GaWC Research Bulletin 465

(Z)

Urban Systems

M. van Meeteren^*

Abstract

Urban systems research utilizes the language of systems theory to grasp the complexity of the urban and the city. This contribution explains the basic principles of urban systems theory and outlines some important pitfalls. Depending on research focus and priority, urban systems have been bracketed geographically and conceptually in different ways in different time periods. In the past, defining the city as a bounded entity was uncontroversial and the difference between inter-urban and intra-urban processes seemed self-evident. As urbanization intensifies urban boundaries blur, prompting to continuously scrutinize the scales of urban systems. The usefulness of urban systems thinking is illustrated through an appreciation of the historical urban systems models of American geographer Allan Pred from the 1960s and 1970s. The work of Pred is contextualized in the present by briefly comparing it in terms of continuity and change with world city analysis, an important contemporary application of urban systems analysis.

 

Introduction

At first glance, land utilization in an urban area such as New York and its environs appears to be without rhyme or reason, a confused and baffling welter of anomalies and paradoxes. The land is being used, most of it, very intensively indeed. Nine million people eat and sleep, work and play in the area. But the assignment of the land to the various uses seems to the superficial observer to have been made by the Mad Hatter at Alice's tea party.

Haig (1926, 403)

Descriptions of the city and the urban as a chaotic, anarchic, and disorganized locus of activity abound in novels, movies, records, and even the academic literature, as the above quote by the pioneering urban systems researcher Robert Murray Haig testifies. Urbanists have long been baffled by the creative potential, the economic activity, but also the alienation, poverty and neglect that urban environments seem to foster. It was in the 1920s and 1930s, when the modern city became salient, that a desire to unravel this chaos led to the development of urban systems theory. Walter Christaller, a pioneer of central place theory (Christaller 1933 [1966]), discovered that if you focused on the ways in which people procured their goods and services, a distinctive, 'systemic', pattern of central places can be discerned. The (ideal) central place system takes the form of neat geometrical, i.e. hexagonal, spacing of places and makes this particular urban function legible.

Such orderly and rational conceptualizations of cities and the urban, which are at odds with descriptions emphasizing the chaotic and anarchic, can give a reductionist impression. Nevertheless, they help us understand a facet of the urban which we did not understand before. Urban systems research aspires to make the complexity of the city and inter-city interactions understandable by isolating some of its constituent social processes, and then relating these processes to others occurring both inside the city and between cities and the outside world. For instance, by decomposing the chaos of urban traffic in a structure of origins and destinations, it may suddenly appear as a logical consequence of many individual choices. By isolating the central place dynamic, its influence on other aspects of urbanization can be examined. Through relating flows of information, goods and money between cities, thick knots of inter-city interactions all of a sudden start to make sense.

While the apparent chaos of the city already felt overwhelming to many observers in the nineteenth and twentieth centuries, urban complexity has only become greater since. In the twenty-first century, the urban condition has become so dominant that an increasing number of scholars are suggesting that we should not look at the urban through the bounded imaginary of a city anymore, but to consider it a planetary phenomenon (Brenner and Schmid 2014). The planetary urbanization argument has merit from both analytical and rhetorical angles, as indeed, urban society has an influence in nearly any nook and cranny of the earth's surface. Nevertheless, such a wide-angled position also has disadvantages. The boundedness conveyed through the notion of the city not only provides an important empirical demarcation when studying individual places: it also allows for comparing different places and by extension, for understanding how several places together constitute a bigger whole. Urban systems researchers do not only study "the city as a system" through, for instance, modeling urban traffic, land use patterns, housing, schooling and care needs within the city. They also study the "city in a system of cities", focusing on the inter-urban geographies of flows of people, information and goods between cities and how these flows impact the economies and societies in the city system's constituent individual cities. Although it is less and less correct to see the city as a self-contained entity with specific properties on the inside that are absent on the outside, social phenomena still tend to exhibit geographically bounded thresholds and boundaries. Urban systems research, which describes and models these urban phenomena, can provide continued relevance to the notion of 'the city' as a settlement-based analytical phenomenon, albeit with variable boundaries.

