The World-System’s City System: A Research Agenda

Jeffrey Kentor, University of Utah

David Smith, University of California, Irvine

Michael Timberlake, University of Utah

Introduction

“World cities” (Hall, 1966; Friedmann and Wolf, 1982) and “global cities” (Sassen, 1991) have increasingly attracted the attention of urban-focused social science research since Peter Hall introduced the idea in the mid-1960s. Social scientists working on comparative social change are now concerned with situating these cities conceptually and empirically within the broad currents of the world political economy (e.g., Smith, 1996; Timberlake, 1985). A more recent development among scholars of cities, urbanization, and development is to view city networks as constituting an important structural dimension of the world system. From this perspective, the great cities of the world are organizational nodes in multiple global networks of economic, social, demographic, and information flows. This relational view allows us to begin to think about mapping cities in terms of their structural relationships with one another. This, in turn, suggests a research agenda the objectives of which range from describing the structure of a world network of cities, to identifying and explaining hierarchical relations among world cities, to understanding the “nesting” of the world city network into the broader world-system, to analyzing the connections between particular cities’ places in the global hierarchy and social relations within them.

Leading scholars focusing on world cities contend that economic power inheres in a few key “global cities”, where the world economy’s key functions, such as financial and other producer services (Sassen, 1991), are concentrated. The top cities are followed in the hypothesized hierarchy of world cities by less influential sub-global cities which, nevertheless, are said to “articulate” among large regions of the world economy. The picture emerging from this body of scholarship is that of a hierarchical world-system of cities (see Knox and Taylor, 1995), and though this hierarchy is subject to change, the consensus is that the particular cities at the top of the global hierarchy have changed little in recent history. But this rich and evocative line of scholarship tends to fall short empirically: it rarely is based on actual analysis of data on the relationships undergirding the global network of cities. We need to develop much better indicators of actual links and flows between these great cities in order to evaluate some of this perspective’s most important assumptions, and to develop more accurate descriptions of world city system structure and changes therein. Such a project will provide an alternative strategy for evaluating theories of globalization, one based not on a system of nation states alone, but one defined by examining the contours of a world-wide system of cities. Paralleling the scholarship comparing world cities is another body of urban research focusing on coalitions of actors within particular cities who are compelled by their land-based interests to push “their” cities into competition for more prominent roles in this global hierarchy (Logan and Molotch 1987; Rondinelli, Johnson and Kasarda 1998; Scott 2001). This suggests a promising way to link the “global” and the “local”. In fact, local political actors are increasingly consciously using the language of globalization to justify putting public resources into making their cities more competitive globally (e.g., Saito and Thornley, 2003). With information on the theoretically relevant attributes of each city in the hypothesized network, we will be able to develop and test propositions about how variations in local economic and social relations are related to global network relations. At the same time, there are strong theoretical reasons to look for “upward links” between this hypothesized global city system and other global networks within which this system is assumed to have been produced: the set of global relations in which nation states are the chief constituent parts, thereby evaluating some of the claims about “denationalization” and the “deterritorialization” of the state that are made in the globalization literature.

1. Objectives and Significance

The objectives of the proposed research are to test the implications of theoretical developments about globalization and “world cities” by refining and analyzing three relevant data sets: (1) data on world city network relations from 1980 to the present, (2) an international trade model of the world-system for the same time period, and (3) a database on the theoretically significant attributes of those world cities that are included in the network database. We plan to conduct descriptive and explanatory analyses of these data, and to make these data available for use by other scholars of world cities and comparative social change. The focus of the proposed research is on relations among the world’s large cities and it will involve collecting information on interlinkages between pairs of cities in a hypothesized world network of great cities. The research will also develop measures of theoretically important attributes of each city in the network. We will perform the initial analyses of these data with the aim of providing both a description of the world city network for specific points in time and a description of how it changes over time (based on formal network analysis of these data). We will investigate the way the global city network is nested and articulated into the relational world-system of nations, and offer explanatory analyses of the relationships among the changing structure of the network, other global socioeconomic processes, and theoretically significant attributes of the cities (based on combining the results of the formal network analysis with more conventional variable-based analytic procedures, such as OLS regression).

