GaWC Research Bulletin 426

GaWC logo
  Gateways into GaWC


World Cities in a System of Nation States

R. Musil*


In recent years, several authors have shed more light again on the state and its territorial restructuring as a relevant factor for the formation of world cities. As such, the concept of re-territorialisation stresses the transformation and rescaling of statehood as well as the rise of the local state. However, in this context, Saskia Sassen warns to avoid the endogenity trap: meaning that the world city phenomenon cannot be explained by simply describing it. Instead, it is necessary to examine the structures in behind recent processes to gain a better understanding. By following this idea, it is the intention of this paper to assess the impact of the territorial state system and its historic pathways on the recent world city system. In doing so, the paper will consider three dimensions of statehood – geopolitical power, macroeconomic arrangement and territorial structure – and will estimate the impact of these dimensions on countries’ total gross connectivity within the world city system. The outcome reveals firstly a high explanatory value for the three regression models and secondly that the relevance of “state impact” differs considerable among the countries as well as among the three dimensions of statehood.

Key words: World City, State, Territorial System, Urbanism


During the 1990s, at the heyday of the “globalisation party” (Peck, 2000; in: Keil & Brenner, 2003, p. 261), so-called “de-territorialisation theories” understood urban processes as an interplay between the global and the local scale. Against the background of a rising global financial system and globally integrated production networks of multinational enterprises, the territorial state lost its explanatory value for the dynamics of a globalised economy. The crisis of the Keynesian welfare state and the break-up of Fordist production structures on the one hand (Jessop, 1997, p. 60), coupled with the rise of new (spatial) organisation structures in services and industrial production on the other hand, required new theories to overcome the “territorial trap” (Angew, 1994). Authors like Manuel Castells (1989; space of flows) and Saskia Sassen (1991; global city theory) developed fundamental theoretical approaches for a new understanding of globalisation as a spatial process. Both eclipsed the state and its territorial logic and stressed a relational perspective, where cities were seen as nodes in the network of globalised capitalism.

In recent years, however, several authors have questioned this one-sided perspective through the term of “re-territorialisation” and accentuated the role of the state, public institutions and territorial arrangements for the formation of a global city (Brenner, 2004; Ma & Timberlake, 2012). It is notable that Saskia Sassen herself emphasises the relevance of the territorial state for the functionality of globalisation in general, and the formation of the world city in particular (Sassen, 2006, p. 13). Rather than focusing on the shift of power from the state to supranational scales, she sheds light on the shift between different state institutions; according to her argument, not the state as a whole, but rather certain aspects of it were affected by an erosion process, while other aspects gained more power and influence (ibd., p. 8). In this context, Sassen warns against the so-called “endogenity trap”, which means that social research “cannot understand the x … by confining our study to the characteristics of the x” (ibd., p. 4). In the context of the global city debate, this means that the integration of a city into the world city network cannot solely be explained by relational-global processes, such as the networks of producer service firms. In order to fully understand these dynamics, an analysis must be undertaken of the institutions and the structures of the territorial state.

It is the intention of this paper to examine the factors of the territorial state that influence the integration of a city into the global economy. The basic assumption is that the relational space in the world city network does not exist detached from the territorial state. It is clear that both systems are related to each other (Taylor, 2000, p. 14), but unclear to what extent. By following this perspective, the paper expects to enhance current understanding of the spatiality of globalisation by analysing the territorial impact on the relational global city system. The paper is structured as follows: after reviewing the theoretical aspects of the territorial vs. relational interplay, several impact factors of the territorial state – such as macroeconomic structure, urban pattern, spatial disparities or organisational structure of the state – will be discussed (section 1). The empirical section first gives a descriptive overview of the cohesion between the connectivity of European world cities and territorial structures (section 2) and then presents a model that analyses the impact of the dimensions of the territorial state on a country’s overall connectivity in the world city system (section 3 and 4). The paper ends with a discussion of theoretical implications of the empirical results (section 5).

Territorial implications of the world city system

This section first gives an overview of the role of the state in the global/world city debate and subsequently discusses arguments suggesting that the role of the territorial state be more strongly considered with regards to internal socio-spatial structures within the world cities and their external formation on the global scale.

The Territorial State in the World City Debate

Earlier contributions to the world city debate accord the territorial state a particular role in world city formation. As such, Friedmann & Wolff (1982) formulated that, “World cities lie at the junction between the world economy and the territorial nation state” (ibd., p. 312). This statement assumes at the least an indirect effect of the nation state on the extent to which cities located in that particular nation state are classified as global cities. A further argument formulated by these authors focuses the “public service cluster” and the fiscal capacity of state institutions, and they assumed that these have an important impact on the functional performance of a world city (ibd., p. 320). In this vein, Friedmann (1986) stresses the pressure that world city formation has on state structures, such as welfare, social and fiscals systems, which can be conquered by states in very diverse ways (ibd., 77). In the “world city hypothesis” he further ties the system of world cities with the system of territorial states; for example, on a world city map, the rank of world cities is based on the core-periphery assignment of the world system in the 1980s (ibd., p. 72). Thus, in these early works, the internal structure of the world city and its external relations was seen as being partly influenced by the territorial state system.

Subsequent de-territorialisation theories that had a seminal influence on the research field of world cities do not include the state as an explanatory factor. For instance, Manuel Castells’ new spatial category, the virtual-globalised “space of flows”, dominates the “space of place” and transforms it into a new type of city, the “informational city” (Castells, 1989). This transformation, which is caused by the revolution of information technology and the rise of a globalised capitalism, does not recognise the territorial state as relevant actor. The same is true for Saskia Sassen’s global city theory (Sassen, 1991). According to this theory, the global relevance of a city is determined by its role as a marketplace for highly specialised knowledge, which is provided by advanced producer-oriented service firms (ibd., p. 9). Both the informational city and the global city theories exclude the explanatory contribution of the territorial state or the welfare system for globalisation processes, in particular on the urban scale. Furthermore, the self-conception of these theories is based on the idea of overcoming the “territorial trap” (Angew, 1994), which seems to be a diametral antagonism to the relational logic and, as such, hinders the analysis of global processes.

