructed plus two data sets from other researchers. The latter two were not collected for measurement within the interlocking network model but do produce comparable results as will be shown below. All data sets used in this paper are listed in Table 2.
I have divided the data sets into four broad spheres of activity that define economic, cultural, political and social globalizations respectively. Of course, any scrutiny of this division of activities will show it to be quite fuzzy but it is a useful starting point because it emphasizes the wide range of data that are brought together in this paper. And, importantly, the data sets, although varying in size, are all relatively large. This is vital for measuring nodal size and interlock connectivity which both involve aggregating activity values: the more institutions the less effect particular idiosyncrasies of particular firms are reflected in the results.
The economic data sets are from two sources. First, there are the original GaWC data on advanced producer firms previously analysed in detail in Taylor (2004a). Here I break down the data into six sectors to explore patterns below the general producer service level. Second, I will report on results from Alderson and Beckfield's (2004) large-scale study of firms in cities. They use data on headquarters and subsidiaries of 446 of the world's 500 largest corporations for which they could find the necessary information.2
The cultural datasets also comprise two separate information gatherings. First, I use data collected by Stefan Krätke (2002) on media conglomerates and previously analysed in Krätke and Taylor (2004). These data record the number of affiliates each conglomerate has across cities. Second, data on design professional firms in engineering and architecture have been collected as activity matrices (i.e. like Table 1).3 From these data, architectural firms have been identified and analysed separately (see also Knox and Taylor (2004)).
There are three separate political data sets that represent the scales at which governance primarily operates. At the global scale data have been collected on UN agencies that operate through cities across the world.4 At the national scale, the representations of states through cities across the world is recorded as presences in capital cities and other important cities. Finally, at the local level, memberships of three world-wide organizations of cities is recorded - WACLAC (World Alliance of International Associations of Cities and Local Authorities) which engages mainly in UN lobbying (including ATO (Arab Towns Association with 384 members', CITYNET with 114 members, EUROCITIES with113 members, and Metropolis with 78 members); Global Cities Dialogue with 142 members; and World Federation of United Cities with 1380 members. Only the first two data sets constitute activity matrices; the latter data are included to indicate local-level processes but only limited analysis is possible as detailed below.
The social globalization material encompasses two contrasting data sets. First, NGOs have been selected from 2001-2002 Yearbook of International Organizations from categories that relate to social movements: Human Rights, Humanitarian and Environmental. From these lists, NGOs for which adequate information could be found were used to create an activities data matrix.5 For further details and analysis see Taylor (2004c). Finally, I draw upon the comprehensive study of scientific papers by Matthiessen et al. (2000) who analyse millions of scientific papers and their citations. I focus upon multi-authored papers in their work that show academic connections between cities and their universities across the world.
These 16 data sets and subsets each allow me to estimate the importance of cities in terms of size attributes and situation relations. Although three of the data sets (multinational enterprises, city government organizations, and scientific papers) are not in the form required for a GaWC analysis, results from these data are adapted to indicate the importance of cities through their site and situation. Thus I am able to provide a uniquely comprehensive assessment of the importance of leading world cities across four spheres of globalization.
THE SITE AND SITUATION OF LEADING CITIES IN WORLDWIDE NETWORKS
I present empirical results from all 16 data sets in the same way to facilitate comparison. For both site and situation the top five cities are identified and ranked. Also for each set of five cities, the levels of size and connectivity of the cities ranked from 2 to 5 are presented as proportions of the top city's scores. This allows evaluation of degree of top city dominance across different spheres of activity. In this section results are reported and briefly commented upon, overall assessment of the results is left to a later discussion wherein subnet articulation cities are also included. Results are presented by spheres of globalization.
Economic globalizations. General indications of the leading world cities in economic globalization are shown in Table 3. It is noteworthy that four cities appear in every top five: London, New York, Paris and Tokyo. This is entirely consistent with the consensus on leading world/global cities in the literature. But there are important differences between the listings. In advanced producer services, London and New York clearly dominate in both size and connectivity but for multinational corporations it is Tokyo that dominates for size and New York for connectivity, with London a distant second and Tokyo falling to fourth. The latter result shows that, while Tokyo has more corporations, those in New York have more links across the world. Breaking down the advanced producer service results into sectors also shows a variety of city lists (Table 4). As expected, London and New York dominate all lists with New York overhauling London for advertising and management consultancy. Tokyo joins with London and New York to form a troika at the top of the banking/finance sector lists. Hong Kong, as gateway to the booming Chinese producer service market, is also prominent across several lists.