Trying to make sense of the urban chaos in this way can feel somewhat disenchanting, as it seems to reduce the fascinating object that is the city to a conflation of, for instance, traffic, land values, economic activities, amenities and social reproduction. Additionally, the act of simplifying complex urban realities in a system or model carries the risk of reification, or the assumption that an abstract category unproblematically corresponds with realities on the ground. Such reduction can make urban planners overconfident since it suggests the pretense of control over something as big and complex as the city.

Despite these drawbacks, having analytical tools that can illuminate the phenomenon of the city, even if only partially, is better than being engulfed in darkness. Being able to estimate the (future) needs of a population in terms of housing, schooling, shopping and mobility is useful. Knowing how cities together form larger wholes through their interactions can provide important information on how the global economy is spun over the face of the earth and how planning might contribute to slightly changing the ways in which this global economy is construed. This paper therefore makes the case that the urban systems approach continues to have merit. The paper proceeds as follows. It first outlines the basic categories of the systems approach and examines how these map onto the urban realm. How urban systems research needs to deal with the complexities of scale and function is considered thereafter. Urban systems theory is subsequently illustrated with the work of Allan Pred and world city research, before conclusions are drawn on its contemporary relevance.  

A Systems Perspective

Textbook descriptions on what a "system" or a "systems perspective" is tend to be simultaneously remarkably abstract and vague and yet deceivingly simple. For instance, Huggett (1980, 1) describes a system as a "set of interrelated parts" that consists of three basic components—elements, states and relations—which can either be material or conceptual. Systems research tries to isolate these components (elements, states and relations) in order to grasp the system as a totality. Most system descriptions have an "outside" as well: the environment. This environment can influence the elements, states, and relations of the system under study, but the environment is not specified in the same precise language as the system and its components. When the environment exchanges matter and energy with the system, as all social systems do, the system is considered to be open. Consequently, any description of the system and its constituent elements, states, and relations needs to be regarded as provisionary, for unexpected interactions can make the actual working of the system different than expected.

The language of systems theory contains a large variety of abstract concepts to describe and qualify a system’s relations, elements, and states. Typical terms associated with systems theory are feedback loops, inputs, outputs, resilience, bifurcation, and entropy. A detailed discussion of all these terms is beyond the scope of this contribution: the key point here is that they allow for the description and comparison of concrete phenomena through abstractions in terms of systems. Such abstraction, in turn, provides a framework to compare entities that refer to different concrete phenomena. For instance, while traffic, social networks, trade, and email interactions refer to rather different concrete social formations, they can all be abstracted and compared as different instantiations of "urban interaction systems".

The notion of systems theory as a mode of abstract description underlines the importance of understanding systems theory as a particular language that describes the world, or even more elaborately, as a specific way of seeing (Angelo 2016). The constructivist conception of systems theory advocated in this contribution builds on the seminal systems formulation of Blaut (1962), which has the advantage of theoretically bridging seemingly incompatible philosophical positions in geography (Van Meeteren 2016). When conceptualizing an urban system, some aspects of social reality are named, defined and demarcated, while others disappear from view. It is possible and even likely that those elements that are currently coded out of view in a system's abstraction, and thus rendered part of the environment, might later become politically relevant or controversial, prompting the reformulation of the system. As with any language, a systems perspective is adequate for describing some social phenomena, but less suitable for describing others, for which a different social-scientific language might be more appropriate. Good candidates for a systems perspective are generally urban phenomena where it is relatively easy to analytically separate the constituent elements, states and relations—i.e. they contain subsystems which show a relative autonomy. In particular traffic systems, trading systems and distribution systems have traditionally been fruitful applications of urban systems theory.

