GaWC Research Bulletin 5

INTRODUCTION

Large and significant cities have fascinated social scientists over the last century and this is indicated by the range of terms used to describe them: imperial cities, primate cities, great industrial cities, millionaire cities, world cities, global capitalist cities, international financial centres, mega-cities and global cities are all well-known designations. This variety in terminology reflects both the diversity in the nature of cities and differences of approach to the study of cities. Although often closely entwined, there is a basic division of approaches which can be easily identified. There is a demographic tradition which is largely interested in the sizes of cities and a functional tradition which treats cities as part of a larger system. The former tradition is today represented by the mega-city project, which is exploring the human and ecological implications of contemporary and future huge population concentrations. The functional tradition is to be found in studies of world and global cities which are interpreted as integral to contemporary globalization processes. In this paper we will be concerned solely with the latter cities.

It should be noted that this conceptual distinction does not identify discrete classes of cities: New York and Mexico City, for instance, are both mega- and world cities. Despite such overlap - obviously there is a tendency for demographically large cities to be economically significant cities - the differences between the two approaches means that their respective rosters are distinct: Calcutta is a mega-city but not a world city, Zurich is a world city but not a mega-city. However, such discussion of rosters of cities is problematic in the case of world cities. Whereas mega-cities can be easily defined in terms of a given population threshold, which cities qualify for 'world' status has never been so clearly specified. Hence, while it is obvious that cities like London and New York are world cities, as we move to less significant cities such as Manchester and Minneapolis for example, there is by no means any consensus as to their status in this context. It is the purpose of this paper to construct a roster of world cities.

The usual way of treating these cities below the Londons and New Yorks is to cite them as national, regional or even 'sub-global' in their functional reach. This hierarchical approach is itself problematic since it relies on specification by city ranking rather than actual inter-city relations (Taylor, 1997). It is also somewhat doubtful given the pervasive nature of globalization. As a recent special issue of Urban Geography (Knox, 1996) has indicated, 'medium cities' have just as much need to respond to globalization trends as their larger neighbours. Hence, in this study we do not approach the definition of a roster in terms of working our way down a possible hierarchy - the latter is considered a separate, albeit closely related, research question. Rather, we consider the global capacity of cities in terms of selected services they provide. Using key advanced producer services, we consider firms which have a global competence and enumerate their presence in cities across the world. Global capacity is then defined empirically in terms of aggregate scores and interpreted theoretically as concentrations of expertise and knowledge. In this exercise, we find 55 world cities and another 68 cities showing evidence of world city formation. The paper is divided into two main sections. In the first we review the work of others in defining world cities. Within this functional tradition of studying major cities we identify four main approaches. However, the problem with this collection of approaches is the variety of criteria used; they range from being very specific to being quite subjective, and sometimes even vague, specifications of world city status. This exercise is useful for presenting the state of play in defining world cities but most of all, it illustrates clearly the need for a systematic consideration of the question of world city status. This is what we attempt in the second section. Using Saskia Sassen's (1991) argument that it is advanced producer services which are the distinctive feature of contemporary world city formation, we focus on four key services: accounting, advertising, banking and law. Cities are evaluated as global service centres in each of these sectors and aggregation of these results provides a measure of a city's global capacity or world city-ness. From these scores we define 10 'Alpha' world cities, 10 'Beta' world cities and 35 'Gamma' world cities. In the conclusion we briefly evaluate our results comparatively and for their utility in future research.

FUNCTIONAL APPROACHES TO DEFINING MAJOR CITIES

From the seminal work of Peter Hall (1966) to the comprehensive analyses of London, New York, Tokyo and Paris in the mid 1990s (Llewelyn-Davies (1996), or international financial centres towards the end of the millennium (The Economist, 1998), the central facet of the world city literature has been to rank cities according to their disproportionate geoeconomic power in the world-system. There has been broad consensus as to which cities are located at the top of the world city hierarchy (Table 1) but below this apex there is a wide range of opinion on which other cities qualify for world or international status (Table 2). Some of the variety exhibited in Table 2 relates to different criteria used in identifying world cities. Four major approaches have dominated the literature and before we consider the range of cities which have been cited as having world or international status we need to briefly review these different types of study.

