GaWC Research Bulletin 198

GaWC logo
  Gateways into GaWC

A later version of this Research Bulletin has been published in Geography Compass, 2 (2), (2008), 559-574.


Please refer to the published version when quoting the paper.


Mapping Global Urban Networks: A Decade of Empirical World Cities Research

B. Derudder*


This survey article reviews the harvest of empirical world cities research since its formal inception about a decade ago. Taylor’s (1997) delineation of ‘the problem of evidence’ in this literature is used as a cue to assess the merits and inadequacies of recent empirical studies. To get a handle on the literature as a whole, a taxonomy of empirical approaches is developed. Major features of the dominant approaches in general and some important studies in particular are discussed. This discussion is thereupon used to analyze the extent to which these studies have managed to construct meaningful measures of transnational inter-city relations. This analysis reveals that significant steps have been taken, but some new points of attention can be raised, such as the lack of suitable theoretical reference points in some recent empirical studies and the disproportionate focus on a relatively narrow range of formal economic processes.


About a decade ago, leading geographer Peter Taylor (1997) launched an ambitious global research proposal under the umbrella of the Globalization and World Cities group and network (GaWC, Above all, Taylor’s paper was a plea to social science researchers to become involved in a large-scale project on trans-state processes, with a particular focus on ‘world cities’, which can provisionally and very generally be defined as the command and control centres of the global economy, i.e. metropolitan areas that function as organizing centres for the interdependent skein of material, financial, and cultural flows that together sustain contemporary globalization. In the decade that followed the publication of Jon Friedmann’s (1986) seminal ‘World City Hypothesis’, world cities research had gained considerable currency. With the benefit of hindsight, it is easy to determine the basic reason for this considerable success: because of its unambiguous focus on potentially more meaningful scales of economic and social activity (‘transnational’ rather than ‘national’ scales), world cities research had the intrinsic capability to break free from the iron grip of territorial states on the social imagination.

However, in spite of this literature’s anticipated potential, Taylor (1997) argued that the harvest of the first “decade of world city research” (Friedman, 1995) was somewhat disappointing. One of the major reasons for this situation was already implicitly spelled out in Smith and Timberlake (1995a), who usefully constructed a typology of the processes they thought world cities researchers should be studying. With three forms of processes – human, material, information – and four functions of these processes – economic, political, cultural, social – they suggested 12 different types of evidence for mapping world city formation. However, in practice, such analyses were very difficult because of a notable gap between this ‘data wish list’ and publicly available information sources. Put simply: world cities research was held back by a seemingly structural gap between the conceptual notion that cities increasingly derive their importance from myriad transnational functions and the actual evidence for such trends, and Taylor’s (1997) rationale for founding GaWC was explicitly based on this observation: one of the main goals of this virtual research centre (co-directed by Jon Beaverstock) was to offer some solutions to this ‘problem of evidence’, whereby GaWC-affiliates and many other researchers have followed suit by devising alternative approaches to the empirical study of world city formation. In this survey article, I aim to review the harvest of this decade of sustained empirical world cities research at large. In this respect, it can be said that this paper equally aims to complement the recently published Global Cities Reader co-edited by Brenner and Keil (2006), whose worthwhile analysis of the world cities literature is almost completely void of a discussion of the empirical research.

It should be noted from the outset that this literature review is most certainly not intended as an attempt to answer the potentially meaningless question which cities can be labelled as ‘world’ or ‘global’ cities. Rather, I will examine how the various researches have tried to overcome the lack of suitable data sources in empirical world cities research. To this end, the paper is divided in two main sections. The first section outlines a taxonomy of empirical approaches. The proposed classification is based on the data sources employed in the analyses, whereby information on infrastructure networks and corporate organization is deemed dominant. This taxonomy gives the reader a feel of the direction the empirical research has taken in recent years, but its prime purpose lies in its heuristic potential: this classification facilitates an integrated assessment of the various studies, which is the subject matter of the second section. I therefore return to Taylor’s (1997) paper to evaluate how and to what extent the literature has managed to use ‘relevant’ data. The last section recapitulates the main findings and raises some avenues for future research.

Towards A Taxonomy of Empirical Approaches

The Emergence of a Research Agenda

The starting point of this literature review is a paper by Short et al. (1996: 698) in which it was observed that the privileged position of cities such as London, New York and Tokyo was “more often asserted than clearly demonstrated” in the empirical world cities literature. The authors clearly objected to the pessimistic viewpoint advanced by founding father Friedmann (1995: 23), who had purported that “we lack unambiguous criteria for assigning particular cities to a specific place in the global system” (see also Friedmann, 1986: 71). Short et al. (1996) bemoaned the notable absence of proper empirical underpinnings, and set out to search for relevant and measurable criteria along the lines of the Smith and Timberlake (1995a) paper. This led them to the identification of five indicators of world city-ness, i.e. the presence of (i) major financial institutions, (ii) corporation headquarters, (iii) telecommunications and (iv) transportation infrastructures, and (v) global cultural events. The balanced coalescence of the different measures resulted in one of the first world city rankings obtained through fully fledged empirical analysis rather than informed speculation.