The proposed research reflects our longstanding goal to understand the global dimensions of urbanization and urban change. This project focuses on understanding large scale social change in light of global processes and structures. It also involves examining the articulation between global structural change and social structures and processes manifest at the local level. In many ways the research can be seen as an extension of a project that began in one co-PI’s dissertation exploring the cross-national relationship among urbanization, international dependence, urban labor force structure, and social quality (Timberlake, 1979; Evans and Timberlake, 1980). Collaboratively, we further developed this line of work in the early 1980s under the auspices of Christopher Chase-Dunn’s NSF-supported research aimed at framing urbanization processes within an international political economy perspective. This project resulted in a number of theoretical and empirical scientific publications exploring the relationships, on one hand, among urbanization patterns, global urban hierarchy, and the growth of particular cities and, on the other hand, among urbanization patterns, national-level socioeconomic and political trends, and international dependency and world-system relations (e.g., Timberlake and Kentor, 1983; Chase-Dunn, 1985; Kentor, 1982, 1985; Timberlake, 1985; Nemeth and Smith, 1985a). Much of this research viewed urbanization and urban hierarchy partially as outcomes of national level processes, such as industrialization, and global processes, such as foreign investment dependence. Later, the long-term project took a new turn. Instead of viewing cities and urbanization as derivative, we began to view cities as crucial basing-points in a global network. From this perspective, inter-city relations are, in part, constitutive of “globalization.” We developed this line of inquiry in several theoretical and empirical publications (Smith and Timberlake, 1995a, 1995b, 1998, 2001; Shin and Timberlake, 2000). The findings of these efforts are discussed in more detail below.

The roots of our efforts are found in important themes in comparative sociology, urban ecology, and urban geography. It resonates with the work of other scholars who are currently attempting to shed light on the ways in which cities are involved in the macro-level processes associated with what now is commonly termed as “globalization.”

World City Networks & Hierarchies. McKenzie and other early urban scholars saw cities in terms of a system of cities, related to one another along a dimension of power. Some of his work is prescient of much more recent research: global integration, world-wide hierarchies of dominance, and competition and change were all important themes in it: “Old centers lose their relative importance as new factors enter to disturb the equilibrium....New centers of dominance are arising...” (McKenzie, 1927). More than a generation later, the same logic appears in the work of scholars of urban planning such as Peter Hall (e.g., 1966) and John Friedmann who are the immediate pioneers of the now extensive body of research on world cities. In 1986 Friedmann produced a ranked list of key world cities, providing a figure that “maps” the linkages between them. He argues that world cities can be located in a global hierarchy based on their positions in the global geographic nexus of economic power: “Cities can be arranged hierarchically, roughly in accord with the economic power they command (1995: 25-6). Moreover, because cities can rise and fall in this hierarchy it becomes important to recognize “the existence of differences in rank and investigate the articulations of particular world cities with each other” (23). Friedmann labels world cities as either “primary” or “secondary,” according to location within the world-system core or semiperiphery. Thus, he provides a rough map of the world city system based on the functional importance of these city-nodes. New York, London, Tokyo and Los Angeles are among the highest ranking “core primary cities” in his scheme – an intuitively appealing result. Saskia Sassen probably provides the most thorough treatment of “the world city hypothesis” in The Global City: New York, Tokyo, and London (1991). Here she provides a global overview of her version of the notion, detailed case studies of the three great metropolitan centers, and a synthetic conceptual argument, including an overview of contemporary globalization processes and the dynamic roles that world cities play in them. She claims various functions relating to the “command and control” of today’s world-economy concentrate in global cities, even as manufacturing is increasingly dispersed to ever more far-flung regions of the world. These cities are the “command posts for the world economy,” the sites for global finance and other specialized service firms, the sites of key innovations, including innovations in services, and they are “markets for the products and innovations produced” (Sassen 1991:3-4).

A number of other case studies have largely substantiated John Friedmann's notion that there is a distinct category of "global" and "world cities" housing activities and organizations that exert international coordination and control (for example, King 1990; Ross and Trachte 1990; Sassen 1991, 2002), and there are fine examples of more historical comparative case studies of cities in the context of global political economy (for example, Rodriguez and Feagin, 1986; Smith and Feagin, 1987; Hill and Feagin, 1987; Feagin, 1988 ). Castells (1989, 1994) pulled various strands of the global restructuring/world cities literatures together. He emphasizes the key role of “information technology” as an underpinning for contemporary globalization, arguing that “the national-international business center is the economic engine of the city in the informational global economy” (1994:29). Castells’ work predicts that dominant global cities will rise to commanding heights by developing the infrastructure that is required to “capture” key information flows, leading to spatially defined urban hierarchies. Therefore, understanding urbanization requires a network-based relational view of how structural similarities among cities and social change within cities are influenced by world‑systemic processes.