Cosmopolitan Characteristics and the Multinational Corporate Economy

The first phase of work concerned the very early proponents of world city research who identified the strategic domination of certain world cities in the world-system by analysing and ranking the locational preferences and roles of multinational corporation (MNC) headquarters in the 'developed' world (Hall, 1966; Hymer, 1972; Heenan, 1977). Following the work of Patrick Geddes (1915), Hall's (1966) analysis of London, Paris, Randstad, Rhine-Ruhr, Moscow, New York and Tokyo, has been widely cited as the starting point for studying the global urban hierarchy. For Hall (1966), these cities were atop the urban hierarchy because of their (global) functional capabilities, with respect to power and influence in: politics; trade; communications; finance; education; culture and technology. Whilst Hall's (1966) work placed the concept of world city onto the agenda of contemporary urban studies, it did so under the auspices of urbanisation, or cosmopolitanism, rather than reflecting world city growth as an outcome of the uneven geographies of capital formation in the world system (see Brenner, 1998). In our eyes, the most significant, yet under reported, theorist of world city formation and the global urban hierarchy during this period, was the economist Stephen Hymer. For Hymer (1972), the top management corporate functions undertaken within MNC headquarters "must be located close to the capital market, the media, and the government ... because of the need for face-to-face contact at higher levels of decision making ... [and] ... applying this scheme to the world economy, one would expect to find the highest offices of the multinational corporations concentrated in the world's major cities" (as shown in Table 1). 

World Cities and The New International Division of Labour

Building upon Hall (1966) and especially Hymer (1972), this second approach centered upon the decision-making corporate activities and power of MNCs, in the context of the new (spatial) international division of labour discovered in the late 1970s (Frobel et al., 1980). This block of work, which includes Cohen (1981), Friedmann and Wolff (1982), Friedmann (1986), Glickman (1987), Feagin and Smith (1987), Godfrey and Zhou (1999) and to a lesser extend Knox (1995a-b) and Thrift (1989) for example, has not only enriched the 'theoretical' approach taken to world city studies, but has also been a major catalyst for the extension of research into the 1990s (see Knox and Taylor, 1995). Of this batch of writings, two major pieces stand out as taking empirical research forward in the development of world city rankings with the global urban hierarchy. First, we have the contribution made by Cohen (1981) who ranked the main locations of the largest 198 non-US corporations into a global hierarchy, with Tokyo and London at its apex (Table 1), with New York ranked alongside them as the three "predominant ... world centers of corporations and finance'" (page 308). Below them, he identified Osaka, the Rhine-Ruhr, Chicago, Paris, Frankfurt and Zurich as the so called "second-level" world cities. Second, and arguably the most important cornerstone of global urban hierarchy writing is Friedmann's world city hierarchy, which differentiated between primary and secondary cities in core and semi-peripheral countries. Friedmann (1986) based his hierarchy upon the idea of major cities as 'control centres' of capital in the new international division of labour. The world city hierarchy was based upon an analysis of several key criteria: major financial centre; headquarters for TNCs (including regional headquarters); international institutions; rapid growth of business services sector; important manufacturing centre; major transportation node; and population size. As Friedmann (1986, 72) admits "not all criteria were used in every case, but several criteria had to be satisfied before a city could be identified as a world city of a particular rank". Many have argued against this casual empiricism (for example, see Taylor, 1997), and Friedmann (1995) has readily acknowledged that constructing a stable urban global hierarchy is difficult but nevertheless he does present: a ranking of thirty world cities based upon their articulation with the global economy. In descending order, Friedmann's ranking of spatial articulations (world cities) identifies: Global financial articulations (e.g. London); Multinational articulations (e.g. Miami); Important national articulations (e.g. Paris); and Subnational/regional articulations (e.g. Osaka-Kobe).

