GaWC Research Bulletin 333

This chapter is written from a world-systems approach with a particular city-centric perspective (Taylor 2004). The latter derives from the Globalization and World Cities (GaWC) research programme that focuses on cities in globalization. The former foregrounds structural change, arguing that the modern world-system is in its demise phase so that our times are indeed special times as ‘new worlds' are necessarily being created. Which is where cities enter the story. The prime governance instrument of the modern world-system has been the inter-state system based upon mutually recognised sovereignties of territorial polities. It is possible that we are just beginning to experience an erosion of territorial sovereignties and their replacement by new mutualities expressed through city networks. This is what the rise of globalization as a contemporary, dominant ‘key word' might be heralding.

It is the purpose of this chapter to explore this possibility of structural change from a mosaic (states) spatial structure premised on boundaries to a spatial structure of nodes (cities) premised on flows. Of course, we cannot know if such a fundamental transition will occur for some decades in the future but we can measure current developments that show the rise of world city networks as an organizational skeleton of contemporary globalization. And that is my modest aim here. The argument is developed in six sections as follows:

The two empirical results sections are the core of the argument. It is hard to research at a global scale whilst eschewing states (and their ‘state-istics' in UN publications) and therefore much written on cities in globalization is evidence-light. In this chapter very many cities appear in tables showing measurements of activities in governance and business that are unique to GaWC researches.

Alternative forms of social organization: markets, hierarchies and networks

In political economy approaches to International Relations the usual framework is to compare and contrast markets and states. The problem with this is that networks tend to be neglected for not being central to understanding markets or states. In this chapter I follow the classic statement on social organization by Powell (1990) that identifies three basic forms of social organization: markets, hierarchies and networks. I also draw on Thompson (2003) who has embellished and extended Powell's original ideas.

Table 1: Three forms of social organization. Source: Derived from Powell (1990) and Thompson (2003).

Table 1 shows the basic distinguishing features of the three forms of social organization. Since I will be employing a network structure in the following section, in discussing Table 1 I will begin with the network column. Thus networks are premised on a horizontal social structure in contrast to hierarchies but they are not as de-centralized as markets. The social agents that operate to produce and reproduce the structure are essentially interdependent in networks, highly dependent in hierarchies, but wholly independent in markets. Their basic actions are reflected in these social relations: reciprocity based on trust for network makers, custom with rules for hierarchy makers, and contracts backed up by law for market makers. These actions produce distinctive social relations for networks in the form of cooperation, whereas both of the other social organizations have competitive relations, unequal for hierarchies and equal for markets. All these organizational attributes lead to the key element of each organization: mutuality is at the heart of all networks; bureaucratic logic solidifies hierarchies; and the price mechanism makes markets work. From these it follows that the antithesis of their respective organization is atomization for networks, anarchy for hierarchies, and monopolies for markets. In this manner Powell and Thompson provide us with a powerful yet simple taxonomy for understanding how social relations are ordered and maintained

The final row in Table 1 indicates the sphere of activity in which each of the social organizations has been mostly studied. Obviously economics is focused upon markets; somewhat less formally, hierarchies are studied in the realm of politics; and most networks studies are of social networks. But this does not necessarily follow from the former listed attributes. In practice, these social organizations have hybrid forms – for instance, networks with hierarchical tendencies – and therefore they should not be corralled into disciplinary homes. One way of breaking out of these disciplinary identifications of the three organizations is to view them concretely through their key geographical representations.

In Table 2 the spatial representations have been interpreted in terms of cities and states. Thus spatial networks come in two forms: business service networks of cities and diplomatic networks of states. In terms of hierarchies, cities are represented by that most famous of economic geography models, central place theory with its hierarchy of cities and towns; whereas within states, bureaucratic functional hierarchies in the centre are supplemented by a hierarchical spatial organization of territory into administration and services areas at different hierarchical levels. For markets, cities are considered to be key sources of externalities providing locational advantages to city-based firms, which contrasts with states' ‘comparative advantage' theories of international trade in which markets are territorialised as ‘national economies'. In GaWC research we have largely prioritized the top row (cities) in Table 2; in this chapter, GaWC research is reorientated to the final column to prioritize networks. This is not an argument that dispenses with markets and hierarchies but rather it is one that uses networks as its starting point. This can be justified on the basis of previous relative neglect of networks as an organizational form but there is a much stronger reason to turn to networks: contemporary globalization.

