This Research Bulletin has been published in
A Borsdorf and C Parnreiter (eds) (2003) International Research on Metropolises: Milestones and Frontiers. Wien: Verlag der Österreichischen Akademie der Wissenschaften, 29-42.
Part I THE NEED TO GENERATE DATA
1. Three Classic Evidential Blemishes
I begin by highlighting one small part of three very important books that discuss cities and globalization. Each selection consists of a quite surprising evidential blemish. These otherwise astute commentators on things global each come a cropper when they try and illustrate how world cities constitute a geography of globalization.
Two of the three evidential blemish examples use Friedmann's (1986) diagram of the 'world city hierarchy' and therefore a brief note on its veracity is appropriate here. First, the fact that Friedmann's model was immediately criticized for its paucity of supporting evidence (Korff, 1987) seems to have been generally overlooked due to the pedagogic utility of his world city hierarchy diagram. Second, according to Abu-Lughod (1989, 32), the origin of this mapping was 'a base map provided by Japanese Airlines'. No doubt this explains why the three cities with most connections in the original diagram are in the Pacific Rim (Tokyo, Singapore and Los Angeles) and not in the North Atlantic region (e.g. New York and London). That Friedmann's (1986) preliminary framework for world city studies should have had such longevity, as shown above, is clearly an indictment of the literature's lack of focus on measuring inter-city relations.
Conclusion: even in the best books, there seems to be an inherent problem with saying something soundly empirical about inter-city relations at the global scale (Taylor, 1999).
The concept of KBE envisions a link between knowledge 'creators' and 'users', the actors whose ingenious works restructure our contemporary 'global economy'. They comprise the quaternary sector of professionals, including R&D scientists and producers' service experts, and the quinary sector of decision-makers, who control and lead the globally linked KBE. The quaternary and the quinary sectors concur with Florida's (2004) three Ts: talent, namely the creative human capital reflected by the number of people engaged in creative occupations; technology, which measures the level of innovative activity associated with high tech concentration; and tolerance, reflecting the quality of a place as liberal and tolerant, one that is capable of attracting skilled people able to generate new ideas. Atop the creative class and the quaternary and the quinary sectors perch Beaverstock et al.'s (2004) global super-rich and Sklair's (1991; transnational capitalist class (TCC), a wealthy elite whose foremost objective is to secure the smooth functioning of the global economy.
2. The Culprit: Stat-istics
The common term for social data is 'statistics' a term that derives directly from the word state. This is, of course, no accident: large-scale data collection on human activities has its origins in state needs and continues to be dominated by states: hence my portrayal of it as state-istics.
Unlike the natural sciences, within social science there is little or no 'big science' where very large sums of money are committed to solving theoretical problems. The latter enables natural scientists to concentrate on developing measurements specifically designed for their theoretical purposes. In social science, most data that is collected relates to small-scale cumulative scientific activity. To get an evidential handle on big issues, researchers normally rely on the statistics that are available, that is to say, already collected. Collection is carried out usually by a state agency for the particular needs of government policy, not, of course, for social science research. But the problem is much more than the possibility of having to use unsuitable data. Basing 'big social science' on state-istics means that the state defines the basic dimensions of the leading edge 'macro' social research and therefore the framework within which most social research is conducted. This embedded statism within most very large-scale social data sets is a major reason why the information we want for describing inter-city relations is not available.Three characteristics of urban studies stem from embedded statism.
In conclusion: for the large-scale study of the inter-relations between world cities there is no alternative but to generate your own data.
Part II STUMBLING TOWARDS A SOLUTION
The Globalization and World Cities (GaWC) Study Group and Network was set up to contribute to solving the world city data problem. There are two strands to this work: qualitative studies that focus on a small number of cities to assess their relations (i.e. not to compare attributes of cities), and quantitative studies that attempt to measure the whole network (i.e. a global urban analysis). Here I focus on the latter. Normally this is presented as a logical progression from model to measurement but in practice it was an iterative, trial and error, process starting from quite modest beginnings. There are four main stages to reaching the point where we can measure the world city network.
