GaWC Research Bulletin 35

GaWC logo
  Gateways into GaWC

This Research Bulletin has been published in F M Zimmermann and S Janschitz (eds) (2001) Regional Policies in Europe: Key Opportunities for Regions in the 21st Century Graz: Leykam, 49-64.

Please refer to the published version when quoting the paper.


Regionality within Globalization: What Does it Mean for Europe?

P.J. Taylor


"Cities have shaped the imagination and the life of Europeans"

Arnaldo Bagnasco and Patrick Le Gales


It has become part of our conventional wisdom that globalization is a threat and/or challenge to the twentieth century welfare tradition of politics based upon ethics of equity and social justice. There is, of course, much evidence to support this concern notably the increased economic polarisation occurring at all geographical scales over the last two or three decades. The question then becomes how can we defend our welfare states from the global market or, put another way, how can we prevent the harsh 'Anglo-American' form of capitalism taking over the world. The more sophisticated answers to this conundrum avoid simple dependency relations and bring together state-market with global-local as dynamic causal nexuses. Such arguments create political spaces through which to confront and/or modify virulent globalization. Though important, I am going to elide this train of thought and experiment with a different approach. This derives from an alternative focus within the multifarious concept of globalization. The common emphasis on global scale (supra-state) is complemented by an analysis of global connections (trans-state). Hence the nature of globalization is treated, not as simply magnitude or extent, but as linkage and flow.

My starting point is the premise that the most important theoretical and practical question confronting us in the new millennium is how spaces of places (spatial mosaics) interact with spaces of flows (spatial networks). From this perspective, and with the great benefit of hindsight, we can now identify 1957 as the most portentous year of the mid-twentieth century. Its dual spatial legacy is first, the supra-national European Union from the Treaty of Rome and its creation of a European Economic Community, and second the trans-national global financial system from the City of London invention of the Eurodollar market. Both are critical components of contemporary globalization but with very different geographies. The former is a coalition of states and hence is essentially territorialist in nature, the latter is a linking of markets and hence is essentially about connections. To build upon my starting premise requires going beyond the territorialist thinking which has dominated arguments for equity and social justice - usually and revealingly described 'welfare state' - as epitomised by the EU's Social Charter. Global network thinking has to be brought to the fore to provide a new perspective on how globalization impinges on such areas of theory and policy.

The space of flows I focus upon is the world city network. In this emphasis I follow a rich tradition of European urban research, epitomised by Bagnasco and Le Gales recent book of essays, but I add a particular concern for the global. Studying world cities is appropriate here for two specific reasons. First, the literature on world cities highlights a particularly deep economic internal polarisation consequent upon world city formation. This represents a very concrete manifestation of the challenge of globalization for traditional national welfare concerns. Second, Europe is a peculiar world region in its density of major metropolitan centres, each with world city potentials. Thus, this paper attempts a move from commonplace conventional state-centric analysis to an alternative city-centric analysis. This, however, is much easier said than done. The conventional approach has the whole weight of the embedded statism of social science underpinning it. Hence the first section below charts a route from territorial state to city network node as geographical entities for analysis. This concludes with a specification of the world city network which has important empirical/analysis and theoretical/policy implications. The second section focuses on analysis providing a review of some initial findings at both the global and European scales. The main feature is the interweaving of hierarchical and regional patterns within the network. The third section considers the implications of these findings for theoretical understanding suggesting both policy limitations and potentials. Under conditions of contemporary globalization there can be no 'Europe of cities' - a regression to territorialist thinking - but there can be indispensable European cities, or indeed leagues of such cities, with the power to set their political agendas. In a short conclusion this experiment in rethinking is put back into its broader context.


It is a truism to say that most cities are much older than the states that contain them. Nevertheless, our basic image of the world is one of states as represented by the familiar mosaic of the world political map. This is a 'metageography' which impinges upon discourses far beyond academic geography. Most social science is prone to excessive state-centrism. Certainly social scientists whose macro-studies are empirically dependent upon 'official data' - customarily called statistics although I prefer to add the hyphen ('state-istics') - typically produce comparative analyses of states. But this is more than a data problem. As a metageography the state world mosaic is part of our taken-for-granted world, it is, therefore, typically unexamined as it enters our social theory. Hence the fundamental disciplines of the social sciences - sociology, economics and political science - incorporate a largely unexamined discourse of spatial congruence with society (= national society), economy (= national economy) and polity (= nation-state) all presumed to exist within the same physical boundary. This is what I have called embedded statism and is the basis of our widespread tendency to territorialist thinking in the social sciences.