In the historic perspective, Taylor (1995) identifies three historic phases of a changing mutual benefit between states and cities: first, the creation of territorial states since the 16th century; second, the nationalisation of this container state since the mercantilist era (19th century); and, finally, the demise of territoriality, caused by the incongruence of the container-state and the global economy (ibd., p. 54). The author concludes the characterisation of the initial phase as follows: “Which leads us to ask whether London as world city needs Britain, or New York as world city requires to be part of the large territorial state that is the USA” (ibd., 59). This quotation is meaningful in the following aspects: first, it understands the global de-territorialisation process as the dissolution of the general mutual benefit for cities and territorial states, according to which the first is detached from the latter. Second, any explanatory value of the nation state is disclaimed here. Further, it expresses a perception of historic processes as independent stages, similar to historic materialism. This historic approach hinders the view on trajectories and historic pathways that might connect these stages, insofar as the (historic) logic of the former period determines the development of the more recent one.

From a methodological perspective, world cities and their global networks demand new data and approaches, which have been developed by the GaWC research group and allowed it to overcome the “methodological territorialism” (Brenner, 2004, p. 38). As “state-istics” do not allow relational analysis of sub-national entities on a global scale, Taylor et al. (2003) developed a database based originally on 100, later 175, globalised producer-oriented service firms (Derudder et al., 2010). This new methodological approach was coherent with the theoretical assumptions of global city theory and opened a vast field of research questions and inspired countless empirical studies on the integration of cities into the world city network. However, with regard to Taylor’s question above, we must ask how the dominant role of London, New York or Tokyo in the world city network can be traced back to a small number of producer-oriented service firms, while the size of the national economy, its historic territorial structures and its geopolitical power do not seem to constitute relevant variables. In the following section, the potential influential factors of the territorial state on the global city will be discussed.

The State as a Relevant Category in World City Research?

Although these de-territorialisation theories and methodologies had an important influence on urban research and established a new field of (world city) research, some authors have asked whether these concepts did not “throw out the baby with the bath water” (Keil & Brenner, 2003, p. 258). For them, the territorial state is a dominant actor that produces new forms of territorial arrangements – coined “re-territorialisation” – and as such is an agent between global and local processes (ibd., 264). Thus, globalisation leads to a new regulation of territorial entities – rapport territorial – which affects urban processes (Schmid, 1996, p. 29). From this perspective, world cities are a “product” of a transformation of institutions and spatial scales of statehood.

While these authors emphasise the transformation of territorial structures and governance, we also should consider public institutions and territorial arrangements that have not been affected by global transformation and still follow the logic of a fordistic container state. As such, the different types of national welfare systems, which are frequently a product of “national idiosyncrasies”, essentially remain in their historical arrangements (Esping-Andersen, 1990, Kaufmann, 2003, p. 10). Indeed, further governmental subdivisions, which are often based on a history that reaches much further back than the paths of nation-building (Flora, 2000), are guarded by strong regional movements, elites and interest groups. In contrast, new territorial arrangements such as “regional territories” (Priebs, 2004, p. 701) or “local states” tend to be much less stable (Lipietz, 1998, in: Becker, 2007, p. 231). The permanent reconfiguration of governance and planning regions in Great Britain (see for instance Pain, 2011) underlines this argument. Subsequently, more light should be shed on the effect of these historic territorial structures that follow more a pre-fordistic logic than the relational/global logic, and therefore affect the formation of world cities.

If globalisation did not lead to an erosion of the territorial state, but rather to a reorganisation of statehood as well as to a continuation of historic arrangements and institutions, we must examine its contribution to the formation of the global city. The following section follows Sassen’s endogenity thesis and elaborates on the explanatory power of the territorial state on the two main assumptions of the global city theory: their external relations (global position and spatial pattern) and their internal socioeconomic structure (social and spatial polarisation).

Territorial State and Social Polarisation in the World City

The concept of a globalised, “new type of city” (Sassen, 1991, p. 4), which assumes the rise of increasing socio-spatial dualism (“dual city”) alongside global integration, provoked critique. Social scientists refused to recognise the assumption of increasing social polarisation, in particular in the case of European world cities (Hamnett, 1996, 2003, Häussermann & Roost, 2000, p. 89). Empirical case studies for European cities display an increase of highly skilled professionals in London, Paris and Amsterdam, which has led to a rising income gap, but the extent of disparities – as well as their causes – differs between American and European cities (Hamnett, 2003, p. 102). In sum, two factors were crucial: first, the capacity of European welfare systems; second, the size of the public sector (Hamnett, 1996, p. 108; Fainstein, 2001). In the case of European cities, social inequality shows a very heterogeneous pattern, which resists – at least in parts – the thesis of a global trend towards a dual city.

Several authors (Fainstein, 2001, Hamnett, 1996, Häussermann & Haila, 2004) stress the relevance of the (European) welfare systems as an explanatory factor for low social disparities, albeit their variety between the countries. For Letho (2000), the impact of European welfare systems on urban development goes far beyond the aspect of social disparities. For him, the welfare state influences urban development in several ways, as it determines several social dimensions of a society (ibd., p. 115): It governs the redistribution of income and welfare; it enables female participation in wage labour; it offers jobs in social services; it decreases economic risks; and supports the consumption potential of households. Finally, it enables households to finance home ownership and a private car, which has a significant impact on the spatial pattern of urban development. Without going too much into detail, we can conclude that the public social welfare system influences the social structure of a (global) city far beyond the aspect of social security.

In regards to the welfare systems, Dieleman and Hamnett (1994) highlight: “It is important that, in examining the impact of the processes of globalisation on global cities, we do not lose sight of their national backgrounds, contexts and cultures” (ibd., 363). At this point, this statement shall not be limited to social (dis-)order of global cities; rather this argument shall be widened for the formation of the global city in general.