Globalization and network society

Networks may have been neglected relative to markets and hierarchies but they have most certainly not been ignored. Network power is recognised (e.g. Deibert 2000), infrastructural networks are studied (e.g. Zacher and Sutton 1996), and social networks have been found to be central to understanding contemporary global governance (e.g. Roseneau 1995). These are all important, especially the latter, and can be interpreted as symptoms of the rise of what Castells (1996) calls ‘network society' and more recently ‘global network society' (Castells 2009). His argument is that globalization marks a transition from industrial society to a new informational society. His work brings networks to the fore as the distinguishing feature of our contemporary world.

Castells' argument has a critical spatial dimension. The communication revolution has fundamentally changed the relationship between time and space. In previous societies social organization was largely facilitated by a spatial organization that enabled simultaneity: people coming together to interact through being in the same place at the same time (e.g. a weekly market place). Recent developments in communication (from the1970s) have enabled simultaneity to be virtually created worldwide without people having to physically come together. This is a new form of society: network society.

Global network society is premised upon ‘spaces of flows' superseding ‘spaces of places' as the basic spatial organization of society. By this he means that social power now resides primarily in informational flows rather than physical places. And the classic example of this transfer of power is to be found in new relations between cities and states. The modern interstate system with its inherent territoriality is a means of capturing power and concentrating it into particular places that are nation-states. This is a construction of a space of places that underpins international relations and is spatially reified in that most familiar of all maps, the world political map of states. It is this world of boundaries that is being transcended by network society as spaces of flows. However, this virtual world of information flows requires grounding in specific places – nodes - where necessary command and control functions, and innovation and development processes, occur. Castells (1996) draws on the seminal work of Sassen (1991) identifying global cities as these critical nodes through which spaces of flows are created and maintained. Sassen recognised that the communication revolution enabled enormous dispersion of economic production but, importantly, that this in turn generated a need for new concentration in creative management and organizational functions. Thus her global cities are great clusters of network makers. It is these network makers in the headquarter offices of global corporations, and in the professional, creative and financial service firms, that convert the virtual global potential of network society into real global business practices. It is this economic globalization that is seen as severely undermining the fundamental role of the nation-state as manager of its own ‘national economy' (e.g. Held et al 1999). Or in Castells' terminology, a new global space of flows is replacing an old national space of places.

GaWC research builds on the insights of Castells and Sassen but without necessarily accepting all their prognoses. With its empirical focus, GaWC has attempted to measure the spatial mechanisms and processes that Castells and Sassen describe.

City networks and how to measure them

In order to convert the key ideas of Sassen and Castells on cities into a model whose features can be measured, it is necessary to specify precisely what I have called the world city network (Taylor 2001). The purpose of the model specification is to guide data collection from which meaningful network measures can be derived.

The world city network is an interlocking network. This is an unusual arrangement of flows in that it has three network levels instead of the usual two. In most networks there is simply the net level (e.g. a neighbourhood gang) and the nodal level (e.g. the individual gang members as nodes), with latter identified as the agents (network makers) and the former being the outcome (network). In an interlocking network there is a third sub-nodal level. Thus in the world city network, the cities are the nodes and the net level is the inter-city relations at the global scale, the network itself. At the additional sub-nodal level there are firms that are the network makers – following Sassen these are advanced producer service firms (e.g. advertising, commercial law) that provide the necessary information (expertise, knowledge) to enable the global economy to function. Note, therefore, that unlike most networks, the nodes (cities) are not the agents, network making takes place in the cities but it is not city authority that creates the world city network, this is done by the businesses that use the world city network: the firms produce and reproduce the network through their day-to-day work practices (e.g. in planning and implementing a global advertising campaign or in writing a contract recognised across several jurisdictions).

It follows that the world city network can be specified as follows (from Taylor 2001). There are m advanced producer firms located in n cities worldwide. A service value matrix is defined as m (firms) x n (cities) filled by service values v_ij which indicate the importance of firm j's office in city i.; the larger v ij, the more work carried out in city i by firm j. This means that a large city office has the potential for generating large work-based information flows. Potential intra-firm work flows between offices in different cities a and b is given by

For every city there are n-1 such links to all other cities across all m firms. Thus the total potential intra-firm work flows to city a is

This is called the network connectivity of the city. It is a basic network measure that defines the degree of integration of a city into the world city network. Results showing this measure will be reported in the next section.