In conclusion: these data form the basis of all our current work on the world city network and therefore the methodology will now be described in detail (from Taylor et al., 2002).
Part III THE BIG MEASUREMENT EXERCISE
The specification is clear on data needs: the starting point for world city network analysis is a matrix V where vij is the 'service value' provided by firm j in city i (Taylor, 2001, page 186). There are three stages. First, the process of gathering the appropriate information is described. The method employed is described as 'scavenging' since any information that can inform the data needs is recorded. Second, the conversion of this multifarious information into comparable data across firms is described. The data are produced by devising a uniform scale of service value that is then applied separately to the specific information gathered on each firm. Third, this data are used to derive specific measures of cities in the world city network. Measures of total service provision in cities and the global connectedness of cities are both computed.
The starting point is to find basic information on where major service firms are present in order to select those firms pursuing a global strategy. Using experience from previous experiments in this field, a firm is deemed to be pursuing a global locational strategy when it has offices in at least 15 different cities including one or more cities in each of the prime globalization arenas: northern America, western Europe and Pacific Asia. Having met this condition, selection of firms is quite pragmatic. Starting with rankings showing the top firms in different sectors, firms are selected on the basis of the availability of information on their office network. In addition, since one obvious research interest is comparison across different service sectors, firms are only included in the data in sectors for which at least ten firms can be identified. Using these criteria, 18 accountancy firms, 15 advertising firms, 23 banking/finance firms, 11 insurance firms, 16 law firms, and 17 management consultancy firms have been selected. These constitute the "GaWC 100", the global service firms at the heart of this research exercise
Although the starting point is firms, the information collected defines networks. Many global service firms exist as 'groups'. For instance, in accountancy there are alliances of medium-sized firms constituted as networks in order to compete globally with the very large firms that lead this sector. In other sectors, take-over activity has led to a corporate structure of core firm plus subsidiaries with the latter providing distinctive services as an additional dimension to the main service provision, for instance, as the investment arm of a mainstream bank. Sometimes the latter structure straddles the sector boundary such as banks owning insurance companies. Such firms are treated here as a single network and allocated to the core company's sector. Basically the networks are defined by the world-wide service contacts provided for clients on a firm's web site. Thus the GaWC 100 constitutes a large sample of global service networks.
A few of the larger firms have branches in many hundreds, even thousands, of cities and towns. The data collection has been restricted to the more important cities for two reasons. The first is analytical: the more cities included the more sparse the final matrix will become with nearly all the GaWC 100 networks not present in the smaller cities and towns. The second is theoretical: the interest is in the more important inter-city relations, ultimately the world city network. Nevertheless, it is also important not to omit any possible significant node in the world city network so that a relatively large number of cities need to be selected. Additionally, it is necessary to ensure that all continents are reasonably represented. The final selection of cities is based upon previous experiments and includes the capital cities of all but the smallest states plus numerous other cities of economic importance. The resulting set consists of 316 cities. It is these cities that are used in recording information on the global service networks of firms.
In selecting the cities to be included in the data collection the main concern has been to avoid excluding any city that may have important global service functions. Thus we have selected many more cities than we expect to use in subsequent detailed analysis of the data. The final selection of cities is based upon previous experiments and includes the capital cities of all but the smallest states plus numerous other cities of economic importance from across all continents. The resulting set consists of 316 cities. This is, of course, a very large number of cities and we are satisfied that it is a large enough selection to ensure no major omissions. It is these cities that are used in recording information on the global service networks of firms.