Although I suspect few researchers would now turn to this paper to copy its approach or make use of its results, the importance of Short et al.’s (1996) paper should not be underestimated. For one thing, it ascertained that empirical analysis was a bonafide strand of investigation in world cities research. Furthermore, the indicators proposed by Short et al. (1996) largely reflect the approaches that would come to dominate the empirical research as a whole, i.e. (i) an approach focusing on corporate organization (e.g. data on the number of corporate headquarters) and (ii) an approach focusing on infrastructure (e.g. data on the size of a city’s transportation infrastructure). Indeed, in spite of profound differences in actual information sources, data analysis techniques, and ways of presenting and framing results, most empirical studies can in hindsight be subsumed under this bifurcation. The success of both approaches can, of course, be traced back to their direct appeal: the corporate organization approach acknowledges that cities derive their status in large part from the presence of key offices of important firms, while the infrastructure approach recognizes that world cities require the presence of vast enabling infrastructures. The next paragraphs present a more detailed elaboration of the premises of some important studies within both approaches.

The Corporate Organization Approach

Empirical studies carried out within the corporate organization approach are basically large-scale assessments of the urban geography of global corporate organization. A prime example is Beaverstock et al.’s (1999) Roster of World Cities. In this paper, the authors presented a straightforward inventory of world cities based upon their level of advanced producer services. Very much in a Christallerian tradition, Beaverstock et al. (1999) identified and graded global service centres for accountancy, advertising, banking/finance and law. The aggregated results produced a roster of 55 world cities in which three levels of importance were discerned (the now well-known alpha, beta, and gamma categories of world cities).

This GaWC-paper became an instant classic, but in practice its rather simple approach was quickly abandoned in favour of a more refined framework initially laid out in Taylor (2001). This alternative methodology equally employs information on the presence of advanced producer services firms, but the data are treated in a totally different way. That is, they are manipulated as to obtain guestimates of the strength of particular inter-city relations, a transformation that is based on the assumption that such firms ‘interlock’ cities through their intra-firm communications of information, knowledge, plans, directions, advice, etc. to create a network of global service centers. Building on this specification, Taylor et al. (2002a) gathered information on the location strategies of 100 global service firms across 315 cities, after which assorted variants of this and similar datasets were analyzed with standard data analysis techniques (Taylor et al., 2002b,c; Derudder et al., 2003), while the transformed datasets were analyzed with network-analytical tools (Rossi and Taylor, 2005; Derudder and Taylor, 2005). The crucial point here is that these studies aspired to be more than mere service inventories: they are empirical assessments of world cities as nodes in a global service space of flows.

A whole range of other studies equally made use of information on the corporate geography of large firms, but in contrast to GaWC analyses they refused to discriminate between different types of businesses. Rather than exclusively dealing with global service firms, then, these researches made use of data sources bringing together leading multinational firms irrespective of the exact nature of their activities (such as the well-known ‘Fortune 500’ list). A prime example is an article by Godfrey and Zhou (1999), which was originally intended as a critique of the excessive focus on global corporate headquarters in world cities research. The authors argued that reliance on global headquarter locations alone – as in the Short et al. (1996) paper – distorts the analysis: key decisions are frequently relegated to regional headquarters, hence their proposal to equally include information on lower rungs of decision making (in this case regional headquarters of MNCs listed in the ‘Fortune 500’). Irrespective of this proposed refinement, however, it is clear that Godfrey and Zhou (1999) acknowledge that information on the spatial organization of MNCs can be used to identify world cities.

In parallel with the methodological shift within GaWC research, some researchers began to criticize such headquarter inventories (be it global or global/regional). Alderson and Beckfield (2004), for instance, argued that Godfrey and Zhou-like (1999) studies are in reality less-than-ideal assessments of the control potential retained in world cities; actual control and its spatiality remain essentially obscure in such an approach. Only a genuine assessment of the interconnections generated by multinational enterprises, Alderson and Beckfield (2004) claimed, allows deriving sensible mappings of world city-formation. To achieve this, the authors collected data on the headquarters of large multinational firms and all their subsidiary locations. Based on this information, they produced a directed (from headquarter to subsidiary) and valued (number of directions to and from a city) matrix linking 3692 cities across the world. In contrast to Godfrey and Zhou (1999), Alderson and Beckfield (2004) were able to apply a set of network-analytical tools that allowed them to evaluate the position of world cities through their inter-city power relations.