This specification guides the data collection to operationalize the model. It is necessary to collect information on the office networks of firms so as to grade offices in different cities into variable levels of importance (Taylor et al 2002). Such information can be retrieved from firm's websites but the information comes in many different forms that have to be codified to make them comparable. In the work on advanced producer services, for every firm cities were coded from 0 to 5; 0 representing no presence of a firm in a city and 5 indicating the city housing the firm's headquarters; all other offices are coded between 1 and 4 based upon evidence of their importance. The result is a service values matrix with cells having values varying from 0 to 5.

This specification was derived to model a particular business world city network as described previously. But business service firms are not the only network makers that operate through cities. There are numerous networks of cities created by non-business agents that have taken advantage of the communications revolution. Business service values can be generalised to office activity values and the above model specification and data collection still applies. Below I report interlocking network analyses for three examples of non-business network makers working through cities that are relevant to understanding contemporary global governance:

These three world city networks will be compared to the original business services world city network below.

Table 3: Parameters of four world city networks. Source: Taylor et al (2002); Taylor (2005b).

Table 3 shows the details of the data collections for four world city networks introduced above. Notice the size of the evidential basis upon which results below are reported: there are over 60,000 individual pieces of information collected in the form of network activities across cities. As noted above these are codified variables that range from 0 to 5 for business firms. Other network activities were coded slightly differently due to their particular characteristics as follows. For diplomatic networks the capital city locations of foreign affairs departments are scored 5 and all other offices (embassies, consulates, missions, tourist and trade bureaus, etc.) are scored 1. For UN agencies, headquarter cities were scored 3, regional offices scored 2, and others scored 1 (field, branch, area, and national offices, advisory team locations, representatives, etc.). NGO offices were scaled by their degree of responsibility: international offices scored 4, regional offices 3, national offices 3, and sub-state offices 1. In all cases no presence in a city is recorded as zero.

These data were all collected between 2000 and 2004; for further details see Taylor et al (2002) and Taylor (2005a, 2005b). Thus all the results reported below are for networks generated at the beginning of the twenty-first century.

Measuring Networks I: City Connectivities

In this section I focus on the network connectivities of cities that can be computed from the four activity matrices. To assist comparability, these results are reported as proportions of the city with the highest connectivity. The latter is the city that is most integrated into the particular network, remaining cities have varying degrees of lower network integration. This provides for a ranking of cities for each network: the top 25 cities in each network are ranked in Table 4. These show quite different global urban geographies that will be described in turn.

Table 4 : Leading cities in worldwide activity networks (2000-03). Source: GaWC data sets (www.lboro.ac.uk/gawc).

That diplomatic networks centre on Washington will come as no surprise but the ordering of cities in this city network is not a simple list of capital cities mirroring the importance of their respective nation-states. Although Tokyo ranks second in line with Japan's economy, Germany with the third ranked economy has its capital city down in 7^th position. Otherwise there is a general bias towards Europe where capital cities of relatively small countries appear with more diplomatic connectivity than capitals of much larger third world countries (e.g. compare Vienna with Brasilia). There are few non-capitals: New York is the main case and obviously features as the locale of the UN Assembly but it is still placed behind Ottawa. The other main transnational capital, the EU's Brussels, is integrated into this network at a lower level than the capital cities of the EU's four major states. Note also the peculiar case of Hong Kong, not a capital but a traditional city gateway, which ranks alongside its capital city Beijing. However, overall these results clearly confirm this to be primarily an international network. The fact that this listing of cities shows the shallowest gradient (Brasilia ranked 25^th has a much higher connectivity than cities ranked 25^th in the other three networks) also reflects the formal equality in the interstate system consequent upon mutual sovereignties. But note the absence of African cities from this list.

The UN network is perhaps more surprising in the connectivity rankings. Although New York houses the UN Assembly it is not at the centre of the UN system of institutions: Geneva is by far the most integrated city in this network. Furthermore this leading city dominates its network much more than for the other networks since its second ranked city, Brussels, has less than two-thirds of Geneva's connectivity compared to second-ranked cities in the other networks all having over 90% of their leading cities' connectivity. This confirms the top down nature of this networking with seemingly strong hierarchical tendencies. However, another feature of the top-down process is that the UN network is the most ‘democratic' with a relatively even spread of cities featuring in the top 25: six cities each from Asia, Europe and Africa, five from Latin America and two from the USA.