It is no coincidence that when urban systems theory was at its most popular in urban studies, namely in the 1960s, these were exactly the kinds of research questions that social scientists were occupied with. Urban systems theory fitted the 1960s 'spatial science' era, which was all about properly guiding the mid-twentieth century modernist and developmentalist urbanization challenges. When the focus of urban research changed toward issues of equity, social justice, and culture from the 1970s onward, urban systems reasoning became more controversial, precisely because these were the topics that the language of systems theory was less attuned to. Although the toolbox and language of urban studies is nowadays much more diverse than it was in the 1960s, the urban systems perspective remains a useful language, complemented by others, to explain those relatively predictable aspects of the urban which continue to exist today.

The above caveats all point to the importance of carefully defining and naming the elements, states and relations of an urban systems model. This phase of conceptualization and demarcating is called "entitation" in systems research. According to Huggett (1980, 29) entitation is the most important yet an often overlooked part of systems research. Making an entitation of an urban system requires the following steps: one has to define the process that is to be understood; to identify which elements are important to understanding that process; and to examine what the (hypothesized) relations between these elements are. Moreover, it has to be assessed whether a part of the defined system has sufficient functional, "systemic", autonomy to merit defining it as a subsystem. Subsequently, feedback loops between systems, subsystems and the environment, where change in the state of an element prompts further change, can be theorized and specified.

One of the strengths of systems analysis is that the hypothesized elements and relationships, although they represent complex interactions in social reality, can be depicted in systems diagrams. Figure 1 shows an example of a simple migration system as it was hypothesized in the 1960s/early 1970s, when urban-rural boundaries were more self-evident than today. However, the simplicity of such a diagram can be deceiving: it might in fact concern highly controversial relationships between theoretical phenomena. For instance, the abstract categories of "urban and rural adjustment mechanisms" play an important role in Figure 1. However, what an "urban adjustment mechanism" is and if it merits an unambiguous and uncontroversial application is far from self-evident. Constructing a logical model of the world in theory and depicting it in a clear and lucid diagram provides no guarantees that the world itself adheres to that logic. After constructing a systems model, the subsequent step is empirical systems analysis, when the ambition is to test whether the hypothesized system has empirical validity. Such testing requires operationalizing the defined components of the system by connecting it to empirical referents. This is the likely moment where the controversies surrounding "urban adjustment mechanisms" will become salient, and another family of methodological problems comes into play. These regard the question of how to deal with (internal) validity, as it is important to assert a degree of scientifically sound correspondence between what is measured and what is theorized. However, discussing these validity issues is outside the purview of this essay.

Figure 1. "A system schema for a theory of rural-urban migration". Redrawn from Mabogunje (170, 3)

A Multiplex, Multiscalar Constellation of Urban (Sub)Systems

A crucial aspect of entitation in urban systems analysis is defining the "systemness" of the phenomenon under study, which might or might not be spatially articulated. However, if the urban system is described in terms of its geography, such geographical articulation, for instance sensitivity to distance or density, becomes important. Some urban systems work at a very small scale. An example of such small-scale urban systemness is the innovative local buzz that is associated with knowledge transfer within cities. This buzz tends to be limited to the science or office park where the knowledge is discussed. Other urban systems, such as the urban labor market with its degree of specialization, might be restricted to the commuting zone, which corresponds with the daily urban system of a particular place (e.g. van der Laan 1998). A central place system (Christaller 1933 [1966]) supersedes the daily urban system and therefore pertains to a larger regional scale, as many specialized central functions are procured on a less than daily basis. A global city region (Scott 2001) such as the Pearl River Delta or the San Francisco Bay area might have its own systematic division of labor based on, for instance, inter-firm interactions that yet defines an even larger scale. Thus, the city or metropolis obtains a difference size and scale depending on the phenomenon under study.