Territorial State and World City Distribution/Formation

Through the lens of the territorial state, the distribution of global cities exhibits a very uneven pattern. For instance, France has one dominant global city, while we can number five German global cities on an average global city rank (Taylor et al., 2011, p. 127). The methodological approach of GaWC can describe these patterns, but global city theory cannot explain them (Taylor, 2004, Taylor & Derudder, 2004, Taylor et al., 2011, overview: Matznetter & Musil, 2012). This section sheds light on the influential factors of territorial state that might support the explanation of these patterns.

At first, the spatial distribution of world cities reflects the existence of varied historic urban systems. For Europe, Hohenberg & Lees (1995, p. 247) differentiate between the “Rhenish model”, the densely urbanised European core region; the “Parisian model” with a dominant city; and the “Peripheral model”, characterised by a weak urban structure. The uneven spatial pattern of urbanisation goes far beyond the formation of territorial statehood, and even influenced this process (Flora, 2000). Stein Rokkan traces the process of state-building – aside from factors such as cultural-religious homogeneity – back on heterogeneous urbanisation in Europe (ibd. 79, p. 103): while the peripheral-rim regions facilitated the formation of centralised stable territorial entities (like Spain, Scandinavia, France), this process was more insecure, interminable and federal in the “city-studded” core (e.g. German states, the Netherlands or Switzerland; ibd., p. 179). The historic urban system in Europe can be seen as a stable development path, in which historic structures influence recent urbanisation processes as well as the spatial formation of statehood.

The outcome of this historic process is a heterogeneous mosaic of spatial territorial organisation: states differ in their degree of federalism (the right of sub-national units to decree laws and taxes) and of decentralism (autonomy of public expenditure on the sub-national level) (Kaiser & Ehlert, 2006). While its effect on regional disparities has been the focus of several studies (Rodriguez-Pose & Ezcurra, 2009, Shankar & Shah, 2003, Ezcurra & Pascual, 2008), the impact of territorial organisation on global city formation has not yet been analysed. It is assumed that federal territorial systems offer a larger judicial scope for the formation of (several) local states, while decentralised financial systems and their fiscal redistribution systems allow larger fiscal autonomy to urban systems. For instance, territorial organisation in Great Britain is characterised by a low level of federalism and a high degree of decentralism; it might be interesting examine the influence of this territorial arrangement on the permanent reform of governance and planning regions. Further, the relevance of decentralised territorial structures is considerable: for instance, the complex (and intransparent) Austrian fiscal system (Finanzausgleich) redistributes 13.5 billion euros between territorial entities on an annual basis (Schratzenstaller, 2005). As the two dimensions of spatial territorial organisation (federalism and decentralism) show several forms of combination, the impact on world city formation seems not to be linear. However, it shall be supposed here that the territorial structure as well as the spatial redistribution systems of the fiscal system influence this process and that they more often follow the logic of historic-territorial structures rather than one of a globalised economy.

A further aspect is the divergent structure of national economies; the perspective of “varieties of capitalism” emphasises the impact of state institutions and the differences therein between developed countries for the formation of diverse forms of capitalism (Hall & Soskice (Eds.), 2001, Amable, 2003). Labour and social rights are counted among the relevant impact factors, as are methods of financing and coordinating the market (cf. Hancke (Ed.), 2009). Furthermore, there is no implicit assumption that the form of capitalism and the welfare regime are interconnected. This discontinuity of national economies seems to determine the integration of a city in the global city network: Ma and Timberlake (2012, p. 2) differentiate between states in two dimensions: the dominance of statehood over the market (developmental vs. liberal laissez-faire state) and the degree of economic internationalisation (domestic vs. foreign capital). Their empirical analysis showed different forms and pathways of integration of cities in global or national city systems, depending on the type of national economy (ibd., p. 14). The process of de-territorialisation, the disintegration of national urban system, seems to be an inhomogeneous process, which is affected by national economies.

Beside historic pathways and territorial and macroeconomic varieties, political regulation had a strong impact on the formation of national agglomerations towards world cities, as Keil and Brenner (2003, p. 265cf.) have shown in several examples. For instance, London’s world city formation was a steady effort of various national administrations, both by conservative and liberal governments. Furthermore, early liberalisation in 1980s induced the rise of London’s financial industry (1986: Big Bang) as well as the foreign investments in London’s real estate market, which were supported by new public institutions (Dockland Authority), enterprise zones and new governance structures (Greater London Authority) (ibd., p. 265; Zehner, 2011, p. 47f.). The number of examples of national policies supporting the formation of global city (or not) could be extended arbitrarily. The point is that the process of de-territorialisation required the support of the national government and policies – whether conscious or unconscious – and that this happened in a very diverse manner.

In sum, we can identify several layers of the territorial container-state that have an impact on the inner structure as well as on the external integration of a world city. These “territorial regimes”, which are based on different historic urban systems, different territorial arrangements, different macroeconomic regulations and policies, question the image of a homogenising de-territorialisation process. Thus, we follow Sassens warning about the endogeneity trap: we cannot explain the formation of a global city network without looking at the structures behind it. To go one step further, we will test the explanatory value of territorial state structures regarding the distribution and strength of integration in the world city network. In other words, what is their contribution to the “production” of the global city?

Territorial Pattern of the world city system

How are world cities – through the lens of the territorial state system – distributed? Our sample contains 167 states1 as well as the 357 cities that are included in the 2008 GaWC database2. Thereby, cities with a weak integration into the world city network (connectivity) are also considered in the analyses. The large sample of cities used in this paper allows more insights into the structure of national urban systems and also into the integration of several (peripheral) states. The distribution of these world cities as well as the sum of connectivity on the national scale displays a very uneven pattern: at the bottom of the global state system, 43 countries do not show any world city-ness, while on top, in just ten countries we find 167 cities that are integrated into the world city system and hold almost 44.9 % of the worldwide connectivity (see tab. 1).