Cultural globalizations. Although the categories comprising this activity sphere are quite different, their top five lists show similarities (Table 5). Again London and New York dominate. However, in media they are rivalled in importance by Los Angeles and Paris. In architecture other US and Pacific Rim cities come to the fore (but not Hong Kong). This reflects the importance of the Pacific Asian urban development market serviced from both the USA and Australia. However, overall, the dominance of London in architectural activities is quite exceptional.
Political globalizations. The results for the three scales are quite distinctive (Table 6). At the global scale UN agencies bring in 'third world' cities for the first time. The particular differences between site and situation are also noteworthy is the case. Whereas New York and Geneva totally dominate the site measures, for situation Geneva is number one and New York is not in the list (it is actually ranked sixth). New York in the UN network is the clearest case of an exceptionally large node whose connections are less than expected. For the national scale, diplomatic missions are pretty well as expected, with Washington, DC topping both lists followed by capitals of major countries and New York because of its UN HQ function. The list for organizations of local governments is totally distinctive mixing European and 'third world' cities. Although different from the cities listed for UN connectivity, this does suggest that it is regions outside the Americas that spawn cities with political practices that are global.
Social globalizations. Although London tops three of the four listings in Table 7, two quite distinctive patterns emerge. For NGOs 'third world' cities appear again and, in fact, dominate when it comes to connections. These results do tend to imply that in this 'global civil society' wherein NGOs are seen as pioneering, large nodes are in the 'first world' (control centres) with highly connected nodes in the 'third world' where the practice necessarily is (Glasius and Kaldor, 2002). In contrast, academic links most obviously do not include 'third world' cities. Two features are noteworthy here. First, there is the identification of two large Japanese nodes. However, the nodal size of Japanese cities does not translate into leading connected cities. Second, in terms of connectivity, here we have US cities at last appearing to 'punch their weight' within world city networks.
The results above are summarised in Table 8. Cities are listed that have more than two mentions in the above tables - i.e. they must appear in at least two activities. Just 15 cities are thus identified and London and New York stand out.
SUBNET ARTICULATOR CITIES
The measures deployed so far have been summaries of importance of cities over the whole network. Such measures may miss out important cities that are at the centre of sub-networks within the larger whole. All our descriptions of the world city network have indicated that there are key cities that are strategically positioned in the network at the intersections of regional locations and functional activities (Taylor 2004, Taylor et al. 2004). To identify such cities requires techniques that divide the overall network into coherent parts wherein key cities can be identified. The method I use is principal components analysis.
Principal components analysis is a technique that reduces a data matrix to its major dimensions (Rummel 1970). With an activities matrix the locational strategies of n institutions - the geography of their office network - are reduced to k common service patterns, where k is appreciably less than n. Basically institutions are clustered into groups on the basis of like geographies. It is this technique - R-mode principal components analysis with varimax rotation - that has been applied to the advanced producer service data to good effect (Taylor 2004; Taylor et al. 2004). The results consist of component loadings to show which institutions belong to which dimension-cluster of institutions, and component scores which show which cities are the important within the common locational strategy. The latter are given as standardized measures so that, if the scores are normally distributed, results appreciably above two should be relatively rare. One of the features of using this technique with activities data matrices is that scores are generated that are very much larger than two. This indicates very strong concentrations of activity in particular cities. These are interpreted as subnet articulator cities, cities where concentrations of activities exist within specific parts (dimensions/components) of the world city network.
The technique is illustrated in Table 9 where the data from Table 1 are subjected to such an analysis. The data matrix is reproduced in the table to facilitate interpretation. In this case the geographies of the six initial institutions are reduced to three common patterns (i.e. n = 6 is reduced to k = 3). From the loadings it can be seen that component I represents institutions 1, 3 and 6; component II represents institutions 2 and 5; and component III represents just institution 4. In this paper I am interested particularly in the scores. These show for component I that Nylon has by far the largest score - note that the three institutions loading on this component have activity values of 5 for Nylon. In other words Nylon is the articulator city for this subnet. Similarly Amsfurt is the articulator city for component II and Boslanta is the articulator city for component III, Again there is a simple relation to these two cities having activity values of 5 with the institutions making up the component. In a large-scale analysis relationships between data and scores will be more complex but the general principle will hold - articulator cities indicate a concentration of activity for a given component. Also, since we would expect approximately 95% of standardised scores (19 out of every 20) to be below ±2 in a typical normal distribution, the 15 scores from this simple analysis are relatively low. With larger analyses, some scores will be found above two, some appreciably so.