The Infrastructure Approach

The gist of this second dominant approach is that “[i]nfrastructure networks are often assumed to be important determinants of the economic potential of urban agglomerations” (Bruinsma and Rietveld, 1993: 919). Studies within this approach therefore essentially take on the form of large-scale assessments of the urban geography of infrastructure networks. In analogy with the basic bifurcation in the corporate organization approach (global service firms versus multinational corporations in general), one can distinguish between analyses of (i) telecommunications (e.g. the Internet backbone) and (ii) physical transportation infrastructures (e.g. airline networks).

Arguably the most well-known airline analyses are a number of inter-related papers by Smith and Timberlake (1995b, 2001, 2002). These contributions were a huge step forward compared to earlier such studies, in which it was somewhat naively assumed that air transportation statistics could directly inform patterns of world city-formation. Smith and Timberlake were able to go beyond such descriptive studies through the thoughtful application of a refined network-analytical framework. Their resulting assessments were a formal analysis of flows between about 100 of “the world’s great cities for six time points between 1977 and 1997, focusing on the changes in network characteristics for the entire global city system” (Smith and Timberlake, 2001: 1656).

Smith and Timberlake’s (2001, 2002) work relegated other airline-based studies, but it quickly became clear that their research efforts were in turn hampered by data problems. Some of these data deficiencies were spelled out in a recent commentary by Derudder and Witlox (2005a). One such data deficiency, Derudder and Witlox (2005a) argued, is the lack of information on the actual routes flown by passengers: standard airline statistics capture individual legs of a given trip rather than the trip as a whole. Empirical analyses will therefore likely overstate the importance of cities that function as airline hubs, while direct links between secondary cities and/or cities that are far apart simply cannot be measured. After having criticized earlier studies for their inadequate data, Derudder and Witlox (2005a) present a new and previously untapped dataset that is able to solve this problem. The dataset contains information on bookings made through so-called Global Distribution Systems (GDS), electronic platforms used by travel agencies to manage airline bookings. Because GDS-based databases hold information on actual trips, a more realistic picture of inter-city flows emerges. Transforming the initial database for the purposes of empirical world cities research, Derudder and Witlox (2005a) derived a 306 x 306 matrix that quantifies the relations between the most important cities in the world economy. Although this dataset has not yet been analyzed along the lines set out by Smith and Timberlake (2001, 2002), it clearly offers new possibilities for the empirical research within this approach.

Although airline data have been the preferred option in the infrastructure approach, there have equally been a fair number of studies that make use of information on the ‘network of networks’ (i.e. the Internet). Malecki (2002: 400), for instance, noted that Internet backbone networks tend to concentrate on important cities, which makes it possible to set “the spatial agglomeration of linkages and linkage sites (...) in the context of the urban hierarchy of world cities.” (see also Zook, 2005, 2006). However, although an analysis of these digital networks per se is authoritative, its usefulness in the context of empirical world cities research needs to be tempered from the outset. In the same paper, Malecki (2002) equally acknowledged that the relation between the Internet’s infrastructure and world city-formation is increasingly flawed because of the rise of a set of urban areas that serve as interconnection points between different regions. Rutherford et al. (2004), for instance, point out that Vienna and Prague have good network presence and quite large bandwidth connections because they act as gateways between the core area of western Europe and eastern Europe, which suggests that such cities will be deemed more crucial than they would have been otherwise1. In short, the emergence of ‘new network cities’ in the infrastructure that comprises the Internet makes such data less and less useful in the context of empirical world cities research (Townsend, 2001).

A Taxonomy of Empirical Approaches

There have, of course, been many more empirical studies than those discussed in the preceding paragraphs, but together they give the reader a feel for the shape the empirical world cities literature has taken since Short et al.’s (1996) paper. Table 1 summarizes the main studies along the breakdowns outlined throughout this section2. The table acknowledges that the basic bifurcation between corporate organization and infrastructure is somewhat more complicated because of the presence of a limited number of studies that (i) make use of other types of data (e.g. Taylor’s (2005) analysis of non-governmental organizations) and/or (ii) combine indicators from both approaches (e.g. Beaverstock et al., 2000b).

In Search of Information on Transnational Inter-City Relations

A Double Data Deficiency

In this section, I review how the studies listed in Table 1 have attempted to overcome the ‘problem of evidence’ identified by Taylor (1997). The latter paper is used as a cue to get a handle on the empirical literature as a whole: every study has a different take on things (actual information sources, data analysis techniques,...), so that an integrated appraisal needs to single out a number of issues affecting the entire literature. And this is exactly what Taylor (1997: 323) did when he specified that the empirical research at large was “impeded by a double deficiency of published data being both state-centric and dominated by attribute measurements.” The next two paragraphs deal with both sub-problems in turn.