Selecting firms and cities is relatively straightforward, problems arise when attempts are made to gather information on the importance of a given city to a firm's global service provision. There is no simple, consistent set of information available across firms. The prime sources of information are web sites and everyone is different among the 100 firms. It is necessary to scavenge all possible relevant available information, firm by firm, from these sites supplemented by material from any other sources available such as annual reports. For each firm, two types of information have been gathered. First, information about the size of a firm's presence in a city is obtained. Ideally, information on the number of professional practitioners listed as working in the firm's office in a given city is needed. Such information is widely available for law firms but is relatively uncommon in other sectors. Here other information has to be used such as the number of offices the firm has in a city. Second, the extra-locational functions of a firm's office in a city are recorded. Headquarter functions are the obvious example but other features like subsidiary HQs and regional offices are recorded. Any information that informs these two features of a firm's presence in a city is collected in this scavenger method of information gathering. The end result is that for each of the 100 firms, information is available to create service values in each of 316 cities.
In conversion from information to data there is always a tension between keeping as much of the original material as possible and creating a credible ordering that accommodates all degrees of information across cases. In this exercise, there is very detailed information for some firms and much less for others. This tension is resolved here by devising a relatively simple scoring system to accommodate the multifarious information gathered. A six-point scale is used3 where two levels are automatically given: obviously zero is scored where there is no presence of a firm in a city, and 5 is scored for the city that houses a firm's headquarters. Hence decision making on scoring focuses upon allocating the middle four scores (1, 2, 3, and 4) to describe the service value of a firm in a city. This means that for each firm three boundary lines have to be specified: between 1 and 2, 2 and 3, and 3 and 4.
The basic strategy of allocation is to begin with the assumption that all cities with a non-HQ presence of a firm score 2. This score represents the 'normal' or typical' service level of the given firm in a city. To determine such normality requires inspection of the distribution of information across all cities for that firm. To alter this score there has to be a specific reason. For instance, a city where contact with its office is referred elsewhere will be scored 1 for that firm. In other firms where there is full information on numbers of practitioners, a city with an office showing very few (perhaps none) professional practitioners would also score 1. The point is that the boundary between 1 and 2 will differ across firms depending on information available. The same is true of the other boundaries. Generally, the boundary between 2 and 3 has been based upon size factors, and that between 3 and 4 on extra-territorial factors. For instance, exceptionally large offices with many practitioners will lead to a city scoring 3 while location of regional headquarters will lead to a city scoring 4. In practice, size and extra-territorial information have been mixed where possible in deciding on the boundaries for each firm. The end result is the service value matrix V, a 316 x 100 data array with vij ranging from 0 to 5.
How credible are these data? They are far from perfect largely dependent as they are on what information is available on web sites. But the key issue is the subjectivity inherent in the process of this data creation: the resulting data do not have the key property of inter-subjectivity. That is to say, two people using the same information will not always decide on the same boundaries. Given the nature of the information this is inevitable. One fundamental question arises. Does this issue lead to so much uncertainty in the data that the exercise is irredeemably flawed? There are two answers to counter this concern. First, the means of scoring has been designed to be as simple as possible, pivoting on '2 as normal' and with decision making limited to just three boundaries. Second, the exercise is carried out over a large number of firms so that particular differences will likely be ironed out in the aggregate analyses that the data are designed for. Thus we are satisfied that we have produced credible data for describing the world city network in 2000.
The total service values given in Table 2 measure the site service status of the cities (Taylor, 2001, 184, 186). This is a measure of the size of cities as service nodes in the world city network. The situational status of a city within the network is a relational measure defined as
Na = SS vaj . vij where a ¹ i
so that Na is the nodal connection of city a into the network defined as n cities, i, and m firms, j, with v as the service values in V. (This equation is a combination of equations 4, 5 and 6 in Taylor, 2001, page 187.) Given the range and scope of the data used here, this measure can be reasonably designated as the global connectivity of a city. The sum of all these city connectivities is 4,078,256 (equation 7 in Taylor, 2001, page 187). Individual city values can be expressed as a proportion of this grand total of interlocking connections (equation 8 in Taylor, 2001, page 188).