The NGO network is again quite distinctive. In this case only Washington remains as a US representative and although the continental distribution is similar to the UN network (8 Asian cities, 6 African, 6 European and 4 Latin American) membership is subtly different. With London as the most integrated city we find more sub-Saharan African cities from ex-British colonies. Thus whereas Addis Ababa is Africa's ‘UN city', Nairobi is well known as Africa's ‘capital for NGOs'. The other difference to the UN network is that the NGO network listing of cities shows a much shallower gradient of falling connectivities: San Jose's connectivity of 0.480 would almost make the top ten in the UN network. This gradient is second only to the diplomatic network and, as with the latter, implies a relatively egalitarian network of cities.

The cities featuring in these three governance networks are very different from those listed for the business services network. Dominated by a duopoly of London and New York (NYLON), this network, not surprisingly is hugely biased towards the richer parts of the world: 21 out of 25 top cities (8 from Europe, 5 from USA, 5 from Pacific Asia and 3 from Canada and Australia). Of the remaining four cities, there is Mumbai and three from Latin America; thus most of Asia beyond the Pacific and all of Africa simply do not feature at the top of this city list. Not surprisingly this network is the most distinctive and this will be a topic of discussion below.

Measuring networks II: Sub-net hub cities

World city networks are very complex structures and this is only hinted at in listing city connectivities. None of the four networks we are dealing with consist of simple inter-city arrangements centred on the important cities that head the lists in Table 4. Rather there are uneven patterns of inter-city connections that indicate distinctive sub-nets within the overall network. The way in which GaWC research goes about searching for these sub-nets is to carry out a principal components analysis on the office activity matrices.

Principal components analysis is a member of the factor analysis group of multivariate statistics (Rummel 1970). These are data reduction techniques that are used to simplify large matrices into smaller more manageable matrices. Principal components analysis is the simplest and most popular of these techniques. It converts m variables into k components where k is much smaller than m. This data reduction is shown in Table 5 for the activity matrices used in this paper. Thus the 114 network makers (state foreign offices) of diplomatic networks can be reduced to 12 components that are newly constructed, distinctive, composite variables. Thus the original 114 x 170 cities office activity matrix is reduced to a 12 x 170 cities components matrix that accounts for 36% of the original variation in the data. These 12 components are distinctive patches of dense interactions that I interpret as sub-nets within the overall network. Note that in this case, most (nearly two thirds) of the original variation in the data is not included in the new matrix: this indicates a relatively unstructured overall network with a large unpredictable element. The other three city networks are represented by sub-nets accounting for over half the original variation in their respective matrices.

Table 5: Data reductions. Source: Taylor (2004 and 2005b). The results are derived from a principal components analysis using varimax rotation.

For the purposes of this chapter I will report just the particularly strong sub-nets that were produced by the four principal components analyses. These are identified by the level of dominance of the leading city (like the headquarter city in the original network making variables) within each sub-net component. I will call these the hub cities in the sub-nets. Dominance is measured by the largest city score recorded for each sub-net in the reduced components matrix; I treat hub scores above five as indicating strongly centralized sub-nets. 21 such centralized sub-nets were identified in the analyses but they are very unevenly distributed across the four analyses (Table 6). The contrast is between NGO and business services linkages with the former consisting of ten strong sub-nets and the latter only one. This indicates a more partitioned NGO network of world cities, with its multiple important hubs, compared to the ‘smoother' business services network of world cities. In fact, the cities' scores for the one business services network hub has two of the three smallest hub scores in Table 6. The other two networks have five strong sub-nets each and therefore also depart from the business services networks' more smooth patterning of links. This is potentially an important geographical finding: the economic linkages are more generally global with less distinctive patches of intensive interactions than are found in the non-economic linkages. In other words, economic and political globalizations show different fundamental network structures. This will be considered further in the discussion section below.

Table 6 : Strong sub-nets in world city networks and their hub cities. Source: Taylor (2004 and 2005b). The results are derived from a principal components analysis using varimax rotation.

