This Research Bulletin has been published in Geography, 86 (1), (2001), 51-60.
In his aptly titled Urban World/Global City David Clark (1996) has recently attempted to update urban geography to address new circumstances of globalization. This textbook is built upon two contemporary features: first, the global settlement transition to an urban population majority (urban world), and second, the development of a world city network (global city). In this paper I focus upon the latter and I begin by arguing that Clark's laudable task was appreciably hampered by the way in which urban geography has developed in recent decades.
Urban geography was originally conceived as having two basic areas of study: internal patterns and external relations (see, for instance, Smailes, 1953). This division was carried forward into the 'quantitative revolution' as Burgess's concentric zone model and Christaller's central place theory. However, since the 1960s there has been a strong 'internalist turn' in urban geography. This is explicitly illustrated by Bassett and Short's (1989) charting of 'development and diversity in urban geography' in which external relations of cities are conspicuously absent. Even in Johnston's (1980) precocious text City and Society which features 'world cities' and 'quaternary' sector influences on city development, external relations feature in the historical chapters but are then omitted from the remaining analyses. Of course, these internalist biases did not treat cities as isolated entities, as Johnston's title indicates, cities were viewed as embedded in their particular societies. Hence the 'geography' was to be found within cities, with social theory, in its various guises, on the 'outside' as explanation. Little or no place here for geographies of external urban relations, for urban spheres of influence or relations between cities. This is a first prime obstacle Clark faced in producing an urban geography relevant to a globalising world.
In the broader realms of urban studies, however, the concepts of 'world city hypothesis' (Friedmann, 1986) and 'global city' (Sassen, 1991) have been developed which provide the theoretical raw materials for re-inserting external spatial relations into urban geography. But there is a critical problem with this literature: a severe case of empirical deficit (Short, et al., 1996). When it comes to relations between cities there is much rhetoric on world city hierarchies, networks and the like but very little evidence of their existence (Taylor, 1999). The paucity of data on relations between world cities is the Achilles heel of world city studies (Beaverstock, Smith and Taylor, 2000a; Taylor, Walker and Beaverstock, 2000) and this is the second prime obstacle for Clark's (1996) updating. In fact, his chapter on world cities is archetypal featuring rankings of cities by population and corporate headquarters but with no evidence of the 'space of flows' which the world city network constitutes (Cassells, 1996).
The purpose of this paper is to provide empirical evidence of the external relations of major cities under conditions of contemporary globalization. To do this I return to ideas of 'urban influence' in service provision which were current before the internalist turn. The first section below considers old concepts in the light of globalization and world city formation. It is found necessary to invent the concept of 'hinterworld' as the particular form of hinterland for world cities. In the second section, data collection for mapping urban hinterworlds is outlined and a means of measuring such worlds is described. The third section provides an empirical record of the hinterworlds of the leading ten world cities. Comparisons between these cities' external relations suggest particular elements of the geography of globalization. The main finding of the hinterworld comparisons is the dual dominance of London and New York which is considered in detail in the fourth section of the paper. This is not a surprising finding, of course, but it has never previously been shown geographically and mapped. In the concluding section implications of these empirical results are discussed: electronic communication does not herald the end of geography, it creates new and interesting global geographies.
URBAN INFLUENCE: HINTERLANDS AND HINTERWORLDS
Delimiting the region around a town or city for which it provides services was a major research topic in the urban geography of the 1950s and 1960s. Variously called sphere or zone of influence, urban field, tributary or catchment area, and umland or hinterland, I will follow Johnson (1967) and use the latter term. Information for defining hinterlands could be obtained from the centre (e.g. mapping the circulation of a city newspaper) or from outside (e.g. mapping the shopping trips of rural residents). Although the purpose was typically to determine discrete regions through drawing boundaries, in practice two general findings were commonplace: first, hinterlands were found to overlap (Johnson, 1967, 87), and second, within hinterlands there were variations in intensity of influence, typically a distance decay pattern from the centre (p. 89). Both of these findings are relevant to world city formation within globalization.
In the empirical research reported below I use Sassen's (1991; 1994) observation that it is the concentration of advanced producer services within major world cities which marks them out as new urban phenomenon within a globalising world. Advanced producer services are the necessary service inputs to enable clients to operate efficiently in the global economy. Including services such as banking/finance, advertising, insurance, commercial law, management consultancy and accountancy, the providers are themselves major corporations. Typically, advanced service firms have a network of offices across the leading cities of the world so that they are able to combine intensive local knowledge (e.g. law in a particular national jurisdiction) with their extensive general expertise (e.g. trans-jurisdictional use of English Law and/or New York State Law in commerce) to provide a 'seamless' service to their global clients. It is the widespread development of office networks of producer service firms across the world in the 1980s and 1990s which I refer to as world city network formation (Taylor, 2001a).