Conceptualizing Linkages Among the World’s Cities. Though the most obvious kinds of flows among cities are those of an explicit economic nature, such as commodity flows, there are other important ways in which cities are connected. Other economic linkages among cities in the human form include flows of labor, sales and producer‑services personnel, and managers. When people migrate or immigrate from one city to another in order to find employment they represent labor flows. An immigrant from Tijuana to Los Angeles, drawn into the garment industry there, is an example. When a manager travels from corporate headquarters to the site of a branch plant in another city, this represents an economic flow in the human form. The primary material objects linking cities can be traced in the flow of commodities, both in the production process and in the distribution process. Some cities house important value‑added processes for commodities that are shipped into the city from other urban locales. Cotton fabric is shipped daily to factories in New York, Los Angeles and other cities with garment industries. Other cities house break‑of‑bulk transportation facilities for moving commodities to smaller cities where they will be purchased by consumers. Examples of economic flows taking the form of symbolic communication include business‑related internet-based communication, telecommunications (e.g., telephone calls, faxes, TELEXes, etc.) mail orders, other business‑related mail, etc. In principle, detailed information measuring different kinds of flows could be used to produce one or more "maps" of the world system of cities. Figure 1 (adapted from Smith and Timberlake, 1995a) illustrates one possible conceptual organization of the various flows which link cities, weaving them into a global network. (Of course, the same relations also characterize inter-national connections in the world-system, too.)

Figure 1. Typology of Global Inter-City Relations

FUNCTION FORM

Human Material Symbolic

Economic labor migration commodity faxed orders

flows

Political ambassadors arms shipments threats

Social family migration remittances personal mail

Reproduction

Cultural dance troupes blue jeans Hollywood films

Network Analysis. Both world‑system analysis and the literature on world cities evoke network imagery and are filled with references to “flows,” “exchanges,” “nodes,” and other words and phrases that evoke network relations. For example, Lo and Yeung describe “the functional world city system” in terms of various linkages, claiming that “world cities are at the points of convergence of these networks and thus acquire growing centrality and importance. Network functions are embodied in financial flows, headquarter-branch relations, high-tech service intensity, and telecommunications networks” (1998:10; and see Meyer, 1986). Although many employ relational imagery, few scholars of global cities attempt to systematically study these types of networks. Most choose instead to focus on case study techniques that describe the ways that particular world cities take on “command and control” functions within global networks.

While the leading theorists may be unlikely to turn to high‑powered statistical analysis to test their models of global structure and hierarchy, quantitative network analysis is particularly well‑suited for this purpose. With appropriate data, this methodology allows us to simultaneously analyze multiple patterns of flows, exchanges or linkages between cities (or other nodes), revealing the complex patterning of connections between them as well as the structure of the entire network. It gives researchers a powerful tool to examine global flows of people, commodities, capital, information, etc. A number of researchers have used network analysis to examine the structure of the world‑system (Snyder and Kick, 1979; Steiber, 1979; Nemeth and Smith, 1985b; Smith and White, 1992; Sacks, Ventresca and Uzzi, 2001), but few have used it to explore the nature of the global system of cities.

Network analysis measures a variety of formal properties of structures and relationships. Two characteristics of networks are of particular interest for studying global city systems. First, the idea of structural or relational equivalence in networks, usually operationalized using various forms of "blockmodeling", is the most familiar technique in previous network research on the global system. Such research seeks to identify the actors (usually nations) that fit together into broad groups or "blocks" according to the way they are similar to each other in terms of their patterns of ties or flows (exchange of diplomats, provision of aid, international trade, migrants, investment, etc.) and different from other sets of actors in these same respects. Snyder and Kick (1979) and Nemeth and Smith (1985b) used the popular CONCOR (Convergence of Iterated Correlations) algorithm (see White, Boorman, and Brieger 1976) that gauged structurally equivalent blocks; later Smith and White (1992) used a relational distance algorithm, called REGE, which more accurately captures the idea of role similarity. These measures of equivalence provide a technique for grouping nations or cities (or, in principle, other geographic units for which data are available) together according to similarities in their pattern of exchanges. Theoretical notions about the link between a city's functions and its role in the global urban system suggest that cities should group into rough "levels" corresponding to their positions in the world city hierarchy (analogous to the “world-system positions” into which nations group).