A clear, positive relationship between the total strength of the national connectivity and the number of global cities can be asserted. Of course, we must consider a strong size effect, which is emphasised by a strong R² among the world cities and the countries’ overall connectivity (see fig. 1, left graph). However, besides this effect, it can be supposed that high national connectivity cannot be achieved by a single city. The highest connectivity of a country with just one global city is the city-state Singapore (rank 23), followed by Ireland (rank 36) and Thailand (rank 37). Thus, high country connectivity seems to coincide with the spatial dispersion of world city-ness.

Table 1: Top-10 countries by GaWC connectivity


Number of GaWC cities

GaWC conn. (national sum)

Share of global GaWC conn.

Share of global GDP (%)

Population 2011(in 1,000)

Share of world population 2011 (%)














































































(Source: GaWC database 2008)

An additional indicator that shows the degree of the “horizontal” integration of a country into the global city system is the share of the most important world city on a country’s total connectivity. For instance, New York is the second most important world city, but its share of the total connectivity of the US makes just 8.4 %. Less surprising is Frankfurt, which holds just 16.5 % of the vertically structured German urban system (Taylor et al., 2011, p. 127), but even Paris, the dominant capital of France’s vertical urban system, holding rank four in the global city system, makes less than half (48.7 %) of the country’s global connectivity (fig. 1, right graph). However, it must be considered that the question of world city centralisation in the national urban systems concerns just a small number of countries: only in 44 countries we can indentify more than one world city, and in just 17 countries is the share of subordinate (national second-tier) world cities more than 50 %. Either way, the spatial pattern of national second-tier global cities and its correlation to the total countries integration seems to be a question that should be further explored.

Figure 1: Relationship between the national connectivity, the number of world cities and the share of the country’s first-tier world city

(Source: GaWC database 2008)

Methodological aspects: territorial variables for the relational urban system

Although the impact of the state on the connectivity of world cities is a relevant question, very little empirical work has addressed it. As a rare exception, Taylor shares empirical facts on the relation between the territorial state and the world city network (2000, p. 14). By cross-tabulating the system of states against the roster of world cities, he focuses on three aspects: first, the distribution of world cities (regarding their world city-ness) over the states, whereby he refers the size of the state and its degree of decentralism. Second, he focuses on the size of national markets; he assumes “a positive relationship between the size of national economies and the distribution and levels of world city formation process” (ibd., p. 15), which seems to be affirmed by the descriptive analysis in section 2. Finally he focuses on the market potential, measured by GDP growth rate, which has an explanatory value for world cities located in transformation markets. In a more recent publication, Taylor (2011) further traces back the distribution of world cities and the primacy of global connectivity on historic patterns of urbanisation, but does not go beyond its description (ibd., p. 332). Although these papers neither elaborate further analysis nor focus more on the explanatory power of relational data for transnational processes, they are a rare example for linking world city data towards the system of territorial states.

It is the intention of this paper to enhance the country perspective and to examine the impact of “container-state related” factors on the spatial distribution and the strength of integration of a city in the world city network. Proceeding with Keil and Brenner (2003, p. 264), we consider the double function/role of cities: first, they act as coordinates of territorial power and layers in the institutional system of states; second, they are nodes in the global network of a capitalistic system. Following this insight, one should not expect to explain connectivity solely through state related variables – which would be a fall-back into the territorial trap. To overcome this, the explanatory value of variables such as the concentration of the national urban system, the territorial organisation of the states or the geopolitical power of the state shall be estimated.

Research Design

Three dimensions of statehood will be differentiated in order to analyse the impact of the state, its territorial structure and the national economy of the world city system: first, the geopolitical position of a country in the global state system including military power, integration into international organisations, and the size of the national market will be considered. The second dimension focuses on the “macroeconomic arrangement” of statehood: the role and quality of public institutions, economic policy and the size of welfare system, which all differentiate between more market oriented, laissez faire states and more regulated states. The third dimension describes the territorial-spatial structure of statehood. This differentiation makes in so far sense, as the disparities on the global scale between the countries might superimpose differences in the internal territorial structure or the macroeconomic arrangement of the countries; the analysis of these two questions even does not make sense for all countries. Thus, on a global scale, the question of the territorial organisation of public institutions makes little sense if a country’s statehood merely exists on a rudimental level – which is of course the case for many underdeveloped countries. For these reasons, the analysis of the impact of the three dimensions of statehood on the global city-ness of a country will be on different spatial scales: the first dimension on the global scale that contains all countries where world cities are located; the second and third dimension is limited to 34 OECD countries.

As the distribution of the number and strength of world cities internationally are the subject of our study, the number of world cities and their connectivity are the dependent variables; both will be aggregated on the national scale. To consider the size-effect between countries, the dependent variable will be normalized for the countries size of population (global city-ness per million inhabitants) in the first model. As the urban/regional system and the territorial structure of the state date back to a pre-globalisation period, a clear direction of influence from the state system (independent variable) to the world city system (dependent variable) might be expected. These relational data derive from the GaWC database, which is based on a network of 175 producer-oriented service firms and their global networks, surveyed in 2008 (Taylor et al., 2011). As earlier surveys – from 2000 or 2005 – are not comparable, a time series analysis is not possible (Derudder et al., 2010). The independent variables will be discussed in the following section.

Research Question and Hypothesis

Based on the theoretical discussion on the relationship between global city-ness and the state, following research question are formulated.

  • Does the geopolitical power in the global system of nation states have explanatory value for countries’ integration into the global city network? A high explanatory value might be expected, at least, because (soft) geopolitical power (Nye, 2005) goes hand in hand with a country’s wealth level.

  • Which state related factors of the “macroeconomic arrangement” – e.g. economic policy, social structure of welfare system – influence the world city integration of a country? Do a liberal economy with social disparities and a weak welfare system correlate with strong integration into the global city system? If we follow the theoretical assumption of Saskia Sassen and research on the question of world city governance (Brenner, 2004), a clear influence in the direction “liberal state – strong connectivity” is to be expected.