The Internationalisation, Concentration and Intensity of Producer Services

In contrast to contextualising the formation of world city hierarchies within the new (spatial) international division of labour, a third approach has firmly associated the cities within the urban hierarchy with their propensity to engage with the internationalisation, concentration and intensity of producer services in the world economy. Here, in particular, we can identify the pioneering work of Saskia Sassen in such works as The Global City (1991), and (1994a) Cities in a World Economy. For Sassen (1991, 1994a-b), New York, London and Tokyo stand as the triad of global cities in the global economy because, "... these cities now function ... as highly concentrated command points in the organisation of the world economy ... as key locations for finance and for specialist service firms ... as sites of ... production of innovations ... and ... as markets for the products and innovations produced" (Sassen, 1991).

For Sassen (1991), the concept of the global city has emerged because of two interrelating factors: the globalization of economic activity, and the organisational structure of the producer service and finance industry itself - "rather than a detailed analysis of the economic base of the cities themselves" (Sassen, 1991). With respect to the globalization of economic activity, translated as being the shift to services and finance on a global scale, Sassen (1991) believes that these processes have brought "about a renewed importance of major cities as sites for certain types of production, servicing, marketing, and innovation." In particular, the internationalisation of both the producer service sector and financial system has made cities vital centres for the "management and coordination" of economic power in the global economy: particularly, New York, London and Tokyo. Paralleled with the globalization of economic activity, Sassen (1991) suggests that the rapid growth, specialisation and agglomeration of producer service firms and the organisation of the financial industry itself has to some extent been responsible for the formation of global cities. The locational preferences of producer service activities, like, for example, accountancy, advertising and banking, are then helpful in conceptualising the agglomeration and centralisation of management functions in global cities. As Sassen (1991) comments producer service firms "obtain agglomeration economies when they locate close to others that are sellers of key inputs or are necessary for joint production of certain service offerings."

In this case, the high concentration of producer service corporate functions in London, New York and Tokyo, can be explained by the array of potential customers that are found within these cities: the corporate headquarters of both manufacturing and other service firms, Government Departments, NGOs, and foreign firms.

World Cities as International Financial Centres

Finally, a fourth approach identifies major cities and their relative positions through rankings of international financial centres. The pioneering work here has been by Howard Reed (1981). Using a multivariate analysis of nine banking and financial variables ¹ and 41 related cultural, economic, geographical and political variables in 76 cities (80 in 1980) in 40 countries, between 1900-1980, Reed (1981) identified a taxonomy of financial centres with five hierarchical levels in 1980. In descending order, it read as: Supranational Financial Center (London); Supranational Financial Centers of the First Order (New York and Tokyo); Supranational Financial Centers of the Second Order (e. g. Amsterdam); International Financial Centers (29 cities); and Host International Financial Centres (39 cities). Recent analyses of the banking and financial prowess of international financial centres have continued to rank London, alongside New York and Tokyo, as supranational centres (see Table 1). Moreover, a series of articles in The Economist's (1998) special survey of financial centres, titled 'Capitals of Capitals' has reinforced the polarization that exists between the supranational centres of London, New York, Tokyo, Hong Kong, Paris, Frankfurt and Singapore, and the rest (including the likes of Chicago, Zurich and Sydney) (also see Sassen, 1999).

We can return now to Table 2 which lists 79 cities identified by 15 sources drawn from all four approaches we have just described. Of these cities, 25 are mentioned by just one source; these can be treated as idiosyncratic selections or the identification of highly specialist but limited functional centres such as Luxembourg and Panama City as international financial centres. This still leaves 54 cities only four of which get unanimous endorsement: London, New York, Paris and Tokyo, which, unsurprisingly, dominate Table 1. Hence there is much confusion below the very highest rankings of world cities. Our purpose is to add some clarification to this situation by creating a new roster of world cities in a theoretically-directed and empirically-transparent manner.