A systems perspective entails an understanding of the urban in functional terms: it is the functional geography, instead of the morphological or political geography, of the phenomenon theorized in systems terms that defines the analytical geographical boundaries, although these boundaries tend to become fuzzy once they are operationalized empirically: exceptions and outliers are to be expected. This functional perspective implies that the geographical outside, i.e. the city/environment or inter-city system's boundary, will differ from urban function to urban function. Geographies of labor markets, transport systems, retail systems, institutional boundaries, daily activity spaces, and social communities will all have different geographies and hence different boundaries. The concrete empirical object we call "city" is in fact an intersection of many different interrelated and imbricated urban systems working on different scales, from the neighborhood to the global scale. Urban systems are inherently multiplex (Burger et al. 2014), and complicate the commonsensical notions of the city as a clearly bounded unit. It is because of this multiplex reality that urban systems research benefits from taking into account different scales and their interactions simultaneously (Burger et al. 2016; Taylor and Derudder 2016; Van Meeteren 2016).

There is a long tradition in urban systems research that operationalizes multiscalar analysis in a nested fashion, by studying "cities as systems in a system of cities" (Berry 1964). Although geographical scales are not necessarily nested and can overlap and intermingle in far more complex ways, this nested conception has the big advantage that it allows utilizing (social) network analysis to analyze urban systems. When we can safely assume that the city is a bounded unit, or a node, for analytical reasons, we can examine the interactions between cities more easily through the lens of "urban networks" (Van Meeteren et al. 2016). This network conceptualization, in turn, allows for detailed analysis of the division of labor and the flows of goods, people and information between places.

An Example of Urban Systems Analysis: Allan Pred's Historical Models of the US Urban System

To illustrate the usefulness of an urban systems approach, this section will discuss the work of geographer Allan Pred (1936-2007) on US urban systems. In a series of articles and books published in the 1960s and 1970s (e.g. Pred, 1965, 1973, 1977), Pred tried to grasp the genesis, historical development and functioning of the system of cities in the United States. Pred did not concern himself too much with the boundary problem of city demarcation discussed above, and used a definition of city systems that did not problematize the fuzziness of the intra-urban/inter-urban distinction. For Pred a system of cities is a:

national or regional set of cities which are interdependent in such a way that any significant change in the economic activities, occupational structure, total income or population of one member city will directly or indirectly bring about some modification in the economic activities, occupational structure, total income or population of one or more other set members. (Pred 1977, 13)

By analyzing the evolution of communication systems, trade flows, occupations, and organizational networks in the United States, Pred was able to understand the respective contributions and benefits of single cities contributing to that larger urban system.

An important aspect of Pred's work is that he contextualized his urban systems models historically (see also Johnston 1984), allowing him to identify path dependencies of urbanization, where historically grown advantages impact new rounds of urbanization in the future. Additionally, he had a keen theoretical and empirical interest in understanding how the process of urbanization changes through time. For each different era of urbanization in the United States, such as the "mercantile era", the "early industrial era", and the "late industrial era", Pred's urban systems model was adapted to cope with the technological advance that had emerged between the eras.

Figure 2 reproduces a subsystem of Allan Pred's urban system's model, which was first introduced in an article in Geographical Review in 1965 (Pred 1965, 165). The mechanism depicted in Figure 2 featured as part of his more elaborate models that describe specific epochs (e.g. Pred 1973, 192; Pred 1977, 90, 175), where it had the role of a subsystem amid other urbanization dynamics such as real estate development, trade, and profit repatriation through the networks of corporations.

The central process that the urban systems model in Figure 2 describes are the economic-geographic consequences of a new or enlarged industry on an urban economy. The driving force of this process is "circular and cumulative causation", theorized by Swedish economist Gunnar Myrdal (1957), which refers to a feedback effect: once an urban economy is growing, it is likely to start a self-reinforcing growth spiral, increasing a city's dominance over the rest of the inter-urban system. Much urban-economic research, from the 1960s all the way to the present, can be characterized as a search to identify mechanisms that trigger, change and stop this self-reinforcing feedback loop of urban growth.