The Lack of Transnational Data

It is important to note from the outset that Taylor’s (1997) call for the use of transnational data should not be mistaken for the suggestion that states do not matter in world city-formation3. Rather, it means that data should not be a priori shaped by states. The observation that transnational inter-city relations are heavily influenced by states is viable, but it should be a result rather than an artefact of the data. Although some studies have been able to circumvent the iron grip of states on data production, others still suffer from this deficiency.

This is particularly evident in studies making use of infrastructure data. For instance, in their review of earlier airline studies, Derudder and Witlox (2005a) argued that the lack of transnational data was a second major problem in such studies (in addition to the lack of information on actual routes): standard airline statistics detail international rather than transnational flows. One obvious consequence is the downgrading of US world cities, because important connections such as Los Angeles–New York and Chicago–New York are not incorporated in the ensuing analysis. This clearly comes to the fore in a study by Rimmer (1998), in which Chicago only appears on one of his maps as a ‘fourth-level’ link to Toronto, while Dublin appears on all maps because of its ‘first-level’ link with London. However, nobody would want to argue that Dublin is more important than Chicago as a world city; it only appears to be that way when one relies on international rather than transnational data (see also Taylor, 1999). Another example of such lingering state-centrism is apparent in Smith and Timberlake (2002: 123), who where confronted with the lack of information on the volume of air passenger traffic between Hong Kong and London. This important link did not feature in pre-1997 databases of the International Civil Aviation Organization (ICAO) because flights between London and Hong Kong were considered to be ‘national’4. While the London–Hong Kong route and the downgrading of US cities are obviously extreme examples, they clearly reveal how difficult it has been to overcome the problem of state-centric data within this approach (similar remarks can be made with respect to studies using telecommunications data). That said, the GDS-based dataset presented in Derudder and Witlox (2005a) is truly transnational in that it does not discriminate between national and international flows, which reaffirms this dataset’s potential in this research field.

Similar observations can be made with respect to studies that make use of information on the corporate geography of large firms, albeit that things are a little less clear-cut here. It is useful to bear in mind that multinational corporations (MNCs) are different from (still rare) transnational corporations (TNCs) in that they maintain strong links with a host state. This implies that if an empirical study exclusively uses data on global headquarters of MNCs, as in Alderson and Beckfield (2004), it will not be able to escape some degree of state-centrism. One obvious consequence is that cities in core states will be privileged over others, e.g. the marginal position assigned to cities that play a significant regional role but have relatively few global MNC headquarters (e.g. Sao Paulo and Singapore). Another consequence is that Tokyo will be overvalued compared to the likes of New York, because for Japanese MNCs participation in the global economy still means going through a lot of government channels given a fairly regulated national economy. Although Tokyo’s suggested dominance will ultimately reflect a fair amount of actual inter-city relations, it is clear that an exclusive focus on global headquarters of multinational firms somehow distorts this analysis.

Multiple ways to work around this problem have been developed, whereby all ‘solutions’ entail the identification of multiple levels of corporate decision making and/or importance. The most straightforward solution has been Godfrey and Zhou’s (1999) proposition to include regional headquarters in the empirical framework, which has the simultaneous effect of downsizing Tokyo’s dominance and appreciating the importance of cities such as Sao Paulo and Singapore. More refined solutions have been put forward by Rozenblat and Pumain (2006) and Taylor et al. (2002a). Rozenblat and Pumain (2006) base their empirical analysis on a survey of MNCs. The collected data include information on the location of the various sites and the ownership ties that link these sites together in a single corporate organization. Their approach is quasi-identical to that of Alderson and Beckfield (2004), albeit that the simple global headquarter-subsidiary relation is replaced by a multi-level approach. In practice, up to five successive levels of transnational branching were identified, which arguably has the effect of ironing out national regulation mechanisms from the data. Taylor et al. (2002a), in turn, were able to circumvent state-centrism through the specific way in which they collected their data from the Internet. Advanced producer service firms, the focus of their data gathering effort, have benefited immensely from the technological advances in computing and communications in that this has allowed them to broaden the geographical distribution of their service provision. Locational strategy is an integral part of their public marketing and recruitment policies, and therefore quite transparent. Typically the web sites of such firms provide an option to select the addresses of the various offices, and this is exactly how information was gathered. In addition to information on a firm’s presence in a given city, all information that could aid in identifying the importance of the office to a firm’s network was written up. Taylor et al. (2002a) were thus able to estimate different layers of importance through a uniform scale of ‘service values’ vij ranging from 0 (non-presence) to 5 (global headquarters). As in Rozenblat and Pumain (2006), this multilevel approach has the likely effect of ironing out national idiosyncracies as long as large data sets are used.