There are, of course, good reasons for the development of a state-centric social science. The political combination of nation and state in the nineteenth century created numerous important practices and institutions which did indeed operate to produce the spatial congruence which social scientists have assumed. But it was never the whole story as Wallerstein's world-systems analysis has attempted to show. However, it has been Castells and his identification of the rise of network society from the 1970s which has provided the critical challenge to territorialist thinking consequent upon embedded statism. The combination of computing and communication technologies has produced the enabling mechanisms for the massive growth of trans-national processes and the consequent production of contemporary globalization. Castells' thesis is that the new space of flows has now transcended the old space of places. From a world-systems perspective I think that globalization has revealed a space of flows which has always been there albeit recently greatly enhanced by the new technologies. Whatever temporal position we take on network society, it is incumbent upon us to rectify the inherent territorialist bias in social science in order to understand properly contemporary globalization.

From the 1970s it became commonplace to compare states with large firms. The latter were dubbed multinational corporations and, by comparing annual sales to GNP, many were found to rank economically above numerous countries. It is in the study of these firms that the space of flows has inevitably figured. But firms themselves do not constitute an alternative metageography to the states mosaic. Instead we need to focus upon the cities through which firm's operate. In Castells' formulation world cities are a critical component of his middle level - primarily the economic linkages which exist between infrastructure networks and cultural flows - in network society's space of flows. World cities, as originally conceived by Friedmann, are the control and command centres through which multinational corporations organise their global production. Subsequently Sassen has identified 'global cities' which carry out the traditional role of cities as service centres but at a global scale. However, there is a crucial difference beyond mere scale of activity. Under conditions of contemporary globalization corporate servicing has itself become a cutting edge activity in the world economy. Global cities are the sites of production of advanced producer services by firms which are themselves large corporations with their own global strategies. In what follows I combine Friedmann's global vision with Sassen's identification of a key mechanism in world city formation.

Adapting a phrase from the early work of Berry, I am interested in world cities as networks within a network of world cities. Most of the literature has explored the internal structure of world cities, the social networks or economic reflexivity or knowledge complexes which make world cities such distinctive places in the world economy. It is the operation of these processes which bring in their train the economic polarisation which also marks these cities. But this is to focus upon the nodes at the expense of the flows, on world city formation but not world city network formation. It is the latter, of course, which is crucial for using cities to explore the global space of flows. But to achieve this we need a precise specification of the world city network to allow its empirical description and analysis.

I specify a world city network as a tripartite structure with the world economy defining the network level, with cities as the nodal level, and with global service firms as the sub-nodal level. This defines an interlocking network where the sub-nodal level provides the connections which hold the system together. Global service firms each have their own location strategy which has involved setting up an office network across world cities. Connections between any pair of cities can be derived from the quantity and quality of service firm offices the two cities share in common. For instance, at the simplest level we can ask the question, if you go into the office of a service firm in city X, what is the probability that the firm will also have an office in city Y? Such information is directly relevant to service company's claims to provide a 'seamless service' whose quality can be assured through remaining 'in house'. Of course, such measures of connections between cities do not have to be limited to elementary presence data, sizes of offices, for instance, can be taken into account. The point is that by specifying the world city network in this way, inter-city matrices of connections can be created based upon the global location strategies of advanced producer service firms.

The world city network has not been previous formally specified in the literature. The usual approach has been to assume an urban hierarchy and then to rank cities by some relevant attribute. Therefore, this specification of an interlocking network has important fresh implications for both theory leading to policy and to analysis leading to findings. I consider both in turn, starting with the latter.


The analyses reported here are based upon inter-city matrices derived from the office networks of 46 global service firms in accountancy, advertising, banking/finance and law. All but the last analysis deal with 55 world cities identified in an earlier exercise. Hence the data matrix is 46 (firms) x 55 (cities) and the connection matrices are 55 (cities) x 55 (cities). The final analysis deals with 53 European cities including all of those in the world roster but also others which showed some 'evidence of world city formation' in the earlier study.