Figure 2. "The circular and cumulative process of industrialization and urban-size growth". Redrawn from Pred (1965, 165)

The relations depicted in Figure 2 describe two cyclical feedback mechanisms that intend to explain the process of circular and cumulative causation. The first cyclical feedback mechanism works through the multiplier effect. This multiplier effect results from new factories and their laborers, who create new local demand for amenities, transportation, construction, and other services, leading to increased urbanization. As new thresholds are met in the city, new amenities and industries become economically viable, which in turn create new possibilities for an individual city to proliferate itself in the wider urban system. The second feedback loop depicts the transformative potential of the urban system through innovation and knowledge circulation. Pred was a pioneer in defining this loop, which has only become more important for urban and regional economic geographers since. As a city becomes a more diverse and bustling place, the chance that new inventions are developed increases. These new inventions become innovations once they find an economically viable use, which becomes again more plausible when local demand is high. Demand, in turn, is positively influenced by a higher urban critical mass. Together with multiplier effects that might make new activities economically viable, innovations can stimulate new or enlarged industries, which again drive the circular and cumulative process of urbanization. A later version of the model (depicted in Pred 1973, 192) also spells out that, through demarcation of an additional subsystem, this circular and cumulative growth of the economic base of cities may enhance the accumulation of local earnings and capital, which subsequently can drive real estate and construction booms. This enhanced accumulation (Pred 1977, 90) can give rise to a secondary multiplier, which effectively creates a third feedback loop of circular and cumulative causation. 

Central in Pred's models of urban systems is the notion that cities are part of larger trading, industrial, and corporate networks. It is the division of labor between cities in these networks that explains the development of the urban system as a whole, impinging on the constituent cities. Crucially, the relative importance of these respective networks changes over time. For instance, Pred’s mercantile model (Pred 1972, 192), valid for urban development in the USA between 1790 and 1840 when urban industrial activity was still limited, explains urban systems development in the USA primarily through the position of cities in countrywide trading networks. Models about later periods progressively include the influence of industrial employment and local demand. As technology made inter-city interactions progressively easier, the relative weight of the components in Pred's urban systems models gradually changed. With the passing of time, Pred considered the geography of information, knowledge, and innovation diffusion throughout the city system increasingly important compared to the emphasis on the production and trade of tangible goods that are center stage in his models about earlier rounds of development of the US economy.

The latest model of urban systems that Pred (1977) elaborated concerns "advanced economies". The notion of “advanced” refers to the state of the economy in the mid to late twentieth century, when American monopoly capitalism was at its zenith, and the recent round of globalization had not yet taken off. The influence of US multilocational corporations (e.g. General Motors) is central to this analysis. Pred's advanced economy model emphasizes the "parasitic" potential of inter-city interactions. Profits can seep away through the networks of multi-branch corporations that will decrease the local multipliers associated with those profits. In return, new knowledge and technology might become available in the urban region that would not be available otherwise. While Pred made no decisive judgment to what extent these positive and negative feedback loops ought to be weighed against one another, his research does raise the question whether having better connections to the rest of the urban system is beneficial by default. In this sense, Pred's later urban systems models have similarities with colonial development models that explain global geographical exploitation in terms of center-periphery relations. World city theory (Taylor and Derudder, 2016) has been developed through the cross-fertilization between urban and world-systems research and shows many continuities with ideas already present in Pred's models.

Nevertheless, Pred's "advanced" economy urban system is already 40 years old and some assumptions of the model seem dated, and therefore inadequate to understand the present. While the multilocational corporations dominating the model are still very important in the contemporary economy, they have profoundly changed in organizational and geographic scope, prompting revision of Pred's model if used today. In many instances, the multilocational corporation has become a global transnational corporation, bringing new geographic scales into view. Additionally, the geography of knowledge diffusion throughout the networks of transnational corporations is likely to be influenced by the knowledge geographies and virtualization produced by the Internet. Although many scholars have argued for the continuing importance of localized interactions for knowledge diffusion (e.g. Storper 1997), a contemporary model would need to accommodate manymore contingencies related with 'network-type' organizations and looser organizational forms associated with knowledge transfer. Moreover, not all knowledge is so sticky that it can only be processed in the city, as many forms of specialized knowledge are shared in virtual communities.