The Lack of Network Data

The second empirical deficiency highlighted by Taylor (1997; further refined in Beaverstock et al. (2002) and Taylor (2004a)) concerns the lack of data on inter-city relations: although world cities are generally deemed to form a network, it is very difficult to sustain this empirically. The problem, as Alderson and Beckfield (2004: 812, emphasis in original) recently put it, is that researchers using attribute data must “assume what they set out to establish: cities are situated in a ‘system’, and some cities – as a result of the position they occupy in this system – are better situated than others.”

The fundamental contrast between attribute and relational data comes to the fore when comparing the data employed in the studies of Short et al. (1996) and Beaverstock et al. (2000b). Both papers are similar in that they combine different measures to obtain an overall picture of world city-formation (Table 1), but the way in which they structure their information differs fundamentally: Short et al. (1996) restrict themselves to various measures of what is contained within cities (e.g. number of corporate headquarters), while Beaverstock et al. (2000b) seek to construct indicators that give an idea of what flows between cities (e.g. the volume of skilled inter-city migration). This bifurcation runs right through the taxonomy presented in Table 1, although it should be emphasized that there has been upsurge of relational analyses in recent years (see, for instance, the evolution within the corporate organization approach from Godfrey and Zhou (1999) and Beaverstock et al. (1999) to Alderson and Beckfield (2004) and Derudder and Taylor (2005))5.

In the main, studies carried out within the infrastructure approach have had the easiest task in circumventing this deficiency: transportation, for instance, is basically all about connections and flows, which makes it relatively easy to construct relational matrices. As a consequence, all infrastructure-based studies listed in Table 1 have in one way or the other been able to make use of matrices detailing the volume of inter-city flows. The most interesting solutions to this deficiency, therefore, have been developed in the corporate organization approach: trying to acquire actual information on intra-firm flows (however conceived) is met by insurmountable problems, so that relational data essentially need to be created. Two different approaches have been proposed, whereby the basic difference between them can be traced back to the different types of firms they refer to (i.e. service firms versus multinational firms in general).

The first approach is Taylor’s (2001) construction of inter-city matrices out of a matrix containing ‘service values’ vij. These service values are, as we have seen, essentially standardized measures of the importance of a city i to the transnational network of a service firm j. The backbone of the proposed transformation can be traced back to Sassen’s (2001: xxi) observation that specialized service firms need to provide a global service which has meant a global network of affiliates or some other form of partnership. Taylor uses this observation to specify the world city network as a series of ‘inter-lock relations’ r ab,j between two cities a and b in terms of firm j. This relation can be computed based on the service values v a,j and v b,j of this firm j in both cities:

r ab,j = v a,j . v b,j (1)

The conjecture behind conceiving this product as a surrogate for actual flows of inter-firm information and knowledge is that the more important the office, the more connections there will be with other offices in a firm’s network (Taylor, 2001: 186). This approach seems reasonable when the following assumptions are made (see Derudder and Taylor, 2005: 72-73). First, offices generate more flows within a firm’s network than to other firms in their sector. This is inherently plausible in a context where protecting global brand image through providing seamless service is the norm. Second, the more important the office, the more flows are generated and these have a multiplicative effect on inter-city relations. The first part of this assumption is very plausible again. The second part reflects (i) the fact that larger offices with more practitioners have the capacity to create more potential dyads, and (ii) the hierarchical nature of office networks where larger offices have special functions like control and provision of specialised knowledge. The total connectivity ra of a city can be computed through aggregating these inter-city links rab,j across all firms and all cities in the dataset:

r a = ∑ jb r ab,j (2)

The second approach has been devised in the empirical research that makes use of information on large MNCs per se, i.e. the studies by Alderson and Beckfield (2004) and Rozenblat and Pumain (2006). The latter paper builds on a long – but in the Anglo-Saxon literature largely neglected – tradition of studies on transnational urban networks. Rozenblat and Pumain (2006: 131) “share Taylor’s view that there is a need for more studies (...) that draw on databases of inter-urban flows,” albeit that they propose to derive such flows from (multilevel) ownership relations in MNCs across space. In particular, inter-city relations are derived by interpreting a subsidiary owned by a multinational firm as “a directed interaction between the city where the headquarters are located and the city where the subsidiary is owned.” Cities are therefore essentially examined through their aggregated position in such ownership networks, in which the total (directed) relation between two cities rab is computed by aggregating all ownership links between firms headquartered in city a and their subsidiaries located in city b. Similar to Taylor’s (2001) approach, attribute data are transformed so that informed estimates of transnational inter-city relations become available.