With such a large data set, the first need is to search for order in the data. This has been achieved by using principal components analysis. The first finding is that the data is quite requiring nine components to produce a reasonable description. This belies the idea that globalization has produced a homogeneous set of world cities. The results are summarised in Figure 1. The second point to make is that there is a reasonable degree of regionality in the pattern with all the key globalization arenas (northern America, western Europe, Pacific Asia, Latin America and eastern Europe) represented. However this analysis does not define a simple world regionalization as is most obviously illustrated by the locations of the European world cities in the diagram. There is a western European cluster which includes Brussels and Paris but it is not very encompassing of its arena's cities. Other major European cities, in particular, are to be found in a transnational service cluster which includes Zurich and Frankfurt alongside the likes of Tokyo and Hong Kong. London, on the other hand, is far away with New York in a small global city cluster. Even in the case of the few eastern European cities in the data, two clusters are found, centred on Moscow and Prague respectively. What does this complexity mean? The plausible assumption used here is that cities with like corporate service mixes will be more connected to each other. Hence the component-clusters of cities indicate sub-networks within the overall system of cities. The sub-network pattern that emerges is an interweaving of region and hierarchy.

The hierarchy implied in this data should not, however, be assumed to take a simple nested form. This is because under conditions of contemporary globalization hierarchical nested patterns are unnecessary. For instance, German world cities have no need to go through the leading German world city, Frankfurt, to reach London: a service firm office in, say Dusseldorf, can and will connect directly with its London office. Furthermore, with current global communication, cities no longer have discrete service hinterlands when it comes to providing advanced producer services. In our data all cities are connected to all other cities. Therefore instead of using the traditional hinterland concept with its separate city service regions, we have now to define the world-wide 'hinterworlds' for each city. Every city is in every other city's hinterworld. Thus hinterworlds are distinguished by the density of service offered across the world. These are found by asking the question, entering an office in city X, what level of service can you expect in city Y? For instance, from any office in Minneapolis you can expect a good service doing business in London, whereas from any office in London, chances of a good service doing business in Minneapolis is much less. This is because almost all global service firms with offices in Minneapolis will also have offices in London so that a seamless service can be provided but the same is not true visa versa. Such measures can be found for all pairs of cities. In Figure 2 the hinterworlds of Minneapolis and London are shown illustrating the least and most dense global patterns of service.

There are, of course, many hierarchical processes operating at the sub-nodal level within the firms themselves. This varies immensely between firms; there are hub and spoke organisation of offices, partnerships and franchises producing relatively autonomous sets of offices, and geographically specialised offices that operate as regional headquarters. The latter is of interest here and in fact turns out to have a relatively simple structure (Figure 3). In the firms which have offices with clear regional responsibilities only five world cities stand out: London (for Europe, Middle East, Africa), New York (Americas, North America), Hong Kong (east Asia), Singapore (south east Asia), and Miami (Latin America). Note that London (for Middle East and Africa) and Miami are extramural regional centres which is, of course, quite possible under the contemporary communication regime.

Finally at this global level, there is another way the connectivities can be simplified. Inverting measures of connectivity creates 'corporate service distances' (the more similar two cities are in their service mix, the shorter the 'distance'). Such an inter-city matrix can be subjected to a multidimensional scaling which generates best-fit spaces to the distances for different dimensions. With two dimensions an alternative 'world map' of cities is produced in corporate service space. Just such a map is illustrated for our 55 cities in Figure 4. This recreates the interweaving of regionality and hierarchy, the latter reflected in the overall centric structure. In this space, the three main globalization arenas are distinctly separate. In the case of Europe the western European region is clear, stretching out from London and Paris at the centre. Only Moscow and Istanbul lie below the horizontal axis as the most exceptional of European cities on the world stage.