  • Are decentralised states more successful in global city integration? The descriptive analysis above indicates a clear impact of federalism and decentralism on the integration of a country into the world city network.

The explanatory value of statehood on world city-ness – three multi-linear regression models

To estimate the impact of the three dimensions of statehood, the following sections present multi-linear regression models for each of them. Each section contains a descriptive analysis of the independent variables, the regression model and its interpretation.

Dimension 1: The Global System of the Nation State

Do variables that indicate the hard and soft power of a country in the global state system (Nye, 2005) explain the distribution of the gross connectivity per inhabitant in each country? We must be aware that the substantial wealth gradient has a significant impact, which can be shown by the GDP per capita (R² 56.0 %), but this is not in the focus here. Of course, the wealth level determines the geopolitical influence of a country, so it is considered more indirectly – e.g. the military expenditure per capita. Further variables that indicate the soft power of a country include membership in international organisations or the openness of the society and its institutions (share of female representatives in parliament). Additional economy-based indicators include the GDP (market size) and the patents per capita (innovatory potential of an economy). Two indicators that represent the level of statehood are the share of taxes in the GDP and the dummy variable of federalism-unitarism.

Table 2: Descriptive Analysis of independent variables model 1








log GDP 2011 in mill. USD







log patents per Mio. inhabitants







log military expenditure per cap. 2011 in USD







log public aid expenditure 2011







log share of female representatives in parliament 2011







log tax rate 2011 (share of GDP)







log no. of membership in internat. organisations 2011







dummy federalism-unitarism








The multi-linear regression model (tab. 3) contains all variables described above, except those that do not have significant explanatory value, e.g. the tax rate and membership in international organisations. While the model contains only six variables, it has a strong explanatory value – at least 67.4 % of the dependent variable – for 106 countries. Surprisingly, the variable military expenditure per capita has the strongest explanatory value, even in the first three models; the second variable with a strong explanatory value is the GDP, showing a negative Beta coefficient; thus, the market size negatively impacts the (size-normalized) dependent variable. This is an interesting point: while the absolute connectivity correlates positively with the size of country or of its economy, this is not true if we eliminate the size-effect; small economies seem to have a disproportionately high connectivity. Military expenditure and marked size together explain 53.1 % of the dependent variable (model 2); when additional variables are included, the GDP becomes the highest explanatory value (model 5 and 6). The other four variables in the model raise the R² for just 14.3 %, whereby the patent density has the strongest explanatory value in this group, followed by the public aid expenditure, which is measured in absolute terms (million USD in 2011) and is therefore highly dependent on size and welfare. The share of female representatives in parliament and the dummy variable (federalism-unitarism) gives the weakest, but a nevertheless significant contribution to the model. Although it is not associated with geopolitical power in a narrow sense, it indicates the level of development of statehood; furthermore, this variable is a hint for the relevance of the third dimension of statehood.

In sum, two variables – military expenditure and GDP – have the strongest explanatory value; this simple regression model explains more than two thirds of the distribution of our dependent variable across 106 countries, suggests a strong impact of the global state system on the world city system.

Table 2: Multi-linear regression model of geopolitical indicators

stand. Beta-Coeff.

N 106







log military expenditure per cap. 2011







log GDP 2011






log share female representatives in parliament 2011





log patents per mill. inhabitants 2011




dummy federalism-unitarism 2005



log public aid expenditure 2011








Durbin-Watson 2.007 Tolerance and Variance Test negative

dependent variable: gross connectivity per mill. inhabitants

(Source: GaWC database 2008; databases: UNCTAD, UN, COW, WHO, ILO, UNIDO)

Dimension 2: Macroeconomic Arrangement

How does the national economy, including the public sector, social policy, and the structure of welfare system, which is described here as “macroeconomic arrangement”, influence the world city integration of a country? If we follow the assumption of John Friedmann and Saskia Sassen, we might expect states with a liberal economic policy, a small public sector and a weak welfare state to “produce” more successful world cities. The connectivity in each of the 34 countries is the dependent variable. The independent variables (see overview in tab. 3) indicate several aspects of the macroeconomic arrangement: the quality of the welfare system (Gini coefficient after social transfers, public health expenditure) as well as the size of public sector (share of public expenditure of GDP) and the tax strain of incomes (share of net income of total income). Furthermore, economic policy and the degree of liberalisation (barriers in trade and investment, market regulation, share of pension fund within the GDP) are considered. Two additional indicators respect the relevance of trade unions (trade union density) and the general labour productivity (GDP per work hour). At least, also in this model, the dummy variable federalism-unitarism is included.

Table 3: Descriptive Analysis of independent variables model 2








Gini coefficient after social transfers







public health expenditure, share of total pub. exp.







public expenditure, share of GDP







share of net income of total income







barriers to trade and investment







market regulation







share on pension fund on gdp







trade union density







labour productivity (GDP per work hour)







dummy federalism-unitarism








The second model has a very high explanatory value (R² up to 91.3 %), but the outcome of the Durbin-Watson test indicates an autocorrelation between the variables. However, the tolerance and variance tests do not indicate autocorrelation between the independent variables in the model. As mentioned above, however, it is not the intention of this paper to explain the phenomenon of world cities, but rather to identify variables that influence their relevance and spatial distribution. The final model (stage 7) shows two dominant variables: high labour productivity and social disparities (measured by the Gini coefficient) seem to have a high explanatory value for integration into the world city network. The other five variables have a comparatively weak impact: the dummy variable has a positive impact on the dependent variable, which affirms the correlation between the number of global cities and the strength of connectivity (fig. 1). Market regulation and trade union density also have a weak impact on the model, but they both show a negative prefix, which is expected from the theoretical assumption: a liberal, deregulated economy with a poorly organised labour force has a strong explanatory value for the strength of a country’s integration into the global city network. The same is true for the share of pension fund within the GDP, which also has a very weak explanatory value. The tax rate, an indicator for the tax charge and the size of public sector, has a very small (put positive) impact, but is only significant on the 5% level.