This principal components technique has been applied to eight of the 16 data sets shown in Table 2. For economic globalization, I use the whole producer services data (from Taylor et al 2002b and Taylor 2004) plus the subset for banking/finance (other sectors have two few firms). For cultural globalization, I use the media data (from Krätke and Taylor 2004) and both architecture- engineering and the architecture subset (from Knox and Taylor 2004). For political globalization I use the UN agency data and the diplomatic offices data (from Taylor 2004d). Finally, for social globalization I use just the NGO data (from Taylor 2003). Other data sets are not of the required structure for this exercise. Nevertheless there is a good range of analyses covering all four globalizations. I do not provide full details of these analyses here, just the scores to identify articulator cities.
In Table 10 all such scores over four are listed from these eight analyses. For a standardised value, four is an extremely high number and indicates very high concentrations of subnet activity. However using this high threshold still generates 41 cases of articulator cities. These are ranked by size in Table 10. The top six cases, all with scores of eight and above, illustrate the results quite well: there are subsets of media conglomerate activities heavily concentrated in London and Paris, subsets of UN agency activity activities heavily concentrated in Vienna and New York, and subsets of NGO activities heavily concentrated in London and Washington. Note that Vienna, which did not feature in the site and situation analyses, has its own subnet that it dominates. There are seven instances where the articulation is shared between cities: for instance, New York articulates a media subnet with Los Angeles and a law subnet with Washington, DC.
One feature that is particularly interesting from these analyses is the fact that economic articulations have lower scores, less concentrated activity, than political, cultural and social articulations. This implies that the world city network of economic globalization has less hierarchical tendencies than the other city-based globalizations (most producer service analysis scores do not meet the threshold of four). Put another way, non-economic globalizations appear to create very hierarchical subnets whereas economic globalization subnets suggest a more horizontal network pattern. This is an important finding for debates over the nature of globalization and its future (Taylor, 2004e).
A PROVISIONAL TAXONOMY OF LEADING WORLD CITIES
The nature of contemporary globalization is, of course, a highly contested question. But at the heart of most debates there is the notion of an acceleration of transnational activities and transactions that has (the potential) to change the spatial configuration of society. In Castells' (1996) terms there is a dominance of spaces of flows over spaces of places in a new network society. Cities appear in this argument as prime grounding points, locales where globalization is produced and reproduced. Short and Kim (1999, p.9) have provided the best succinct expression of this argument:
Globalization takes place in cities and cities embody and reflect globalization. Global processes lead to changes in the city and cities rework and situate globalization. Contemporary global dynamics are the spatial expression of globalization, while urban changes reshape and reform the processes of globalization.
Cities have always projected their influence over surrounding regions economically, culturally, politically and socially, and today such projections have become global. Thus Jacob's (1984) influential model of the projection of economic processes through cities can be adapted to interpretation of contemporary world cities in a global economy (Taylor 2004a). Similarly, contemporary concepts of global civil society and global governance have been interpreted as processes operating in and through cities (Sassen 2002; Taylor 2004d and e). This is the theoretical context within which the results above should be interpreted. However this paper is essentially an empirical contribution and therefore I will use the method of taxonomy to provide a provisional means of thinking how results link to theory.
Taxonomies are orderings of objects that reflect their similarities and differences thus supplying basic suggestions about the processes of their construction. By looking at objects in the whole, taxonomies provide critical first steps towards understanding. In the case of contemporary leading cities such a method allows me to return to the the terms 'global city' and 'world city' and suggest a differentiation. Hierarchical tendencies have always appeared prominently in analyses of the world city network (Taylor 2004a) and this is strongly represented in the results above. Thus I suggest use of the term 'global city' to describe the top echelons of these hierarchical patterns. This is identified in two ways: first, important global contributions across a range of spheres of activity; and second, very strong global contributions in a particular sphere of activity. The remaining leading world cities also consist of two groups: subnet articulator cities and cities with worldwide contributions in particular spheres of activity. In allocating cities I have used results above supplemented by some findings that did not reach the high thresholds I have used: f or top site and situation cities I have considered top 10 positions. The final taxonomy is shown in Table 11.