Europe is clearly distinguishable within these global analyses but what patterns do we get if we just focus on just European cities? This can be illustrated by a principal components analysis which mirrors the global clustering exercise above. In this case only five components are required to describe the cities (Figure 5). Here the complexity of the global analysis is shed and a straightforward core-periphery ordering of cities is to be found. There are two 'spine clusters' one for major cities and another for minor cities which stretch north-south through Europe in a manner reminiscent of Rokkan's identification of a historical 'city-studded Europe'. This represents the intermingling of region and hierarchy so conspicuous in the global analysis. Outside the spine there are three peripheral clusters, an inner tripartite periphery plus an eastern (ex-COMECON) and a western (British Isles) periphery. One feature of this result is that London is the only British city not in the western periphery cluster: London is not very British. As we might expect, it appears as a major spine city with other leading European cities such as Frankfurt and Paris. But this is not the whole story. Put simply, London is not very well captured at all by this analysis which is shown by its relatively low communality. As well as London not being very British, it is not that European either. Well, of course, we know this from the global analysis and other studies: London is the archetypal global city. To analyse London within just a European context is to truncate the city. And given the previous discussion of hinterworlds this is true of all world cities, it is just so very clear for London with its great depth of service connections across the world. Truncating cities is a certain means for misunderstanding cities. Hence we have to be very careful how we interpret the spatial order of Europe's cities. It is, to be sure, an impressive indication of historical continuity through to contemporary globalization but this result does not an autonomous systematic structure.

There can only be one conclusion from these analyses: we must be very prudent in using the idea of a 'Europe of cities'.


Cities often feature in dystopias, worse-case scenarios for the future. From Petrella's concern for a lumpenplanet beyond favoured city-regions to Giddens' fear of dangerous instability in 'a world of a thousand city-states', cities seem to attract all the opprobrium usually targeted at globalization. These progenitors of civilisation are being cast in the role of barbarians. This is a pity, not just because cities are not going to disappear, but considering there is a growing consensus that they will be of increasing importance in the future. Of course, these concerns are to be expected in intellectual and political worlds imbued with territorialist thinking. By transcending such thinking can we find a more positive future role for cities?

The previous section has provided an empirical glimpse of cities within contemporary globalization. I call it a glimpse because, not only is the world city network only one small part of the global space of flows, even in terms of urban patterns, the processes I have described above depict just a part of a much more complex ordering of settlements - see, for instance Kunzmann's spatial differentiation of European world city regions into 17 areal types. Nevertheless, the network I have described can be viewed as a particularly important, even critical, framework within globalization. The world city network provides a spatial order for an otherwise nebulous globalization spatiality. Hence, although the specification and analysis above are far from complete, they do suggest some key alternative directions that network thinking points to.

The first point relates to the unusual tripartite network structure where the sub-nodal level is the active creator. Thus it is private firms who have produced the world city network. Theoretically, this is an important insight because it counters the tendency to reify cities as would likely occur with a specification of a simple two-level (i.e. no sub-nodal level) city network. Practically, this is critical for policy considerations beyond the obvious need for public-private accommodation which I deal with below under issues of governance. Highlighting the role of the private firms allows us to take the concept of network seriously. Unlike hierarchies which cities are usually assumed to form, a network implies a dominance of co-operation over competition. City competition may exist in the way local governments promote their areas, but in the tripartite network the main process is co-operation within firms through their offices across cities. For instance, with the coming of the euro and the location of the European Central Bank in Frankfurt there has been much speculation of the latter city threatening the European pre-eminence of London. However, from the perspective taken here, London versus Frankfurt is a minor process since most global service firms have locations in both cities and therefore have a vested interest in both cities being successful. Through their location strategies, firms have created a genuine network of cities - that is to say, cities need each other.

The political implication of this is clear: leading service firms in a city cannot be mobilised into a coalition of 'local capital' in competition with other cities. It is for this reason that the return to a sort of 'urban corporatism' is so problematic. This does not mean these firms are apolitical in any formal sense because they can be mobilised into a local city coalition in lobbying the state. In this context they can be part of 'local capital' because this practice is outside the network and in a different political arena. In fact, of course, current evidence shows world cities and states operate very closely together. It was the German government which obtained the Bank for Frankfurt at the negotiations in Maastricht but the best example is the UK's special treatment of the City of London in its 'national' economic policy-making as described in detail by Baker. This highlights the point that the city network should not be interpreted as a replacement for territorial states, rather they can be expected to complement each other as part of global economic and political restructuring. But this cannot hide their essential differences. Unlike European regions which can challenge states when they form cross-border alliances, the city challenge is not territorial. Cities are rediscovering the promise of networks.