Discussion: looking forward by looking back

It could be argued that this paper has not done full justice to the many empirical efforts undertaken in recent years: every study ultimately deserves to be assessed on its own terms. A number of concerns have nonetheless been common to the empirical literature as a whole, and it is exactly these concerns I have reviewed here. To this end, a very broad overview of the dominant empirical approaches was attempted. Using Taylor’s (1997) ‘double deficiency’ of data traditionally being neither transnational nor relational as a yardstick, I pointed to some major breakthroughs and lingering problems. As a whole, the empirical literature deserves some good marks, because it has come up with a number of original approaches and new data sources to work around Taylor’s ‘problem of evidence’. However, now that the empirical world cities literature has more or less dealt with the basic tents of its evidential crisis, it is time to raise some new points of attention.

A first concern is the disproportionate focus on a relatively few large metropolitan centres to concomitant neglect of all other cities. The most trenchant critique along these lines is by Robinson (2002: 536), who complains that “millions of people and hundreds of cities are dropped off the map of much research in urban studies.” This exclusion is from two ‘maps’: (i) the geographical map of world cities wherein most cities in the ‘South’ are missing; and (ii) the conceptual map of world cities which focuses on a narrow range of formal economic processes so that myriad other connections between cities are missing. However, all cities experience contemporary globalization processes, and these processes can therefore not be construed as affecting just a few privileged cities. However, subsequently Robinson (2005: 760) has conceded that recent empirical literature now focuses on “a much wider range of cities around the globe” thus lessening the exclusion from the map.

A second point of attention is the lack of theoretical nuance in the empirical research (Derudder, 2006). Consider the following conundrum: why do the likes of Taylor (2001) and Alderson and Beckfield (2004) bother with laborious data gatherings and fairly complex and assumption-rich transformations when a number of datasets (such as airline statistics) are within arm’s reach? This is the sort of question recently raised by Nordlund (2004) in his critique of the assumptions underlying Taylor’s (2001) network specification. The dataset presented in Derudder and Witlox (2005a), for instance, details the volume of transnational inter-city relations with unsurpassed detail, and is – compared to studies within the corporate organization approach – relatively easy to construct and interpret. The answer to this conundrum is that airline data “incorporate a large proportion of flows that are not part of the new city building processes which are our particular concern” (Taylor, 1997: 325). In other words: airline flows may very well be a prime example of transnational inter-city relations, but no amount of data manipulation can prevent that such statistics have a fairly limited conceptual relation to world city formation (see, for instance, Beaverstock et al., 2000a,b and Derudder and Witlox, 2005b). This relative conceptual weakness is quite different from other researches, which are quite specific in their conceptualization by taking their cue from the theoretical literature. Beaverstock et al. (2002), for instance, move well beyond general aerial flow patterns by embellishing Sassen’s (2001) theoretical research: they construct a proper conceptual model that focuses on four major attendants to the world city network (firms, sectors, cities and states), whose co-efficiencies enact network formation through two nexuses – ‘city-firm’ and ‘state-sector’ – and two communities – ‘cities in countries’ and ‘firms within sectors’. Put simply: refined airline analyses may well remain useful because they feature straightforward information on transnational inter-city relations, but quantitative approaches based on a carefully designed qualitative skeleton as in Beaverstock et al. (2002) take the literature a step forward through employing a more meaningful analytical framework.

Third, and on a more general level, it can be noted that the somewhat weak conceptual underpinnings of infrastructure-based studies point to a more fundamental problem, i.e. the failure to distinguish between the different conceptual backgrounds that exist within the theoretical research. Looking beyond the terminological buzz recently highlighted by Taylor and Lang (2004), one can distinguish between two main disciplinary traditions, i.e. (i) a geography/planning tradition (Christaller/Dickinson/Berry) treating cities as service centers and (ii) a sociology tradition (McKenzie/Hawley/Duncan) treating cities in terms of their metropolitan dominance (Taylor, 2006). The conceptualizations of the two most cited theoretical researchers (John Friedmann and Saskia Sassen) reflect this bifurcation. As a consequence, both conceptualizations are distinct, hence Sassen’s (2001: xxi) reminder that she is dealing with a distinctively different process. The crucial point here is that the different starting points of Sassen and Friedmann are reflected in diverging perspectives on the main features of globalized urbanization: a city’s prime function as node in transnational networks, the key agents in these networks, the spatial structure of the global capitalist system (and its alleged consequences for the structure of the global urban network), and the territorial basis of cities. This basic conceptual bifurcation, now, is equally reflected in the empirical research, more particularly in the two main sub-approaches in the corporate organization approach: Beaverstock et al. (2002b), Derudder et al. (2003) and Taylor (2004a) essentially employ Sassen’s conceptualization through their focus on intra-firm relations in global service firms, while Alderson and Beckfield (2004) employ Friedmann’s conceptualization through their focus on headquarter-subsidiary relations within MNCs. Both types of empirical studies, then, are not so much different solutions to the same problem, but rather parallel paths to understanding similar problems (see also Taylor, 2006). This distinction has not always been acknowledged (Derudder, 2006). Alderson and Beckfield (2004), for instance, (i) compare their results to Beaverstock et al. (1999) and Smith and Timberlake (2001), and (ii) frame their results against Sassen’s (2001) conceptual background. Despite the many merits of their empirical framework, such comparisons make little sense because they deal with different processes altogether. Now that the empirical research has outgrown some of its initial problems, the time has come to establish a stronger relation between specific concepts and the various sources of evidence at hand.