Another time-bound aspect of Pred's models is that he uncritically accepted the notion of a 'national urban system', as did most authors at that time (Bourne and Simmons 1978). Cities (as a subsystem) are nested in the geography of a country (the system), which interacts with the rest of the world (the environment). As the world-economy has become much more integrated transnationally since the 1970s (Dicken 2011), such a notion of a national economy with its distinctive national urban system becomes analytically less and less useful for understanding today's urban systems. Moreover, the risks of a methodological nationalism that veils transnational interactions loom large (Taylor and Derudder 2016). However, even today, the national boundary expresses important institutional boundaries (e.g., tax regimes, legal systems, languages) that systematically influence urban systems. Thus, although a national perspective might be more limiting than it was in the 1970s, it should nevertheless not be neglected. Of course, there was also another reason why national boundaries were convenient for authors such as Pred: states produce national statistics that allow for empirical estimations and testing of urban systems models.

The World-City Network as an Inheritor of Pred's Urban Systems Model

Research on the world-city network through the Interlocking World City Network (IWCN) model (Taylor and Derudder 2016) is a contemporary version of an inter-city urban systems analysis that shows much commonality with Allan Pred's approach. Simultaneously, the IWCN model attempts to overcome some of the conceptual limits associated with Pred's nationally-focused urban systems analyses.

The core assumption of the IWCN urban systems model is that the global networks of Advanced Producer Service (APS) providers are a contemporary indicator of the information and material interactions that are central to the economic inter-city model pioneered by Pred. Through an empirical reconstruction of these firm networks outside the purview of state-based stat(e)istics, estimations on the division of labor between cities on a world scale can be derived, while circumventing the problem of methodological nationalism. However, recalibrating a systems model to take some identified shortcomings into account can result in losing analytical precision on other aspects of the system. For instance, Pred's models intended to comprehensively grasp the economic interaction between a focal city and the national system of cities that focal city is embedded in. This comprehensiveness is exemplified by his specification of the local versus the extra-local dimensions of the model and their respective influences on multiplier effects, leakage of profits, local unemployment and real estate development. In its ambition to circumvent the national bias and provide global coverage compared to older research such as Pred's, the IWCN model has had to compromise. The focus on APS enables data analysis on a global scale but does not allow definite inferences on APS relations with other sectors. Thus, the IWCN model attains geographical comprehensiveness at the expense of functional comprehensiveness.

Concluding Thoughts

Statistician George E.P Box (1979, p.2) famously remarked that "all models are wrong, but some models are useful". This statement definitely applies to urban systems and related models. Box argues that the question of a model of a phenomenon being "true" is largely moot: because any model is a simplification of social reality, it is necessarily not corresponding to that reality and thus, by definition, not "true" in the metaphysical sense.  However, what is important is whether a model of the world is able to adequately contribute to describing, analyzing, and perhaps even explaining the phenomena that somebody is interested in studying. Rightful criticism of the urban systems model can, like the criticism leveled at it in the 1970s, take the form of agenda-setting critique: a model might overlook a concern that is nevertheless politically, socially or ethically salient at that moment. To give an example, urban systems models that focus on describing the economic division of labor between cities will likely conclude that larger divisions of labor and denser transportation networks are beneficial for urban economies. Quite often, such findings lead local governments to want to increase their airline connections as much as possible out of fear of missing out on economic growth. However, it is highly questionable that an urban systems analysis that defines, measures and models the ecological consequences of all that activity would arrive at the same conclusions. Resultantly, if ecological considerations rise in importance, the urban systems models need to be adapted. Different eras and concerns require different (and equally wrong but useful) models. However, although what an urban systems model comprises will change through critiques, such criticism does not necessarily discredit the language of urban systems theory as a useful tool to make sense of the ever-increasing complexity of our urban world. As making sense of this complexity becomes ever more pertinent with the huge social challenges of contemporary urbanization, the concept might even see a renaissance in the coming decades. 