In world city network formation there are no boundaries between cities in terms of their hinterlands. With electronic communication underpinning their office networks (for law examples, see Beaverstock, Smith and Taylor, 1999b, 2000b), every city's hinterland overlaps with every other city's hinterland. For instance, a German firm doing business in Australia can work through the 'local' office of a global bank in Dusseldorf just as an Australian firm doing business in Germany can work through the 'local' office of a global bank in Melbourne. This is what globalization is all about but it does rather problematise the notion of hinterland. With the absence of boundaries, it seems to me that we need a slightly different concept here. I propose we refer to hinterworlds to indicate the geographical scale and nature of service provision by world cities. Thus a city's hinterworld is the global distribution of service connections which lay behind its world city formation1.
With no boundaries, the first empirical task in studying hinterworlds is to map the intensities of service provision by individual cities. I will follow the external approach to studying urban influence, measuring service provision from outside the centre. Just as with local hinterlands, designating hinterworlds in this way requires collecting data 'in the field' as it were, to show the service links.
WORLD CITY DATA: OFFICE NETWORKS OF GLOBAL SERVICE FIRMS
The data for this paper were produced for other projects and are explained in detail elsewhere, for details see Beaverstock, Smith and Taylor (1999a) and Taylor and Walker (2001).2 Here I provide a skeleton description to make this paper self-standing.
The first task is to define the world cities. A roster of world cities has been created by assessing the distribution of offices of 73 service firms across 262 cities. Defining cities as global service centres in four different sectors - accountancy, advertising, banking/finance and law - by the function, size and number of offices, 128 cities were identified as possible world cities. Results from these four sectors were combined to provide a scale from one to twelve measuring the 'world city-ness' of each of the 128 candidates (Beaverstock, Smith and Taylor,1999a). Fifty five cities scoring four and above were deemed world cities leaving the 73 cities scoring one, two or three as only 'cities showing some evidence of world city formation'. We deal only with this roster of 55 cities in the analyses below (see appendix A). Geographically, the world cities fall into three regional groups: 18 from the Americas (only 5 from Latin America), 23 from Europe (including Istanbul, and, for cartographic convenience (see below), Africa's only representative, Johannesburg), and 14 from Pacific Asia (including two from Australasia). Notice the uneven globalization represented by this world city formation; for instance, there is no such city between Istanbul and Bangkok in Asia. This geography is expressed schematically in Figure 1 which is used for presentation of results (hinterworlds) in the next two sections.3
In the original study (Beaverstock, Smith and Taylor, 1999a) the world cities were divided into three strata: ten alpha world cities scoring 10 to 12, ten beta world cities scoring 7 to 9, and 35 gamma world cities scoring 4 to 6. It is the hinterworlds of the alpha world cities - Chicago, Frankfurt, Hong Kong, London. Los Angeles, Milan, New York, Paris, Singapore, Tokyo - which we describe below.
For studying the office networks within world cities we have been more restrictive in the selection of firms. Forty six firms which have offices in at least 15 different cities are deemed to be global service firms (see appendix B). Maintaining an office in any large city is a major investment, possessing a network of such offices implies the development of a global servicing location strategy. For each global service firm their service provision in each of the 55 world cities is designated major, medium and minor (largely based on size and function), and scored 3, 2, 1 accordingly; where a firm has no office, zero is recorded (Taylor and Walker, 2001).
The end-result is a data matrix with dimensions of 55 cities by 46 firms and with cells ranging from 0 to 3 (Taylor and Walker, 2001). This provides the basic information for creating hinterworlds.
The urban hinterworld for a given city y is described in terms of the relative level of service it provides in each of the other cities. The measure of service provided by city y at city x is produced in four stages. (i) All service firms in city x are identified which we call xfirms. (ii) The sum of the highest scores of xfirms (not all firms have scores of 3) is produced. (iii) The sum of xfirms' scores in city y is computed. (iv) Expressing the sum from (iii) as a proportion of the sum from (ii) produces a measure of relative service ranging from 0 to 1. For instance, if the cities share no firms the sum from (iii) will be zero as will be the final measure. On the other hand if all the xfirms in city y have maximum scores, the sums from (ii) and (iii) will be identical and the service measure will be 1. The actual scores computed from our data range from 0.92 (London(y) servicing Dusseldorf(x)) to 0.12 (Minneapolis(y) servicing Singapore(x)).