Alderson A S and Beckfield J 2004 Power and position in the world city system American Journal of Sociology 109 811-851

Beaverstock J V, Smith R G and Taylor P J 1999 A roster of world cities Cities 16(6) 445-458

Beaverstock J V, Smith R G and Taylor P J 2000a World City Network: A New Metageography? Annals of the Association of American Geographers 90(1) 123-134

Beaverstock J V, Smith R G, Taylor P J, Walker D R F and Lorimer H 2000b Globalization and world cities: some measurement methodologies Applied Geography 20(1) 43-63

Beaverstock J V, Doel M A, Hubbard P J, Taylor P J 2000 Attending to the world: competition/co-operation and co-efficiency in the world city network Global Networks 2(2) 111-132

Brenner N (1998) Global Cities, Glocal States: Global City Formation and State Territorial Restructuring in Contemporary Europe Review of International Political Economy 5(1) 1-37

Brenner N and Keil R 2006 The Global Cities Reader Routledge, London

Bruinsma F and Rietveld P 1993 Urban agglomerations in European infrastructure networks Urban Studies 30 919-934

Derudder B 2006 On conceptual confusion in empirical analyses of a transnational urban network Urban Studies 43(11) 2027-2046

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(9) 875-886

Derudder B and Taylor P J 2005 The cliquishness of world cities Global Networks 5(1) 71-91

Derudder B and Witlox F 2005a An appraisal of the use of airline data in assessments of the world city network Urban Studies 42(13) 2371-2388

Derudder B and Witlox F 2005b On the use of inadequate airline data in mappings of a global urban system Journal of Air Transport Management 11 231-237

Friedmann J 1986 The world city hypothesis Development and Change 17 69-83

Friedmann J 1995 Where we stand: a decade of world city research in Knox P L and Taylor P J eds World cities in a world-system Cambridge University Press, Cambridge

Godfrey B J and Zhou Y 1999 Ranking world cities: multinational corporations and the global urban hierarchy Urban Studies 20 268-281

Keeling D J 1995 Transportation and the world city paradigm in Knox P L and Taylor P J eds World cities in a world-system Cambridge University Press, Cambridge

Kunzmann K R 1998 World city regions in Europe: structural change and future challenges in Lo F-C and Yeung Y-M eds Globalization and the world of large cities United Nations University Press, Tokyo

Malecki E 2002 The economic geography of the internet’s infrastructure Economic Geography 78 399-424.

Nordlund C 2004 A critical comment on the Taylor approach for measuring world city interlock linkages Geographical Analysis 36(3) 290-296

Olds K and Yeung H W C 2004 Pathways to global city formation: a view from the developmental city-state of Singapore Review of International Political Economy 11(3) 489-521

Rimmer P J 1998 Transport and telecommunications among world cities in Lo F-C and Yeung Y-M eds Globalization and the world of large cities United Nations University Press, Tokyo

Robinson J 2002 Global and world cities: a view from off the map International Journal of Urban and Regional Research 26(3) 531-554

Robinson J 2005 Urban geography: world cities, or a world of cities Progress in Human Geography 29 757-65

Rossi E C and Taylor P J 2005 Banking Networks across Brazilian Cities: Interlocking Cities Within and Beyond Brazil Cities 22(5) 381-393

Rozenblat C and Pumain D 2006 Firm linkages, innovation and the evolution of urban systems in Taylor P J, Derudder B, Saey P and Witlox F eds Cities in globalization: practices, policies and theories Routledge, London

Rutherford J, Gillespie A and Richardson R 2004 The territoriality of pan-European telecommunications backbone networks Journal of Urban Technology 11(3) 1-34

Sassen S 2001 The global city: New York, London, Tokyo (2nd edition) Princeton University Press, Princeton

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

Smith D A and Timberlake M 1995a Cities in global matrices: toward mapping the world-system’s city-system in Knox P L and Taylor P J eds World cities in a world-system Cambridge University Press, Cambridge

Smith D A and Timberlake M 1995b Conceptualising and mapping the structure of the world system’s city system Urban Studies 32 287-302