With the nationalization of the state in the nineteenth century, cities were faced with the ever-present threat of truncation. It was not only the obvious problem of political truncation - St Peterburg and Vienna were the extreme examples after the demise of their respective empires - but there was also economic truncation with the rise of the anti-city policy of state protectionism. The 1930s represent the highpoint of the use of territorialist solutions to network failures. For instance, imperial preferences attempted to distort London's connections away from, say, Buenos Aires to Cape Town. Globalization has de-nationalised cities in important respects; no longer so truncated, they have transmuted into world cities. As such the EU cannot expect to operate policy through a 'Europe of cities' in the way it has developed a framework for a Europe of regions.

By being networked, world cities are presenting a fundamental challenge to state territoriality. Bringing more 'political space to cities', to use Bagnasco and Gales' phrase, means that this restructuring is not just a reallocation of power among different geographical scales, the structures of power themselves are being changed. Cities look certain to have an important role in future global governance. And this will not be a matter of 'global capital' dominating 'local capital' because in cities capital is both global and local. Hence, it is best thought of a network capital since that is where its fundamental spatial allegiance lies. Cities may be often be 'too frail' on their own to be major players but as an integral part of a critical network, all cities are potentially very powerful. They are special places which are needed to make the network work: without world cities global capital cannot be properly serviced and capital accumulation is seriously threatened. However, being so long under the domination of states, cities have yet to realise their new found power. New powerful mayors representing local government in alliance with network capital, and, through them, with other cities can produce a network governance regime. This may even take the form of leagues of cities actualising the power derived from wealth within sub-sections of the world city network based upon hierarchy and regionality. This is the logic of co-operation within the network.

But how is this scenario necessarily different from the dystopias with which I began this discussion? It can be reasonably objected that democratically local governments form only a part, perhaps just a minor part, in the governance. And the firms can be viewed as not having roots in a city, they just use cities as their anchors within the world economy. Once again such criticism betrays traditional territorialist thinking. The roots of the firms are in the network and the power of the local government derives from the city's position in the network. A further criticism is that the social context for building such political coalitions is impeded by the economic polarisation which typifies contemporary world cities. This is sometimes contrasted with the past political successes of the nation-state as integral state. The latter derived from a Gramscian hegemonic class dominance which is deemed highly unlikely in world cities with their potential for political conflict. However, this fails to appreciate that national hegemonies had to be constructed under very difficult circumstances - acute class conflicts - at least as foreboding of social cohesion as current world city social conflicts. Previously a new civicness was constructed from city to nation, it is not beyond the possibilities that another new civicness can be constructed from city to network.


This paper has been an experiment in some rethinking from a global network perspective. I make no apology for exhibiting an extreme city-centric bias in this work. I was being very purposive in this choice: I found the bias to be necessary for breaking out of conventional territorialist thinking. I believe the approach has succeeded in this initial aim but it has been much less fruitful in stimulating alternative ideas. There are no easy answers to evaluating how globalization is going to affect our lives as citizens in Europe. I hope that my discussion is viewed as more than the speculation which so often substitutes for real analysis and theoretical argument.

As my starting premise made clear, whatever might come out of prioritising a space of flows, spaces of places, most notably states, must be considered ultimately at the same time. Hence my argument has not been to relegate the role of states in defending peoples from an amoral world market, but to query whether defence can be our only strategy? I have tried to argue from a different beginning which may lead to a more positive engagement with globalization. I guess my bottom line is: why should the greedy bad guys have all the best geographies?

Figure 1: Principal component-clusters: regionality and hierarchy among world cities 


Figure 2: The urban hinterworlds of Minneapolis and London 


Figure 3: Regional world cities and their pan-regions 


Figure 4: Corporate service space: a new world map of cities 


Figure 5: Principal component clusters: the spatial order of European cities 


Edited and posted on the web on 6th October 2000

Note: This Research Bulletin has been published in F M Zimmermann and S Janschitz (eds) (2001) Regional Policies in Europe: Key Opportunities for Regions in the 21st Century Graz: Leykam, 49-64