A ROSTER OF WORLD CITIES BY CORPORATE SERVICE CRITERIA

We are going to be very specific in our designation of world cities. Taking our cue from Sassen (1991, 126), we treat world cities as particular 'postindustrial production sites' where innovations in corporate services and finance have been integral to the recent restructuring of the world-economy now widely known as globalization. Services, both directly for consumers and for firms producing other goods for consumers, are common to all cities of course, what we are dealing with here are generally referred to as advanced producer services or corporate services. The key point is that many of these services are by no means so ubiquitous; for Sassen they provide a limited number of leading cities with 'a specific role in the current phase of the world economy' (p. 126). This is a consequence of general corporate growth and reorganisation over the last three decades which have resulted simultaneously in a dispersion of production and more routine office services, and a concentration of advanced services necessary to manage and organise the dispersion. With growth in the scale of these services, opportunities have increased for the rise of specialist corporations servicing other globalising firms. It is the former whose agglomeration needs in terms of information complexes are to be found located in selected cities - world cities - across the world.

Our purpose in this section, therefore, is to produce an inventory of world cities in terms of their provision of corporate services. The basic unit of analysis in this assessment is the service firm. The firms we deal with all have multi-city, multi-state locations and it is the geographical pattern of this organisation which is used to investigate world city formation. For a given city we are looking at the presence of major corporate service firms. By studying several firms over many cities, we are able to obtain the relative importance of a given city in terms of a particular service. By carrying out this exercise for several services, we can evaluate the overall 'world-cityness' of a given city. Thus, our procedures involve three stages. First, we find the global competence of service firms in terms of the geography of their presence in cities. Second, aggregating to cities we find the global service centres for a given sector. Third, aggregating service centre results by city we identify world cities of different degrees of overall corporate service provision.

Ideally, this study should deal with all global corporate service firms in all the advanced producer service sectors. Unfortunately we are not dealing here with data which is automatically in the public realm so that our analyses are dependent upon information availability. Through various investigative means involving trade directories, professional magazines, internal partner listings and world wide web searches, we have managed to put together data for four significant corporate services from which we can provide estimates of a city's global service capacity. The four services - accountancy, advertising, banking and law - are different enough to produce some interesting contrasts between geographical patterns, but the overall similarity of their world-wide locations give us confidence that aggregating the four sets of results produces a sound inventory of contemporary world cities in terms of their service capacities.

We have adhered to the following general principles in producing the service capacities of cities:

1. We have adopted an Adansonian approach to taxonomy (Sokal and Sneath, 1963). This entails two main guides: first, we use as much data as is both available and comparable; and second, we weigh the various data equally. This produces a relatively simple and transparent classification exercise.

2. In terms of recording a city presence we identify 'significant presences' only. Many major service firms, notably in banking and accountancy, have many hundreds of branches most of which will have no relevance to world city formation such as retail banking branches and local accountancy offices. In recording significant presence, we look for principle offices, or find alternative means for defining significance. As a rule of thumb so as not to omit relevant offices, we aim to record between 50 and 150 significant city presences per firm.

3. In converting individual firm presences to city capacities we ignore all cities which have only one significant presence in a given sector. All leading firms have particular histories of global growth which often create idiosyncratic location patterns, notably in their origin country or their initial region of expansion. By insisting on the significant presence of two firms in a city we are, by definition, omitting such particularity from this generic exercise.

4. We treat each sector on its own terms and do not attempt to impose a single set of classification rules. This is because each sector has a different history and pattern of global expansion. In addition, we have different types of data across sectors which would make a single set of rules impossible in any case.

5. While respecting these differences, we do, nonetheless, need to produce a classification of cities comparable across the four sectors. With this in mind, we produce three levels of service presence for each sector: prime global service centres, major global service centres and minor global service centres. It is these three categories which will be used to define the global capacities of cities and thus allow us to offer an inventory of world cities.

As indicated in the last section, the most interesting feature of our final inventory of world cities will be the identification of cities below the top levels. This is a grey zone where it is unclear whether or not we are dealing with world cities or some 'sub- level' of city. Our purpose in devising an inventory of world cities is not to simply replace the grey zone by a strict boundary. Rather, we will keep a 'fuzzy' lower 'frontier' to our list of cities by recording the global presence in cities below our chosen world city threshold. The need is not to eliminate the grey area but to investigate it and that is the spirit in which we have approached this classification exercise.