Relative service scores have a simple common-sense interpretation. They are based upon the global firm's ideal of a seamless service, that is to say, provision remaining within the firm's office network. If a client walks into any service firm office in Dusseldorf to do business in London, the score of 0.92 records the relative level of service with that service firm he or she could expect to find in London. Conversely, a client in Singapore can expect a low level of service for work in Minneapolis, the score of 0.12 will reflect the fact that many firms with Singapore offices are not located in Minneapolis. Business may have to be conducted outside a firm's own office network - smaller law firms, for instance, often have arrangements with local firms in other cities - but the service is not seamless and it is difficult to ensure a high and consistent quality for valued clients. Of course, this measure favours large cities over smaller cities which is the nature of service provision at all scales. But it is not simply a matter of size, there is the matching of firms' office networks across cities and the different weightings of firms' provision across cities which produce a variety of outcomes encompassing more than proportionality with largeness. It is the geography of this variety which creates the distinct hinterworlds of particular cities; these divergences will include differences between cities of comparable size.4
THE HINTERWORLDS OF ALPHA WORLD CITIES
The distributions of intensity of service for the ten alpha cities are shown in Figure 2. Although all hinterworlds include service from all other world cities, only cities with scores above 0.5 are shown on the maps, an arbitrary cut-off point but one that broadly represents where a seamless, high level service is reasonably attainable.
As well as being quite distinct between the cities, the hinterworlds are also complex. However, the following findings are immediately apparent. First, all ten cities have service provision patterns with a reasonable coverage across the three geographical zones: there are no strictly regional alpha cities. Second, there are, nevertheless, differences between the cities in the relative intensities across the world: Chicago and Los Angeles have the least, London and New York the most. Third, of the three geographical zones, the Americas is generally the least well provisioned: this is particularly noticeable for USA cities. The last two points relate to a 'shadow effect' of New York on other US cities reported elsewhere (Beaverstock, Smith and Taylor, 2000a). Non-US firms tend to use New York as their gateway city to the US market resulting in their relative neglect of other US cities (see Beaverstock, Smith and Taylor (1999b) for London law firms behaving in this way).
Continuing the pairing of cities above, the hinterworlds will be described in turn as five pairs of cities with similar patterns of intensity.
1. Chicago and Los Angeles. Chicago has by far the least intensive of all alpha city hinterworlds (Figure 2(a)). Its only area of strengths are other US cities and western European cities. Other alpha cities are conspicuously low in this hinterworld. Chicago seems to be the major victim of the New York shadow effect (Beaverstock, Smith and Taylor, 2000a) and suspicions of its high level world city status (Abu-Lughod, 1995) seem to be justified on this evidence. Although having a more intense hinterworld, Los Angeles (Figure 2(b)) shares many of Chicago's characteristics notably the low levels for eastern European cities and even relatively low levels for Latin American cities. Los Angeles has higher intensity of services with other alpha cities but its main difference with Chicago is the greater coverage of Pacific Asia cities.
2. Frankfurt and Milan. These two western European cities have overall levels of intensities clearly higher than Los Angeles and with a very similar patterns. In both Frankfurt (Figure 2(c)) and Milan (Figure 2(d) the Americas are relatively low especially US cities. Both cities' hinterworlds are European-oriented. The basic difference between them is that other alpha cities are better represented in Frankfurt's hinterworld, reflecting its international finance role.
3. Hong Kong and Singapore. These two cities have slightly higher intensities in their hinterworlds than the previous pairs of cities and Hong Kong (Figure 2(e)) is the first city in this series of maps to have scores above 0.5 with all other cities; Singapore (Figure 2(f)) has three cities with scores below 0.5. As we might expect Pacific Asia is well serviced and, as previously, we can note the low scores for US cities but it is the high scores for European cities which is noteworthy here. The main difference between the two hinterworlds is Hong Kong's higher scores for other alpha cities; like Frankfurt, this will reflect, at least in part, Hong Kong's role in international finance.
4. Paris and Tokyo. Both cities have high intensity and broad coverage hinterworlds: neither Paris (Figure 2(g)) not Tokyo (Figure 2(h)) have particular regional biases except the commonly found lower levels for many US cities. This confirms both as 'global cities' although in the original use of this term Tokyo but not Paris was so designated (Sassen, 1991; see Keeling (1995) for other evidence suggesting Paris' elevation).