Table 4: Multi-linear regression model of macroeconomic arrangement

stand. Beta-Coeff.








market regulation








Gini coefficient after social transfers







labour productivity (GDP per work hour)











trade union density




share of pension fund within GDP



ln tax rate of GDP









Durbin-Watson 0.787 Tolerance and Variance Test negative

dependent variable: gross connectivity

(Source: GaWC database 2008; databases: UNCTAD, UN, COW, WHO, ILO, UNIDO)

In sum, these variables have high explanatory value and confirm the theoretical assumptions formulated above: liberal, deregulated economic policy, a society with high social disparities and weak institutions for labour representatives have a positive impact on the formation of world cities.

Dimension 3: Territorial Structure and the Urban System

How does the spatial structure of a country influence its integration into the world city network? The independent variables considered in our model can be differentiated into two groups (tab. 5): first, those that indicate the spatial structure of the country’s urban system. Apart from general indicators like the urbanisation rate, regional disparities and their change, the degree of spatial concentration is here in focus: the concentration of the three largest cities on the urban and the total population, and the concentration of large cities (more than 0.75 million inhabitants) within the total population of a country. The second group of indicators is related to the territorial organisation of statehood: the share of sub-national expenditures on total public expenditures and its change between 2001 and 2011. Additionally, the share of local tax income, which indicates the fiscal autonomy and the degree of federalism, is considered here; furthermore, the transfer between governmental authorities and the dummy variable of federalism-unitarism is integrated into the model.

Table 5: Descriptive Analysis of independent variables model 3








share of local tax income on GDP 2010







Sub-national public expenditures 2011







change of sub-national public expenditures 2001-2011







intergovernmental transfers on public expenditure







dummy federalism-unitarism







regional disparities. GDP per head 2009







change of reg. disparities. GDP per head 2005-2009







share of urban population 2010







concentration of top 3 cities within urban population







concentration of top 3 cities within total population







concentration of large cities (>0.75 mill.) of urban pop.








The third multi-linear regression model has a high impact on the explanation of the dependent variable, the connectivity per country (R² 80.1 %). It considers each three variables that are related to the territorial organisation of the state and the country’s spatial and urban system. In the first three stages, the variables that are related to the territorial organisation are integrated into the model; together, they explain approximately 39.2 % of the dependent variable. However, by integrating the variables that are related to the urban system, their explanatory value rises gradually.

In the final model, the concentration of the three largest cities has the strongest impact; the negative prefix indicates that decentralised urban systems positively influence the global connectivity of a country. On the contrary, the concentration of large cities (> 0.75 million inhabitants) has a positive contribution, which underlines the size-effect for cities in their function as a world city as well as the size effect of a country. Compared to these two indicators, regional disparities have a weak impact, but their negative prefix constrains the relevance of a balanced urban and regional system for the world city integration of a country. The variables of territorial organisation of statehood, representing the two dimensions federalism and decentralism (see Ehlert et al., 2007), have a comparably low explanatory value; here, sub-national expenditures indicate the degree of decentralism, which seems to have a positive impact, which is affirmed by the dummy variable regionalism. In contrast to this, the second dimension of territorial organisation, the local tax income, which is more an expression of a federal system and is linked with the right to create and levy taxes on the sub-national level, seems to have a negative impact on the formation of a country’s global city-ness.

Table 6: Multi-linear regression model the territorial structure and urban system

stand. Beta-Coeff.









dummy regionalisation






log sub-national expenditure 2011








share of local tax income in GDP 2010







log concentration of top 3 cities on country pop.






concentration of large cities (>0.75 mill.) on urban pop.




regional disparities, GDP per head 2009











Durbin-Watson 1.729 Tolerance and Variance Test negative

dependent variable: total gross connectivity

(Source: GaWC database 2008, databases: UNCTAD, UN, COW, WHO, ILO, UNIDO)

The third model allows a perhaps surprising assertion: large cities in a national urban system with low hierarchies and disparities as well as a decentrally organised territorial structure have a positive impact on the formation of a country’s world city-ness. This outcome reflects a specific size-dependency of the dependent variable, as all other variables are size independent.

Discussion and conclusion

The three models presented here focus on three different aspects of the state system. The first focused on geopolitical power; although the market size and wealth level of states are considered in the variables at least indirectly, it can be concluded that geopolitical power determines the formation of a country’s world city integration. In other words, New York would not be New York if the USA were not a dominant actor on the global scale. Market size, military expenditure and several soft power indicators were identified as relevant explanatory variables. The second model focused on the macroeconomic arrangement of states; the outcome is not so surprising, as liberalised economies with weak welfare institutions, high social disparities and weak interest group for the labour force have an positive impact on a country’s global city-ness. However, the analysis supports the assumption of Hamnett (1996), which states that the world city does not produce social inequality, but rather that socially unequal, liberal societies might facilitate the formation of a world city. The third model focused on the role of the territorial organisation of states and the structure of urban system and regional disparities. As a surprising outcome, decentralised states with low regional disparities, a low concentration-degree in the urban system, but the existence of large cities have an positive impact on the global city formation. A comparison between Germany and France makes this outcome comprehensible: centrally organised, spatially unbalanced France is characterised by a dominant global city, but decentralised Germany with a huge numbers of global cities shows – in sum – a higher total connectivity.

The models presented in this paper estimated the impact of state-related variables on global city formation on different scales (model 1: 106 countries, model 2 and 3: 34 OECD countries), but did not allow for examination of the relation between the impact of the global-relational system and the territorial state system on world city formation. The analysis of residuals might help to throw little light on this question. The world map in fig. 2 shows countries whose total connectivity per inhabitant was over- or underestimated in the first model (geopolitical power). In the blue coloured countries, the model does not adequately explain the dependent variable; in these groups, where we find the most important countries in the global city network, the state system has insufficient explanatory value. On the contrary, in the red and yellow coloured countries, the state system has sufficient explanatory value and might expect a stronger input of the state on the world city formation.