Finally, before we outline our specific methods and results, two points need to be emphasised. First, we have to consider the limitations of our data both in terms of range of service sectors covered, and numbers of firms dealt with in each sector. Different services and different firms would have undoubtedly have produced different results. Second, we have defined general principles of classification but the specific methods used in each sector are inevitable subjective; the data could be ordered in other ways. Our position on these matters is that the evidence of gross similarities within and across service sectors does suggest that our inventory is reasonably robust: differences are likely to be minor and on the margins. Nonetheless we avoid the temptation to 'over-interpret' our findings: all our data is reduced to just three levels of service provision - prime, major and minor.

Global Service Centres in Accountancy

Accountancy is the most highly concentrated service sector: at the time of our survey there were six major firms dominating the global provision. We have obtained suitable data for five of the 'Big Six' (the exception was Deloitte Touche Tohmatsu International)- Ernst & Young, Price Waterhouse, Arthur Andersen, Coopers & Lybrand, and KPMG. These are major organisations and each has many hundreds of offices across the world. For our purposes here, the main task is, therefore, to define significant presences. We have done this in two ways. For the first three listed above we have identified all cities in which the firm has two or more offices. For the latter two firms, we have more detailed information and we use this to identify all cities in which 20 or more partners work. These significant presences of the five accountancy firms identify 150 cities, 72 of which record just a single firm presence. We are left, therefore with 78 cities with services in accounting which meet our classification criteria. These are listed in Table 3.

Of the 78 cities, 13 have significant presences by all five firms and we designate them prime global accountancy service centres. There are, perhaps, two surprises in this list, Atlanta and Dusseldorf, which do not appear on other service prime centre lists. The 38 cities in which 3 or 4 firms have significant presences are designated major global accountancy service centres. The surprise here is the presence of five German cities, which together with two prime centres and another minor centre show global accounting to exhibit a particular concentration in cities of this one country. This is part of a wider pattern with Pacific Asian cities less prominent than in other sector tables below and with a relative over-representation of medium-sized European cities. The remaining 27 cities with only two firms having significant presences are minor global accountancy service centres. This list includes ten European cities which do not appear in any other sector table, for example - Arhus, Gothenburg, Leeds and Turin.

Global Service Centres in Advertising

Advertising is much less concentrated than accountancy and we use as our initial data here a list of the top 50 advertising firms/groups by annual turnover. The locations of the headquarters of these countries are used to define prime global centres. Three cities dominate this list, Tokyo, New York and London in that order, but another ten cities are represented. We have divided the latter group into two sets of five on the basis of the number and importance of the advertising firms located in the cities. Using a reverse rank (i.e. 1st = 50, 2nd = 49, etc.) sum for each city, we find a gap in the sums so that those above become prime centres and those below become major centres. In this way we define 8 prime global advertising service centres. The remaining major and minor centres were defined by branches of selected firms/groups. Suitable data was obtained for 8 firms/groups with significant presence defined by two offices in a city for the larger three and by a single office for the smaller five. Only London, already defined as a prime centre, scored eight in terms of these presences but 23 new cities scored between 5 and 7. These cities along with the five non-prime headquarter centres constitute 28 major global advertising centres. There are 31 cities which have from 2 to 4 presences and these are designated minor global advertising centres. All these 67 cities are shown on Table 4.

The surprise in the prime centre list is the occurrence of Minneapolis, Osaka and Seoul none of whom appear as prime centres in the other sectors. The latter two, coupled with the top ranking of Tokyo, suggests a cluster of longer established Pacific Asian world cities with a particular knowledge concentration in advertising. More generally, our lists show a tendency for firms to have presences in one city, usually the capital, to service a national market, e.g. Amsterdam, Athens, Tel Aviv and Warsaw. This means far fewer medium-sized European cities are represented here compared to accounting centres.