In addition to probing overall structures of networks, the methodology also provides insight into the attributes of particular points. One particularly important concept is that of centrality. Sociologists link centrality in communities or interorganizational networks to power, prestige and economic success (Laumann and Pappi, 1976; Laumann, Galaskiewicz, and Marsden, 1978; Galaskiewicz, 1979; Burt, 1982), while anthropologists and geographers have focused on centrality in trade networks between places (for instance Hage and Harary, 1981; Pitts, 1965, 1979). Formal network methodologists argue that there are several distinct types of "centrality," operationalized using different formal measurement algorithms (see Freeman, 1979). Because of technical problems that are only now being solved, quantitative measurement of point centrality is a complex proposition, especially if the data on the flows or relationships between points incorporate information about the amount or value of the exchange (i.e. as opposed to simply reporting presence or absence of a tie) (see White, 1989). But nodal centrality is obviously relevant to understanding world city systems: the key role of world cities as “junctions of flows” (Harris 1994) should be strongly related to a place's ability to control and broker various types of international exchanges, to serve as a source of information and capital flow, and to act as a magnet for certain types of migrants or high technology.

Operationalizing Intercity Linkages for Network Analysis of the Global City System. The various types of intercity flows discussed above are not highly abstract (see Figure 1, above), so they might appear to be straightforward to measure. However obtaining data on actual flows of people, materials and information between global cities presents a particularly daunting task. Further, the nature of network analysis makes missing data particularly problematic.

Because of the way quantitative data are gathered and compiled, there is a dearth of relational data on all social phenomenon. There is growing theoretical interest in the various social sciences on how social structures are generated through the relationships and linkages between people, localities, institutions, nations, etc. (see, for example, Tilly, 1984). But statistical information still tends to be collected in the attributional form. That is, sources report values on the characteristics of particular units (i.e., an individual's income, a city's population, a nation's gross national product) rather than information about their relationship to others (in terms of either links or flows). Therefore, it is very difficult to obtain data on international flows or links between social units (whether these are cities, regions, organizations, corporations, or nations). So while it is easy to get attributional data on various nation‑states, and there is some good network data at the inter‑national level (commodity trade flows, for instance, see Smith and White, 1992 -- a study we are now replicating and updating), compilations of networks of interactions or flows between world cities are not readily available. This does not mean that these data do not exist or that pulling them together into a useable format is impossible; rather, it means that this task is difficult. The difficulty is compounded by a methodological issue. While network analysis is a powerful tool to understand social structure, it has stringent data requirements. Other types of statistical analysis use relatively simple, standardized ways of adjusting for "missing cases." But incomplete data in network matrices is more difficult to accommodate. An important part of the proposed research will thus involve careful attention to collecting data for as many city pairs as is possible.

Global City Network Data. Economic flows can be embodied in the movement of people when they migrate from one geographic place origin to a distant destination: labor migration. Periodic surveys of national populations for certain periods of time may allow us to piece together a map of global labor flows. Of course, good counts of migrants will not be available for many countries in the world, and when they are, cities of origin may not be coded. But the study of migration is central to demographic inquiry, so this could be a potentially fertile field for data in the future that might allow us to begin to chart a migration‑based world city system grid.

Economic flows in their material form may be operationalized by describing commodity chains linking geographic locales (see Gereffi and Korzeniewicz 1994). The required data would consist of some measure of the nature and volume of commodities by cities of origin and destination. Once again, it’s unlikely that complete data for all of the world's large cities are available, but we can begin to piece together a partial map of the world's city system by using the data that are available and making efforts to find new sources. For example, in this country a Commodity Flow Survey is available from the U.S. Department of Transportation. The survey provides a complete network data base (from systematic sampling of commodity flows) on the value and number of commodity flows by zip code. But, since it can only provide data for cities in the U.S., they are not directly useful in a study of global networks. There may, in fact, be data compiled by commercial interests that would be ideal for mapping the morphology of commodity flows between world cities – information on port-to-port movements of containerized trade, for example. However our efforts to obtain these data have been unsuccessful so far. But we do have complete matrix data, as yet unanalyzed, on air cargo flows between the each nation’s leading airport. We plan to examine these data as part of the project.

Communication among actors in geographic locales remote from one another takes place in the absence of face‑to‑face contact. Firms, businesses, and business people with far‑flung operations and interests use e-mail, telephone calls, Telex messages, Faxes, telegraph and postal mail for such purposes. In principle, it should be possible to sample the volume by locale of origin and destination of some of these communications from existing records. But this information also seems to be difficult to obtain. Scholars of the communications industry and the major telecommunications corporations themselves may be willing to cooperate in generating the necessary data to begin to map the global inter‑city information network (see, for example, Barnet, 2001). In principle, sampling pr