Turning to the particulars in Table 6, the hub cities identified by no means replicate the previous connectivity results (Table 4); instead they complicate our understanding on the networks to varying degrees. The simplest case is for business services: the one hub here is for legal services provided by firms centred in the USA. Law firms generally are the least dispersed of advanced producer services and this is especially the case for American firms – hence this one strong sub-net for business services. For the three non-economic networks I will describe them in the order of their partitioning. Which means I start with the ‘extreme' case of the NGO city network with its multiple dense patches of interactions and their hub cities. Here the broad geography of the high connectivity cities (Table 4) is curtailed: the three leading globalization regions - northern America, western Europe and Pacific Asia – have three hub cities each, leaving Africa with just one and Latin America and Asia beyond the Pacific with no hub cities at all. And Nairobi, Africa's one hub city, is reduced from its second highest rank in connectivity (Table 4) to the second lowest hub score in the principal components analyses. What are being picked up here are the power-relations in the NGO network. Cities beyond the main globalization regions may have plenty of NGO presence but the main offices, directing the network, are not to be found here. For example, many NGOs are present in Harare and Accra and the cities are therefore ranked 9^th and 12^th in connectivity (Table 4) but they are parts of sub-nets directed from elsewhere, Geneva, Brussels and London. Thus the NGO network is not just partitioned; it mirrors traditional core-periphery power structures in the modern world-system. The other two political networks show a similar pattern but with no hub cities beyond the main globalization regions (i.e. there is no Nairobi exception to prove the rule). The UN system shows a distinctive functional partition, for instance Vienna and energy policy and Washington and financial institutions. The diplomatic networks relate to three US strategic patterns and two Pacific Asian new hub cities; the interesting feature here is the dearth of European cities – Paris and Brussels appearing as adjuncts of Washington.

In summary: moving from general network connectivity analyses to specific sub-net hub city identifications has revealed that although the non-economic networks appear to be more open to participation of network making away from the core regions of the world-system, the principal components analyses show these political networks to be less progressive than first appears. The non-economic networks are much more partitioned than the businesses services network of cities so that power relations through hub cities indicates a core-periphery differentiation at least as severe as the economic. In these early years of globalization as network society, this can have a profound meaning for global governance.

Discussion: cities in spatial incongruence

This chapter is overtly empirical but purposive with regard to understanding spatial dilemmas of global governance. What do our spatial findings contribute to this vital topic? The first point to make is that globalization is more than an ‘up-scaling' of economic and political activities: the ‘global scale' is not simply ‘national scale' writ large. The larger scale activities are enabled through a change in the nature of the social space that is being created through globalization. This contemporary emphasis on spaces of flows – Castells' ‘network society' – has been shown to be operating in both economic and political spheres in the analyses above. It has been shown that there are diplomatic, UN, NGO and business service networks that operate through cities across continents – this is global network society in practice.

This practice is entirely different from the modern space of places organization that is the inter-state system. This was premised on a spatial congruence of processes within state boundaries. That is to say, modern governance has been corralling political, economic, cultural and social processes into ‘nation-states' to create national economies, national cultures and national societies administered by national governments. These basic units of governance aspired to control flows at their boundaries. Beyond the boundaries states negotiated flows in international relations. This modern governance has been all about accommodating flows to places. The processes that constitute contemporary globalization are challenging this spatial logic in that places are being accommodated to flows. The latter is a much more complex organization.

The spatial congruence of nation-states provided a very tidy solution to issues of governance: differentiated processes were neatly packaged into more or less manageable units. As has been pointed out many times, this contrasted with pre-modern societies where differentiated social processes generated differentiated social spaces: for instance, political spaces of places in non-national states contrasted with economic spaces of flows through city networks. But globalization is not a post-modern throwback to ‘medieval' organization; to be sure there is a sense in which the economic has ‘escaped' the bounds of states but this new ‘mismatch' is associated with a reversal of power from the political to the economic, something completely alien to ‘medieval' organization. What is required is network governance, an organization that international relations is finding inherently very hard to deal with. Multi-level, multi-agency, multi-sector – it is becoming so complex; this is what the results above are providing a glimpse of. Organization through capital cities in separate states is being superseded by the multiple hub cities of overlapping sub-nets in global networks. Spatial incongruence is the big challenge of global governance.