Five levels of global city are identified. First, and clearly above all others, there are London and New York. All previous research has highlighted the dominance of these two cities in the world city hierarchy (Taylor 2004a) and they emerge here as the most important 'all-round' global contributors. They are followed by three cities that make smaller all-round contribution and with particular cultural strengths: Los Angeles, Paris and San Francisco. Finally, among 'all-rounders' there are seven incipient world cities identified in Table 11. In the second category of global niche cities, the three leading Pacific Asian cities are critical economic nodes in the world city network and there are also three critical nodes that are non-economic: Brussels, Geneva and Washington, DC. Thus a total of 18 cities are deemed to be global, actual or incipient.
The remaining world cities encompass articulator and niche cities. The former are focussed upon subnets and there are 13 distributed between the three non-economic spheres. Classic examples are Vienna at the centre of a UN agency subnet and Nairobi at the centre of a NGO subnet. There are 21niche world cities identified of which seven have important concentrations of economic activities and 14 concentrations of non-economic activities. Frankfurt is typical of the first group with its concentration of banks while Manila is typical of the second group with its concentration of NGOs.
These two sets of cities represent the upper echelons of the hierarchical tendencies in world city networks. To reiterate a point made in the introduction, they do not encompass all globalization processes, all cities as so involved, but they are the key locales that network formation agents are using in their everyday activities that are creating world city networks.
This paper has been a heavily empirical treatise. I make no apology for this. World city research has been notorious for its evidential deficit (Short et al. 1996; Taylor 1999; Alderson and Beckfield 2004). In this paper I have show some pathways taken in overcoming this problem so that it is possible to begin to say something empirically sound about world cities and their inter-relations. To be sure, the techniques used are crude and basic and the same can be said for the data. Certainly I do not pick up key processes that are important: any analysis that does not identify such specifically important world-cultural flows as those to Las Vegas, Mecca and Rome must be suspect. Thus, whereas this paper includes more data than has appeared in any previous text on the subject, it is self-evidently not a comprehensive study of its subject. My study is preliminary in both technique and data - it is hoped both are greatly improved upon in the near future - but it does represent a first evidential step to understanding cities and their interrelations at a reasonably comprehensive level. In Short and Kim's (1999, 9) terms I hope I have helped bring understanding of globalization down from ideas 'pitched at . a stratospheric level' to a concrete illustration of how activities are producing and reproducing global spaces of flows.
ALDERSON, A. and BECKFIELD, J. (2004) Power and position in the world city system, American Journal of Sociology, 109, pp. 811-51.
ALLAN, J., MASSEY, D. and PRYKE, M. (1999) Introduction, in J. Allan, D. Massey and M. Pryke (Eds) Unsettling Cities, pp.1-6. London: Routledge.
CASTELLS, M. (1996) The Rise of Network Society. Oxford: Blackwell.
DERUDDER. B., TAYLOR, P. J., WITLOX, F. and CATALANO, G. (2003) Hierarchical Tendencies and Regional Patterns in the World City Network: A Global Urban Analysis of 234 Cities, Regional Studies, 37, pp. 875-86.
FRIEDMANN, J. (1995) Where we stand: a decade of world city research, in P.L. Knox and P. J. Taylor (Eds.) World Cities in a World-System, pp. 21-47. Cambridge: Cambridge University Press.
GLASIUS, M. and KALDOR, M. (2002) The state of global civil society: before and after September 11, in M. Glasius, M. Kaldor and H. Anheier (Eds.) Global Civil Society 2002, pp. 3-34. Oxford: Oxford University Press.
HALL, P. (1966) The World Cities. London: Heinemann
JACOBS, J. (1984) Cities and the Wealth of Nations. New York: Vintage
KING, A. (1990) Global Cities. London: Routledge
KNOKE, D. and KUKLINSKI, J. H. (1982) Network Analysis. Beverly Hills: Sage.
KNOX, P. L. and TAYLOR, P. J. (2004) Globalization of architectural practice, Journal of Architectural Education,
KRÄTKE, S. (2002) Medienstadt. Opladen: Leske and Budrich.