ACKNOWLEDGEMENT

This work builds upon research conducted at Ghent University between 2012 and 2016. The author would like to thank Ben Derudder and David Bassens for constructive feedback on earlier drafts.

REFERENCES AND FURTHER READING

Angelo, Hillary. 2016. "From the city lens toward urbanisation as a way of seeing: Country/city binaries on an urbanising planet." Urban Studies, DOI:0042098016629312.

Berry, Brian JL. 1964 "Cities as systems within systems of cities." Papers in Regional Science, 13: 147–163.

Blaut, James M 1962. "Object and relationship." The Professional Geographer, 14:1–7.

Bourne, Larry S and James W. Simmons, eds. 1978. "Systems of Cities: Readings on Structure, Growth, and Policy". New York and Oxford: Oxford University Press.

Box, George EP 1978. "Robustness in the strategy of scientific model building." Paper read at the Army Research Office Workshop on Robustness in Statistics held at Research Triangle Park, North Carolina on April 11-12, 1978.

Brenner, Neil and Schmid Christian 2014. "The “Urban Age” in Question." International Journal of Urban and Regional Research, 38: 731–755.

Burger, Martijn J., Meijers, Evert J., and Frank G. Van Oort 2014. "Multiple perspectives on functional coherence: Heterogeneity and multiplexity in the Randstad." Tijdschrift voor Economische en Sociale Geografie, 105: 444–464.

Christaller, Walter. 1966 [1933]. Central Places in Southern Germany. Eaglewood Cliffs, New Jersey: Prentice Hall.

Dicken, Peter. 2011. Global Shift (6th edn). New York: The Guilford Press.

Haig, Robert M. 1926. "Toward an understanding of the metropolis, no.2" The Quarterly Journal of Economics, 40: 402–434.

Huggett Richard. 1980. Systems Analysis in Geography. Oxford: Clarendon Press.

Johnston, Ronald J. 1984. City and Society: An Outline for Urban Geography, updated ed. London: Hutchinson.   

Mabogunje, Akin L. 1970. "Systems approach to a theory of rural‐urban Migration." Geographical Analysis, 2: 1–18.

Myrdal, Gunnar. 1957. Rich Lands and Poor, the Road to World Prosperity. New York and Evanston: Harper & Row.

Pred, Allan R. 1965. "Industrialization, initial Advantage, and American metropolitan growth. Geographical Review, 55: 158–185.

Pred, Allan R. 1973. Urban Growth and the Circulation of Information: The United States System of Cities 1790-1840. Cambridge MA: Harvard University Press.

Pred Allan R. 1977. City-Systems in Advanced Economies. London: Hutchinson.

Scott, Allen J. (ed). 2001. Global City-Regions: Trends, Theory, Policy. Oxford: Oxford University Press.

Storper, Michael. 1997. The Regional World. New York: The Guilford Press.

Taylor, Peter J and Ben Derudder. 2016. World City Network, 2nd ed. London and New York: Routledge.

Van der Laan, Lambert. 1998. "Changing urban systems: An empirical analysis at two spatial levels. Regional Studies 32: 235–247.

Van Meeteren, Michiel 2016. From Polycentricity to a Renovated Urban Systems Theory: Explaining Belgian Settlement Geographies. PhD Thesis: Ghent University.

Van Meeteren, Michiel, Zachary P.Neal and Ben Derudder. 2016. "Disentangling agglomeration and network externalities: A conceptual typology." Papers in Regional Science 95: 61–81.

NOTES

^* Michiel van Meeteren, Department of Geography, Vrije Universiteit Brussels (VUB), Belgium, email: michiel.van.meeteren@vub.ac.be