Smith D A and Timberlake M 2001 World city networks and hierarchies, 1977-1997: an empirical analysis of global air travel links American Behavioral Scientist 44(10) 1656-1678

Smith D A and Timberlake M 2002 Hierarchies of dominance among world cities: a network approach in Sassen S ed Global networks, linked cities Routledge, London

Taylor P J 1997 Hierarchical tendencies amongst world cities: a global research proposal Cities 14(6) 323-332

Taylor P J 1999 So-called “world cities”: the evidential structure within a literature Environment and Planning A 31(11) 1901-1904

Taylor P J 2001 Specification of the world city network Geographical Analysis 33(2) 181-194

Taylor P J 2004a World city network: a global urban analysis Routledge, London

Taylor P J 2004b Reply to ‘a critical comment on the Taylor approach for measuring world city interlock linkages’ Geographical Analysis 36(3) 297-298

Taylor P J 2005 New political geographies: global civil society and global governance through world city networks Political Geography 24(6) 703-730

Taylor P J 2006 Parallel paths to understanding global inter-city relations American Journal of Sociology 112(3) 881-894

Taylor P J, Catalano G and Walker D R F 2002a Measurement of the world city network Urban Studies 39(13) 2367-2376

Taylor P J, Catalano G and Walker D R F 2002b Exploratory analysis of the world city network Urban Studies 39(13) 2377-2394.

Taylor P J, Walker D R F, Catalano G and Hoyler M 2002c Diversity and power in the world city network Cities 19(4) 231-241

Taylor P J and Lang R 2004 The shock of the new: 100 concepts describing recent urban change Environment and Planning A 36(6) 951-958

Townsend A M 2001 Network cities and the global structure of the internet American Behavioral Scientist 44 1697-1716

Zook M 2005 The Geography of the Internet Industry: Venture Capital, Dot-coms and Local Knowledge Blackwell, Oxford

Zook M 2006 The new old thing: e-commerce geographies after the dot-com boom in Daniels P, Leyshon A, Bradshaw M and Beaverstock J eds Geographies of the New Economy Routledge, London

Zook M and Brunn S 2006 From podes to antipodes: positionalities and global airline geographies Annals of the Association of American Geographers 96(3) 471-490



*Ben Derudder, Department of Geography, Ghent University, E-mail:

1. This observation is, of course, similar to the line of argumentation developed in Derudder and Witlox (2005a), where it is argued that the importance of airline hubs is overstated when standard airline statistics are used.

2. It is important to stress that not all studies listed in Table 1 can be designated as clear-cut empirical analyses of world city-formation, mainly because they encompass another scale of analysis. Rozenblat and Pumain, (2006), for instance, focus on the ‘European urban network’. However, it is clear from the outset that their empirical approach is deemed relevant in the context of world cities research.

3. Brenner (1998) and Olds and Yeung (2004), for instance, discuss the multifaceted links existing between world-city formation and state territorial development under conditions of contemporary globalization.

4. Admittedly, Smith and Timberlake (2001, 2002) were able to overcome the London–Hong Kong problem by estimating the importance of this link, while the relegation of US, Canadian, Brazilian and Japanese cities was dealt with through the use of additional data sources. While this allowed them to circumvent the most obvious gaps in the initial, this problem continues to affect major Chinese, Indian and Argentinean cities (among others).

5. This endorsement of relational data does not necessarily mean that attribute measures are totally redundant. The whole idea of spatial interaction modeling, for instance, is that relations between places can be conceived as the product of some measure of the size of these places. This means that attribute measures may in some instances be a good proxy for a city’s connectivity. E.g., data on the capacity of a city’s airport infrastructure allows for a reasonable estimate of its place in the overarching airline network, while an inventory of a city’s level of advanced producer services ultimately informs us on its position in these firms’ networks.

Table 1: A taxonomy of empirical approaches



Corporate organization



Other and/or a combination of indicators


Global service firms


Multinational enterprises


Air transportation




Beaverstock et al. (1999, 2000a)


Taylor et al. (2002b,c)


Derudder et al. (2003)


Derudder and Taylor (2005)

Godfrey and Zhou (1999)


Alderson and Beckfield (2004)


Rozenblat and Pumain (2006)

Townsend (2001a,b)


Malecki (2002)


Rutherford et al. (2004)


Zook (2005, 2006)


Smith and Timberlake (2001, 2002)


Derudder and Witlox (2005a)


Zook and Brunn (2006)


Short et al. (1996)


Beaverstock et al. (2000b)


Taylor (2005)


Edited and posted on the web on 9th May 2006; last update 10th March 2008

Note: This Research Bulletin has been published in Geography Compass, 2 (2), (2008), 559-574