KRÄTKE, S. and TAYLOR, P. J. (2004) A world geography of global media cities, European Planning Studies,
MARCUS. P. and van Kempen, R. (2000) Globalising Cities. Oxford: Blackwell
MATTHIESSEN, C. W., SCHWARZ, A. W. and FIND, S. (2000) Research gateways of the world: an analysis of networks based on biometric indicators, in A. E. Andersson and D. E. Andersson (Eds) Gateways to the Global Economy, pp. 17-30. Cheltenham: Edward Elgar
RUMMEL. R. J. (1970) Applied Factor Analysis. Evanston: Northwestern University Press.
SASSEN, S. (1991) The Global City. Princeton, NJ: Princeton University Press.
SASSEN, S (2002) Global Cities and diasporic networks: microsites in global civil society, in M. Glasius, M. Kaldor and H. Anheier (Eds.) Global Civil Society 2002, pp. 217-40. Oxford: Oxford University Press.
SHORT, J. R., KIM, Y., KUSS, M. and WELLS, H. (1996) The dirty little secret of world city research, International Journal of Regional and Urban Research, 20, 697-717
SHORT, J. R. and KIM, Y-H (1999) Globalization and the City. London: Longman.
TAYLOR, P. J. (1999) So-called 'world cities': the evidential structure within a literature, Environment and Planning A, 31, 1901-04
TAYLOR, P. J. (2001) Specification of the world city network, Geographical Analysis, 33, pp. 181-94
TAYLOR, P. J. (2004a) World City Network: a Global Urban Analysis. London: Routledge
TAYLOR, P. J. (2004b) Parallel paths to understanding global inter-city relations, GaWC Research Bulletin No. 143.
TAYLOR, P. J. (2004c) The new geography of global civil society: NGOs in the world city network, GaWC Research Bulletin No. 144.
TAYLOR, P. J. (2004d) Political geography of global governance and global civil society, GaWC Research Bulletin No.
TAYLOR, P. J. (2004e) Homo Geographicus : a geohistorical manifesto for cities, Review (Fernand Braudel Center), 26, pp. 37-60
TAYLOR, P. J., CATALANO, G. and WALKER, D. R. F. (2002a) Measurement of the world city network, Urban Studies, 39, pp. 2367-76.
TAYLOR, P. J., CATALANO, G. and WALKER, D. R. F. (2002b) Exploratory analysis of the world city network, Urban Studies, 39, pp. 2377-94.
TAYLOR, P. J., CATALANO, G., WALKER, D. R. F. and HOYLER, M. (2002c) Diversity and power in the world city network, Cities, 19, pp. 231-41
TAYLOR, P. J., CATALANO, G. and WALKER, D. R. F. (2004) Multiple globalizations: regional, hierarchical and sectoral articulations of global business services through world cities, The Service Industries Journal, 24, pp. 64-82
1. The Globalization and World Cities (GaWC) Study Group and Network is based at Loughborough University (UK) and collaborates with world city researchers across the world.
2. For a comparison between this research and GaWC researches see Taylor (2004b).
3. Collected for the Metropolitan Institute, Virginia Tech by Erin
4. Collected for the Metropolitan Institute, Virginia Tech by Clare Blake
5. Collected for the Metropolitan Institute, Virginia Tech by Troy Gravitt
Table 1: Pedagogic Case Study I: nodal size and interlocking connectivity
Table 2: Data sets for interlocking network modelling
* data collected for a different analysis by Alderson and Beckfield (2004), and adapted for interlocking modelling.
Table 3: Sites and situations of leading cities in economic globalizations
* from Alderson and Beckfield (2004) Table 3; site is defined by 'outdegree', connectivity by 'indegree'
Table 4: Sites and situations of leading cities in economic globalisations: Disaggregated by service sectors
Table 5: Sites and situations of leading cities in cultural globalisations
Table 6: Sites and situations of leading cities in political globalisations
Table 7: Sites and situations of leading cities in social globalisations
Table 8: Leading world cities by site and situation
(number of mentions in Tables 3 through 7)
Table 9: Pedagogic Case Study II: principal components analysis
Cities with HQ functions (values of 5) are emboldened
(ii) Sub-net articulator cities
High loadings and scores are emboldened
Table 10: Ranking of articulator cities within globalizations
Table 11: Taxonomy of leading cities in globalization
Well rounded global cities
Global niche cities - specialised global contributions
Subnet articulator cities
Worldwide leading cities
Edited and posted on the web on 14th July 2004
Note: This Research Bulletin has been published in Urban Studies, 42 (9), (2005), 1593-1608