The top 10 cities ranked in terms of global connectivity are shown as gross and proportional measures as well as proportions of the highest city connectivity (i.e. London's) in Table 3. (The latter is the form of measurement we have found most convenient and is used in all our studies.) Not surprisingly this table is similar to Table 2 but it not exactly the same: Paris jumps ahead of Tokyo and Milan jumps ahead of Los Angeles while Frankfurt drops out to be replaced by Madrid. What this is indicating is that the important firms in the cities that rise in the ranking are relatively more connected than the equivalent firms in cities falling in the rankings; hence the greater global connectivity of, say, Paris over Tokyo. In terms of comparing the relative utilities of the site and situational measures, global connectivity is an aggregate relational measure and therefore is the preferred means of assessing the importance of cities in a network context. In addition, the situational status of cities is the more analytically interesting since it leads on to the creation of connectivity matrices and more sophisticated data analyses (see Taylor, 2001).
In conclusion: we can now do global urban analysis. The world city network is illustrated as a pattern of nodes in Figure 1. The cartogram includes all cities that have at least one fifth of the highest city connectivity (i.e London's) which includes 123 cities. Treating world cities as global service centers, this is the first time global connectivities of cities have been able to be shown: it is the fruit of much labour.
Postscript: the network model, the data generation methods, and the measurements of city connectivities can be used for other builders of networks across world cities. For instance, using data from media conglomerates brings Los Angeles to the fore while focusing on NGO networks highlights the importance of Geneva and Nairobi.
ABU-LUGHOD, J. (1989) Before European Hegemony: the World System, AD 1250-1350, New York: Oxford University Press
ALLEN, J. (1999) 'Cities of power and influence', in Allen et al. (eds) Unsettling Cities. London: Routledge
ALLEN, J., MASSEY, D. AND PRYKE, S. (eds) (1999) Unsettling Cities. London: Routledge
AMIN, A. AND GRAHAM, S. (1999) 'Cities of connection and disconnection', in Allen et al. (eds) Unsettling Cities. London: Routledge
BEAVERSTOCK. J. V., SMITH, R. G. AND TAYLOR, P. J. (1999) 'A roster of world cities', Cities, 16, 445-58
BEAVERSTOCK. J. V., SMITH, R. G. AND TAYLOR, P. J., WALKER, D.R.F. & LORIMER, H. (2000) 'Globalization and world cities: some measurement methodologies', Applied Geography, 20, 43-63
CASTELLS, M. (1996, 2001) The Rise of Network Society. Oxford: Blackwell
DICKEN, P. (1998) Global Shift. London: Paul Chapman
FRIEDMANN, J. (1986) 'The world city hypothesis', Development and Change, 17, 69-83
KORFF, R. (1987) 'The world city hypothesis: a critique', Development and Change, 17, 483-95
MICHELSON, R. L. AND WHEELER J. O. (1994) 'The flow of information in a global economy: the role of the American urban system in 1990', Annals of the Association of American Geographers, 84, 87-107.
SASSEN, S. (1991, 2001)) The Global City. Princeton, NJ: Princeton University Press
TAYLOR, P. J. (1997) 'Hierarchical tendencies amongst world cities: a global research proposal', Cities, 14, 323-32
TAYLOR, P. J. (1999) '"So-called world cities": the evidential structure within a literature', Environment and Planning A, 30, 1901-04
TAYLOR, P. J. (2001) 'Specification of the world city network', Geographical Analysis, 33, 181-94
TAYLOR, P. J. AND WALKER, D.R.F. (2001) 'World cities: a first multivariate analysis of their service complexes', Urban Studies, 38, 23-47
TAYLOR, P.J., CATALANO, G. AND WALKER, D.R.F. (2002) 'Measurement of the world city network', Urban Studies, 39, 2367-76
Table 1: Top 10 firms ranked by total service value across 316 cities
Table 2: Top 10 cities ranked by total service value across 100 firms
Table 3: Top 10 cities ranked by global connectivity
Edited and posted on the web on 20th January 2003
Note: This Research Bulletin has been published in A Borsdorf and C Parnreiter (eds) (2003) International Research on Metropolises: Milestones and Frontiers. Wien: Verlag der Österreichischen Akademie der Wissenschaften, 29-42