Global Service Centres in Banking

As we have seen in previous discussion above, it is in the finance sector studies where much work on ranking cities has been done. This is a sector, therefore, where we might expect rather less surprises in our analyses. Our purpose here is not just to replicate what has gone before but to produce a listing of centres comparable to the other sectors we are dealing with. For this task we have put together suitable data for 10 banks within the top 25 in the world by assets. For seven of the banks we use office locations to represent significant presence but for the top three, with their many offices, we have to define significance in other ways. For the top ranked bank, Standard Chartered, we have detailed data on branches which allows us to include only those with five or more fax numbers listed. For the second and third ranked banks, Credit Suisse and UBS, we have the problem of numerous branches in their origin country, Switzerland. We solve this by including branches outside Switzerland as significant presences but requiring two branches for a significant presence in a Swiss city. Eliminating cities with a single significant presence produces a total of 68 global banking service centres (Table 5).

We have defined a prime global banking service centre as a city with 8 or more significant presences and ten cities qualify. This list is very much as expected with the 3 leading cities of Pacific Asia, 5 cities in western Europe and 2 in the USA with San Francisco joining New York ahead of Chicago and Los Angeles. Thirty cities with 5 to 7 significant presences are designated major global banking centres and 28 with 2 to 4 significant presences are minor global banking service centres.

Global Legal Service Centres

Law firms are the latest producer service providers to become important players in globalization. The dearth, until recently, of major law firms trading across state boundaries is hardly surprising given that law is very territorial in its practice, principles and organisation. The result has been a quite unique pattern of globalization: an unprecedented UK/USA dominance resulting in part from the international commercial use of English Law and New York State Law. Spar (1997, 12) reports that for 1992 35 of the top 40 law firms were American and British (25 US, 10 London and the remaining 5 were from Australia and Canada). This requires us to define global law centres differently for UK/USA compared with the rest of the world. We define UK/USA cities as legal service centres on the basis of global law firm origins and the rest of the world in terms of the number of foreign branches they have. For this exercise we use data for the top 30 London law firms, all of which have foreign branches, and data for the 100 US law firms in the US top 250 law firms which have foreign branches. The number of cities either housing two or more global law firm headquarters or two or more foreign branches is 72 and they are listed in Table 6.

Eleven cities are designated prime global legal service centres on the basis of either having 7 or more global law firms centred there or having 26 or more foreign branches. The distinguishing feature of this service is the link to politics and therefore it is no surprise that Brussels and Washington, DC are among the prime centres. 28 cities with either 4 or 5 global law firms centred in them or with 5-17 foreign branches are defined as major legal service centres. Finally, 2 or 3 global law firm headquarters or 2 to 4 foreign branches qualifies 31 cities to be minor global legal service centres. The most distinctive feature of this distribution of service centres is the relative prominence of post-Communist eastern European cities which have attracted law firms as a result of their privatisations.

The GaWC Inventory of World Cities

Although intrinsically interesting in their own right, the purpose of the last four descriptions of global service centres was not to add to our understanding of these services per se, but rather to use their levels of service provision to create an inventory of world cities. We have called this listing the 'GaWC inventory' because it forms one of the basic research platforms for future research by the Globalization and World Cities (GaWC) Research Group and Network. We have created the inventory by aggregating the information from the four sector listings of centres. 122 cities are considered in all. A sum is produced for each city by scoring three for a prime centre, two for a major centre, and one for a minor centre. Given 4 sectors, the result is series of estimates of world city-ness ranging from 1 to 12. Cities and their scores are shown in Table 7.

The division of the 122 cities into different classes has been carried out using simple logical criteria. Any city scoring 10 or above must be a global service centre in all four sectors. Where it is a minor centre this would have to be compensated for by the other three being prime. In addition it must be prime in at least two sectors, and the other two would have to be major designations. Ten cities qualify as Alpha world cities. As expected, there are no surprises at this end of the distribution with 4 cities from western Europe, 3 from the USA, and 3 from Pacific Asia. Any city scoring 7 to 9 must be a global service centre for at least 3 of the four sectors and must be a prime or major centre in at least two sectors. Ten such cities qualify as Beta world cities. The same three world regions are represented as for prime world cities but with 'outer' cities appearing such as Sydney, Toronto, Mexico City and Moscow. In addition, a third world region appears: Sao Paulo from South America.