Figure 2: Standardized Residuals – over and underestimation of GaWC Connectivity per inhabitant in the model of geopolitical power

(Source: GaWC database 2008)

The same evaluation of residuals has been carried out for the other two models and cross-tabulated in fig. 3. In this graph, we can differentiate between four sections where the states in section one of both models are overestimated (total cross-connectivity per country); this is true for e.g. Sweden, Portugal and the US; the macroeconomic arrangement as well as the territorial structure is overestimated, which can be interpreted as the strong impact of these dimension of statehood on the connectivity. On the contrary, Finland is underestimated in both models. Thus, the two models are not able to explain the dependent variable sufficiently. In field 4, e.g. Korea, Greece and Poland are overestimated in the macroeconomic arrangement, while the territorial structure has only little explanatory value for the countries’ connectivity.

Figure 3: Standardized Residuals – over and underestimation of GaWC Connectivity in the model of macroeconomic arrangement and territorial structure

(Source: GaWC database 2008)

It is not the intention of this paper to expose the world city system as a product of the state, which would be a fall-back into methodological nationalism (Brenner, 2004, p. 38). However, this paper wants to argue that the world city is the product of two processes: on the one side, de-territorialisation processes, which are carried out by the global financial sector, multinational enterprises, global producer service firms and the technical revolution in transport, communication and the information sector. On the other side – which has not yet been the focus of research – the system of states influences the formation of world city-ness; in contrast to the first, the state-related processes date back to before the era of globalisation and are determined by structures that follow a historic – at the least a fordistic – logic. The aim of this paper was to give insights into the relation of these two processes and the role of the state system.


Amable, B. (2003). The diversity of modern capitalism. Oxford: Oxford University Press.

Angew, J. (1994). The territorial trap: the geographical assumptions of international relations theory. Review of International Political Economy, 1, 1, 53-80.

Becker, J. (2007). Akkumulation, Regulation, Territorium. Zur kritischen Rekonstruktion der französischen Regulationtheorie. Marburg: Metropolis.

Brenner, N. (2004). New state spaces. Urban Governance and the rescaling of statehood. Oxford: Oxford University Press.

Castells, M. (1989). The informational city. Information technology, economic restructuring and the urban-regional process. Oxford: Basil Blackwell.

Derudder, B., Taylor, P. J., Ni, P., De Vos, A., Hoyler, M., Hanssens, H., Bassens, D., Huang, J., Witlox, F., Shen, W. & Yang, X. (2010). Pathways of Change : Shifting Connectivities in the World City Network, 2000-2008. Urban Studies, 47, 9, 1861-1877.

Dieleman, F. & Hamnett, C. (1994). Globalisation, regulation and the urban system. Urban Studies, 31, 357-364.

Ehlert, N., Hennl, A. & Kaiser, A. (2007). Föderalismus, Dezentralisierung und Performanz. Eine makroquantitative Analyse der Leistungsfähigkeit territorialer Politikorganisation in entwickelten Demokratien, Politische Vierteljahresschrift, 48, 243-268.

Esping-Andersen, G. (1990). The three worlds of welfare capitalism. Cambridge: Polity Press.

Ezcurra, P., & Pascual, P. (2008): Fiscal decentralization and regional disparities: evidence from several European Union countries. Environment and Planning A, 40, 1185-1206.

Fainstein, S. (2001). Inequality in global city-regions. In A. A. Scott (Ed.), Global City-Regions. Trends, Theory, Policy (285-298). Oxford: Oxford University Press.

Flora, P. (2000). Stein Rokkan: Staat, Nation und Demokratie in Europa. Frankfurt am Main: Suhrkamp.

Friedmann, J., & Wolff, Goetz (1982). World city formation: an agenda for research and action. International Journal of Urban and Regional Research, 6, 3, 309-344.

Friedmann, J. (1986). The world city hypothesis. Development and Change, 17, 1, 69-84.

Hall, P. A., & Soskice, D. (Eds.) (2001). Varieties of Capitalism. The Institutional Foundations of Comparative Advantage. Oxford: Oxford University Press.

Hamnett, C. (1996). Why Sassen is wrong: a response to Burgers. Urban Studies, 33, 1, 107-110.

Hamnett, C. (2003). Unequal City. London and the global arena. London: Routledge.

Hancké, B. (Ed.). (2009): Debating Varieties of Capitalism. A Reader. Oxford: Oxford University Press.

Häussermann, H. & Roost, F. (2000). Globalisierung, Global City. In H. Häussermann (Ed.), Großstadt. Soziologische Stichworte (pp. 79-91). Opladen: Leske&Budrich.

Häussermann, H. & Haila, A. (2004). The European City: A Conceputal Framework and Normative Project. In Y. Kazepov (Ed.), Cities in Europe. Changing Contexts, Local Arrangements, and the Challenge to Urban Cohesion (pp. 43-64). Oxford: Blackwell.

Hohenberg, P., & Lees, L. H. (1995). The Making of Urban Europe 1000-1994. Cambridge: Harvard University Press.

Jessop, B. (1997). Die Zukunft des Nationalstaats – Erosion oder Reorganisation? Grundsätzliche Überlegungen zu Westeuropa. In S. Becker, T. Sablowski, & W. Schumm (Eds.), Jenseits der Nationalökonomie? Weltwirtschaft und Nationalstaat zwischen Globalisierung und Regionalisierung (p. 50-96). Ort: Verlag.

Kaiser, A. & Ehlert, N. (2006). How and Why Do Political Institutions matter? Federalism, Decentralisation and Macro-Economic Performance in OECD Countries. In: Max Planck Summer Conference on Economic Sociology and Political Economy, 34.