5. London and New York. The hinterworlds of these two cities are qualitatively different from those of the other eight alpha cities: the broad coverage of high intensities is very impressive (Figures 2(i) and 2(j)). These are undoubtedly the premier world cities in the world today. The distribution for London is slightly more dense; we explore differences between these two hinterworlds below.
These ten diagrams (Figure 2) represent the first mappings of hinterworlds. They illustrate clearly the world-wide nature of 'urban influence' under conditions of globalization. World city formation is an inter-city process: behind every successful world city there is a broad and intense hinterworld.
RELATIVE HINTERWORLDS: LONDON AND NEW YORK COMPARED
Within whichever other world city a client is doing business, they can be assured of a very high level of service if that business takes them to London or New York. But as we have seen, their two hinterworlds are by no means identical.
Differences between the two hinterworlds can be simply computed by taking scores for New York from those for London. This shows that for 37 of the 53 other world cities, London's level of service is higher. When these differences are mapped (Figure 3) there is some regionality in the results. Although there is by no means a simple geographical division of the world between these two leading cities, there is a tendency for US cities not to be spread proportionally between New York and London - only Miami appears in the set of cities with the higher differences in favour of London. Also there is a European bias towards London especially conspicuous in eastern Europe. This regionality relates, at least in part, to the roles of cities as pan-regional cities: as well as New York being the gateway to the USA, London operates as gateway to eastern Europe and Miami as gateway to Latin America (Taylor, 2000). However, the particular cities featuring in the bias towards New York by US cities implies a different sort of division between London and New York. Although predictably Washington, DC has the highest difference favouring New York, the other two cities in the highest New York category in Figure 3 are the two other US alpha cities. The other two cities in this category are also alpha cities, Hong Kong and Tokyo. With Frankfurt and Milan in the balanced category between the two cities it means that while London has a majority of all cities on its side of the balance, it has only two of the other eight alpha cities. Furthermore, if we go down to beta cities, five so designated (Brussels, Mexico City, Sydney, Toronto and Zurich) are in the balanced group between London and New York, which at 50% is proportionally high relative to all cities, again at London's expense.
The conclusion from this comparison is that whereas London provides a slightly higher level service on the whole, when it comes to the more important world cities New York is at least London's equal. The strata division of cities between London and New York looks to be more important than the geographical division.
CONCLUSION: NOT THE END OF GEOGRAPHY
I will conclude this paper first with a caveat and second, more importantly, by setting the findings briefly into a broader debate on globalization and geography.
The caveat is a straightforward one. Results are only as good as the data input and in this case we have drawn conclusions from the office location networks of 46 firms. This is certainly a large number in comparison with other studies of world cities but it remains, of course, only a small proportion of all global service firms. Certainly a different set of firms will produce different results; the question is how different?
The basis for choice in the data collection was data availability - firms were included where it was possible to find information on their whole office network. The assumption here is that using 46 firms covering four different producer service sectors will provide a reasonably robust set of results. Certainly the findings seem to be plausible in terms of other knowledge about the cities. It may be that some of the details are dependent on the specifics of the data which is why I have concentrated on the more general patterns. The least robust results are probably those pertaining to the London - New York comparison in the last section: these two cities have very similar hinterworlds and I have described relatively small differences. Nevertheless, like the other results, the findings from the comparison do seem to be feasible. The only test of this assertion is, of course, further study; replication of this search for hinterworlds.
Without replication, I remain confident about the broad findings of this research. At its most general, the study has shown that there are distinctive geographies to service provision in world city formation. Hinterworlds, despite their inherent overlapping, are like hinterlands in being specific to the centre under scrutiny. Globalization is not producing an even pattern of service provision, predictably supplied through an established hierarchy of world cities. Globalization has its geographies just as did previous eras, it is just that the geographies are new, and therefore intellectually challenging and exciting.
This very general finding is important because there has been suggestions about the end of geography (O'Brien, 1992). Certainly financial markets, in particular, are now based upon effectively instantaneous communications and analyses which means distance is no longer a factor in their operations. But the enabling technologies have not eliminated location decision making, rather there are now new combinations of spatial dispersion and concentration (Sassen, 1994): the international financial side of world cities is a prime example of concentration made possible by the new technologies. But this is equally true of all advanced producer services as they contribute to world city formation. What this paper has shown is that such centralisation itself produces further geographies. In the past the intensity of 'urban influence' was largely a matter of distance within a hinterland, such is no longer so for world cities but this does not mean there is not a geography of their influence. This is precisely what I have shown in my empirical study of hinterworlds. Not the end of geography, hinterworlds are part of a beginning of geography, a new geography composed of the geographies within globalization.