As mentioned previously, it is at the bottom end of the scale where uncertainty reigns for designating world cities. We have decided to define cities scoring 4 to 6 as Gamma world cities. All these cities must have be global service centres for at least two sectors and at least one of those must be a major service provision. This definition catches 35 further cities again distributed largely across the three main regions but with another 3 representatives of South America. Africa has its first city in our list, Johannesburg, but there are still no world cities found in south Asia or the Middle East (if we count Istanbul as European). The remaining 68 cities are designated as having evidence of world city formation processes but the evidence is not strong enough to really call them world cities. The interesting cities are those 12 scoring 3 and here we find one Middle East city, Tel Aviv, and two south Asian cities, Mumbai and New Delhi, which perhaps, signify the early stages of filling in the voids on the global world city map (see Figure 1).

CONCLUDING REMARKS

Let us begin these remarks by comparing the GaWC inventory with the previous studies which have identified world cities (Tables 1 and 2). The first point is that there are no major surprises in our listing. Certainly the cities which appear in Table 1 are found to be Alpha world cities in our analysis with only one or two exceptions. Furthermore there is an exact coincidence between the four cities identified by all sources in Table 2 and Alpha cities with maximum scores in our inventory (London, New York, Paris, Tokyo). The interest of our inventory lies in the lower levels of world city-ness, which provides us with a world geography of 'global' service centres. The regional concentration is quite remarkable: this is a concrete expression of what has been called 'uneven globalization'.

As we have noted in another paper (Beaverstock et al., 1999), world city formation has largely proceeded in three world regions, northern America, western Europe and Pacific Asia, which we have termed them the major 'globalization arenas'. Note that the Alpha world cities are relatively evenly distributed to these three regions (3,4,3 respectively) and 7 out of 10 Beta cities have a similar distribution (2, 3, 2 counting Sydney as part of Pacific Asia). The other three Beta world cities represent two minor globalization arenas, two in Latin America and one in eastern Europe (former COMECON). The rise of the latter arena is consequent upon the demise of the Cold War and the subsequent economic liberalisation especially privatisation of state assets which requires new advanced producer servicing. This arena counts for three of the four world cities we have identified which do not occur in the lists of others in Table 2; they are Budapest, Prague and Warsaw which we designate as Gamma world cities. The overall pattern of the 35 Gamma world cities repeats the geography of the higher levels: the three major globalization arenas account for 27 of these cities (8, 11, 8 respectively counting Istanbul as western Europe and Melbourne as Pacific Asia), and the two minor globalization arenas account for another 7 cities (4 in Latin America, 3 in eastern Europe). Geographically, the odd city out is Johannesburg, which might be considered to represent an outlier of European capital. Our inventory most certainly defines a very uneven globalization.

Of course, the basic advantage of the GaWC inventory is its grounding in a large quantity of comparative data thus giving identification of world cities a robust empirical basis. We hope it helps alleviate the confusion illustrated by the many differences among the lists in Table 2. However, while this is the first systematic, multi-sector assessment of cities to produce an inventory of contemporary world cities, we do appreciate that world cities are much more than service centres. Following Sassen, we have focused on the latter because the production of advanced producer services defines a distinguishing feature of cities in contemporary globalization, but there is room for other inventories emphasising other aspects of world city-ness. And there is one added advantage to creating an explicit inventory: it highlights a relatively large number of cities where world city formation processes are shown to be operating thus getting away from research concentration on just a few, allegedly paradigmatic, world cities such as London, New York and Tokyo (finance) and Los Angeles (cultural processes).

Finally, this inventory has been produced by us as a part of the GaWC programme of research looking at relations between world cities. Obviously, the starting point of wider research exercises is to define the roster of objects which are of concern. The GaWC inventory of world cities as presented here can be a starting point for new researchers interested in the hugely under-researched area of inter-city relations in the world-economy. We hope the publication of this roster of world cities stimulates such research.

ACKNOWLEDGEMENTS

We would like to thank the Economic and Social Research Council for funding this research project (R000222050).