Kaufmann, F.-X. (2003). Der deutsche Sozialstaat im internationalen Vergleich. Frankfurt am Main: Suhrkamp.

Keil, R., & Brenner, N. (2003). Globalisierung, Stadt und Politik. In A. Scharenberg, & O. Schmitke (Eds.), Das Ende der Politik? Globalisierung und der Strukturwandel des Politischen (p. 254-276). Münster: Westfälisches Dampfboot.

Letho, J. (2000). Different cities in different welfare states. In: A. Bragnasco & P. Le Gales (Eds.), Cities in Contemporary Europe (pp. 112-130). Cambridge: Cambridge University Press.

Lipietz, A. (1998). Das Nationale und das Regionale. Wie viel Autonomie bleibt angesichts der globalen Krise des Kapitalismus? In A. Lipietz (Ed.), Nach dem Ende des „Goldenen Zeitalters“. Regulation und Transformation kapitalistischer Gesellschaften (pp. 155-184). Berlin, Hamburg.

Ma, X., & Timberlake, M. (2012). World city typologies and national city system deterritorialisation: USA, china and japan. Urban Studies, 2012, 1-21.

Matznetter, W. & Musil, R. (2012), The European City in the Age of Globalisation, Belgeo, 1-2, 2012. (

Nye, J. S. (2005). Soft Power: The Means to Success in World Politics. New York: Public Affairs.

Pain, K. (2011). Cities in Transformation – What Role for the State? GaWC Research Bulletin 384 (1st March 2013;

Peck, J. (2000). Discussant’s commentary on the “Annual Conference of the Institute of British Geographers in Brighton” (Panel on Economies and Politics of Scale).

Priebs, A. (2004). Funktionalräume versus Territorien – Möglichkeiten einer Anpassung politischer Entscheidungsstrukturen an den Regionalisierungsprozess. In W. Gamerith, P. Messerli, P. Meusburger & H. Wanner (Eds.), Alpenwelt – Gebirgswelten. Inseln, Brücken, Grenzen. Tagungsbericht und wissenschaftliche Abhandlungen (pp. 697-706). Bern: 54. Deutscher Geographentag, Bern 2003.

Rodriguez-Pose, A. & Ezcurra, R. (2009). Does Decentralization Matter for Regional Disparities? A Cross-Country Analysis. SERC-Discussion, Vol. 25. London: Spatial Economics Research Centre.

Sassen, S. (1991). The Global City: New York, London, Tokyo. Princeton: Princeton University Press.

Sassen, S. (2006). Territory, Authority, Rights. From medieval to global assemblages. Princeton/Oxford: Princeton University Press.

Schmid, C. (1996). Headquarter Economy und territorialer Kompromiss. Zeitschrift für Wirtschaftsgeographie, 1/2, 28-43.

Schratzenstaller, M. (2005). Neuer Finanzausgleich und Stabilitätspakt – keine grundlegenden Änderungen, WIFO-Monatsberichte, 1/2005, 49-60.

Shankar, R. & Shah A. (2003). Bridging the economic divide within countries – a scorecard on the performance of regional policies in reducing regional income disparities, World Development, 31, 1421-1440.

Taylor, P. J. (1995). World cities and territorial states: the rise and fall of their mutuality. P. L. Knox & P. J. Taylor (Eds.), World cities in a world-system (pp. 48-62). Cambridge: Cambridge University Press.

Taylor, P. J. (2000). World cities and territorial states under conditions of contemporary globalization. Political Geography, 19, 5-32.

Taylor, P. J., Catalano, G. & Gane, N. (2003). A Geography of Global Change: Cities and Services, 2000-2001. Urban Geography, 24, 431-441.

Taylor, P. J. (2004). Is there a Europe of Cities ? World Cities and the Limitations of Geographical Scale Analysis. In E. Sheppard, & R. B. Master (Eds.), Scale and Geographic Inquiry : Nature, Society, and Method (pp. 213-235). Oxford: Blackwell.

Taylor, P. J., & Derudder, B. (2004). Porous Europe : European Cities in Global Urban Arenas, Tijdschrift voor Economische en Sociale Geografie, 95, 5, 527-538.

Taylor, P. J., Derudder, B., Hoyler, M., Pain, K. & Witlox, F. (2011). European Cities in Globalization. In P. J. Taylor, P. Ni, B. Derudder, M. Hoyler, J. Huang & F. Witlox (Eds.), Global Urban Analysis. A Survey of Cities in Globalization (pp. 114-136). London, Washington: earthscan.

Taylor, P. J. (2011). Global Synthesis: National and Sub-Regional Contrasts. In P. J. Taylor, P. Ni, B. Derudder, M. Hoyler, J. Huang & F. Witlox (Eds.), Global Urban Analysis. A Survey of Cities in Globalization (pp. 331-335). London, Washington: earthscan.

Zehner, K. (2011). Von der Hauptstadt des Empire zur Alpha Global City. Die Stadtentwicklung Londons im Zeichen von Globalisierung und Stadtentwicklungspolitik. In W. Matznetter, & R. Musil (Eds.), Europa: Metropolen im Wandel (pp. 39-51). Wien: Mandelbaum Verlag.

Data sources

International Labour Organization, Social Security Department (

United Nations, Department of Economic and Social Affairs, World Urbanization Prospects, the 2011 Revision (

United Nations, World Health Organization, Global Health Observatory Data Repository (

World Bank Database (

United Nations Conference on Trade and Development (

University of Gothenburg, The Quality of Government Institute, QoG-Database (

OECD, Fiscal Decentralism Database (

OECD, Database (


* Robert Musil, Institute for Urban and Regional Research, Austrian Academy of Sciences, E-mail:

1. Mini-states without global cities are not included.

2. Globalization and World Cities (GaWC) Research Network (, unpublished data 2008; Taylor et al. (eds)(2011).

Edited and posted on the web on 19th March 2013