The data used in this paper derives from work on ESRC project R000222050. I thank Jon Beaverstock and Richard Smith for their contributions to that project. David Walker was instrumental in organising the data in the form used in this research and Ian Taylor computed the inter-city scores.
Abu-Lughod, J L (1995) Comparing Chicago, New York and Los Angeles: testing some world city hypotheses, in P N Knox and P J Taylor (eds.) World Cities in a World-System. Cambridge, UK: Cambridge University Press, 171-90
Bassett, K and Short, J (1989) Development and diversity in urban geography, in D Gregory and R Walford (eds.) Horizons in Human Geography. London: Macmillan, 175-93
Beaverstock, J.V. Smith, R.G. and P.J. Taylor (1999a) A roster of world cities, Cities 16, 445-58
Beaverstock, J V, Smith, R G and Taylor, P J (1999b) The long arm of the law: London's law firms in a globalising world-economy, Environment and Planning A 31, 1857-76
Beaverstock, J V, Smith, R G and Taylor, P J (2000a) World city network: a new meta-geography?, Annals, Association of American Geographers 90, 123-34
Beaverstock, J V, Smith, R G and Taylor, P J (2000b) Geographies of globalization: US law firms in world cities, Urban Geography 21, 95-120
Castells, M (1996) The Rise of the Network Society. Oxford: Blackwell
Clark, D (1996) Urban World/Global World. London: Routledge
Friedmann, J (1986) The world city hypothesis, Development and Change 17, 69-83
Johnson, J H (1967) Urban Geography. Oxford: Pergamon
Johnston, R J (1980) City and Society. An Outline for Urban Geography. London: Penguin
Keeling D J (1995) 'Transportation and the world city paradigm', in World Cities in a World System, eds. P L Knox & P J Taylor , (Cambridge University Press, Cambridge), 115-131.
O'Brien, R (1992) Global Financial Integration. the End of Geography. London: Pinter
Short J, Kim Y, Kuss M, and Wells H (1996) 'The dirty little secret of world cities research'. International Journal of Urban and Regional Research 20, 697-717.
Sassen, S (1991) The Global City. Princeton, NJ: Princeton University Press
Sassen, S (1994) Cities in a World Economy. Thousand Oaks, CA: Pine Forge
Smailes, A E (1953) The Geography of Towns. London: Hutchinson
Taylor (1999) So-called "world cities": the evidential structure within a literature, Environment and Planning A, 1901-4
Taylor, P. J. (2000) World cities and territorial states under conditions of contemporary globalization, Political Geography 19, 5-32
Taylor, P. J. (2001a) Specification of the world city network, Geographical Analysis (in press)
Taylor, P. J. (2001b) Visualizing a new metageography: explorations in world-city space, in Festschrift for Herman van der Wusten, ed. G. J. W. Dijkink (in press)
Taylor, P.J. and Walker, D.R.F. (2001) World cities: a first multivariate analysis of their service complexes, Urban Studies (in press)
Taylor, P J, Beaverstock, J V and Walker, D F R (2000) Introducing GaWC: researching world city network formation, in S Sassen (ed.) Cities and their Cross-Border Networks. Oxford: Blackwell Publishers (in press)
1. I do not wish to suggest that hinterworlds have replaced hinterlands; local newspapers continue to operate in their local catchment areas and rural people still shop in the local town. Globalization processes do not supplant existing processes, they add a further dimension to them: new practices operating alongside, and interacting with, old practices.
Note that papers in press in the bibliography can be found at this site as GaWC Research Bulletins.
3. For a discussion on how to present world city distributions as a new metageography, see Taylor (2000b)
4. Note that the methodology presented here does not strictly define a space of flows. Data on corporate service information flows between cities is obviously commercially sensitive and is not generally available. Here we make the plausible assumption that firms will have more intra-firm contacts than extra-firm contacts within their service sector: hence the more similar the complex of service firms in any two cities, the more service information flows will pass between them. For precise specification, see Taylor (2001a).
Figure 1: World cities in three regional blocs (for city codes, see appendix A).
Alpha world cities are shown in bold circles.
Figure 2: Hinterworlds of alpha world cities
(a) Chicago (b) Los Angeles
Figure 3: Differences between London's and New York's hinterworlds
APPENDIX A: WORLD CITIES
(with abbreviations for Figure 1)
APPENDIX B: GLOBAL SERVICE FIRMS
Abbott Mead Vickers (BBDO)
BANKING AND FINANCE
Allen & Overy
Edited and posted on the web on 25th November 1999; last update 5th July 2000