Conclusion: travails of global governance in system transition

This chapter is overtly empirical but purposive with regard to understanding transition dilemmas of global governance. What do our organizational findings contribute to this fundamental topic? Once again the empirics informs the theoretical issues. According to Wallerstein (2004) the modern world-system has entered its demise phase and we are at the beginning of a transition to a new system that will, perforce, be based upon a different social logic to ceaseless capital accumulation. But there is a conundrum in identifying this transition: since incessant change is the hallmark of the current system, how can we distinguish between such ‘ordinary' rapid social change that reproduces the system from ‘extraordinary' rapid social change that is undermining the system? It is not the quantity of change that is relevant but its nature. Thus in the analyses above, the diplomatic networks represent the international political network, the UN network is a supranational political network, and the NGOs are a trans-national political network. Evidently it is the international network that is reproducing modern politics whereas the supranational and transnational network both have potential for undermining this established politics. It is interesting to note, therefore, that it is the international network that shows least hierarchical tendencies compared to the other two political networks, especially the ‘bottom-up' NGO network In other words, the changes in political network structure seem to be producing a more hierarchical political world than that bequeathed by the Treaties of Westphalia. But all political networks are more hierarchical than the commercial world city network based upon business services. The rise of the latter implies less hierarchical global inter-city relations compared to modern international relations.

But there is one feature of the modern world-system that contemporary globalization in all its forms does not seem to be transcending. Wallerstein (2004) describes the modern world-system as a space-economy based upon a core-periphery division of labour that is reflected in the distribution of power in the inter-state system. In the analyses above core-periphery patterns emerge: globalization and its global spaces of flows does not seem to be lessening global inequalities. Quite the opposite has been happening. The business services inter-city relations might be showing relatively less hierarchical tendencies in their network but the rise of this structure coincides precisely with an increasingly polarized global economy. Core-periphery geographical patterns are shifting in global network society but the basic material inequalities are not being radically changed. If economic globalization in network society portends a future alternative system, it would seem that modernity's replacement would be a more polarized world.

There have been strong international pressures in most global governance institutions to support lessening global polarization both politically and materially. They are patently failing and the rise of network society has not helped … thus far. But network society does provide a new potential for supporting a progressive transition. Current global governance networks might be relatively hierarchical and business service networks relatively polarizing, and yet inter-city relations based upon cooperation (mutuality) contrast starkly with inter-state relations based upon competition (realism) and therein lies a progressive future for global governance. The twentieth century's development planning and development aid based upon spaces of sovereign places have been a disaster. Twenty-first century city-network planning and connectivity aid based upon spaces of flows between cities is a global governance agenda worthy of creation.

REFERENCES

Deibert, R J (2000) ‘Network power', in R Stubbs and G R D Underhill (eds) Political Economy and the Changing Global Order, Oxford: Oxford University Press, pp. 198-207

Held, D, McGrew A, Goldblatt, D and Perraton, J (1999) Global Transformations Cambridge: Polity

Powell, W W 1990 “Neither market nor hierarchy: network forms of organization” Research in Organizational Behavior 12, 295-336

Roseneau, J N (1995) ‘Governance and democracy in a globalizing world' in D Archibugi, D Held, and M Kohler (eds) Re-imaging Political Community. Cambridge: Polity

Rummel, R J (1970) Applied Factor Analysis. Evanston IL: Northwestern University Press

Taylor, P J (2001) ‘Specification of the world city network', Geographical Analysis 33, 181-94

Taylor, P J (2004) World City Network: A Global Urban Analysis. London: Routledge

Taylor, P J (2005a) ‘Leading world cities: empirical evaluations of urban nodes in multiple networks', Urban Studies 42, 1593-1608

Taylor, P J (2005b) ‘New political geographies: global civil society and global governance through world city networks', Political Geography 24, 703-30

Taylor, P J, Catalano, G and Walker, D R F (2002) ‘Measurement of the world city network', Urban Studies 39, 2367-76

Thompson J (2003) Thompson, G F 2003 Between Hierarchies and Markets: The Logic of Network Forms of Organization (Oxford University Press, Oxford)

Wallerstein I (2004) World-Systems Analysis: An Introduction. Durham, NC: Duke University Press

Zacher, M W with Sutton, B A (1996) Governing Global Networks. Cambridge: Cambridge University Press

NOTE

^* Peter J. Taylor, Department of Geography, Loughborough University, UK; email: p.j.taylor@lboro.ac.uk

Note: This Research Bulletin has been published in M. Amen, N.J. Toly, P.L. McCarney and K. Segbers (eds) (2011) Cities and Global Governance: New Sites for International Relations Farnham, UK: Ashgate, pp. 201-216.