GaWC Research Bulletin 316

This Research Bulletin has been published in Urban Geography, 31 (7), (2010), 865-884.

doi:10.2747/0272-3638.31.7.865

Please refer to the published version when quoting the paper.

(Z)

Explosive City Growth in the Modern World-System: An Initial Inventory Derived from Urban Demographic Changes

P. Taylor^*, A. Firth^**, M. Hoyler^*** and D. Smith^****

Abstract

This is an empirical quantitative study that uses urban demographic data to identify economic change in cities. Unlike previous demographic studies we provide a strong theoretical framework for the analysis: the research is designed and interpreted through Jacobs' theory of the city as economic process within the framework of Wallerstein's modern world-system. This is the first time Jacobs' theory has been subject to systematic enquiry over big time and big space. We create an inventory of 184 examples of Jacobs' ‘explosive growth' from 1500 to 2005 within the modern world-system. These results are interpreted in terms of systemic hegemonic cycles with first Dutch, and then British, and finally US cities dominating the inventory lists. It is found that Dutch, British and US city spurts are front-loaded in their respective hegemonic cycles: this is strong evidence of hegemony being created and initially sustained through and by these dynamic cities. Thus the paper makes a direct contribution to developing world-systems analysis; in a short conclusion future research on ways of contributing to Jacobs' theory of the city are discussed.

Key words: dynamic cities, economic expansion, hegemonic cycles, modern world-system

Introduction: DEMOGRAPHIC CHANGES IN CITIES

At least since Adna Weber's (1899) seminal study The Growth of Cities in the Nineteenth Century scholars have been fascinated by the rise of great cities in modern society. Massive urbanization appears to be integral to the unfolding of modernity culminating today with the historically unique circumstance of the majority of humanity being urban dwellers. The story of how this came about has been told many times. What we offer in this paper is a new take on the process that advances the argument beyond the demographics to consider the underlying economic change that created modern urbanization.

The process of urbanization is an established topic in demographic and population geographical analyses (Lawton and Lee, 1989) but this concept is too general for this study. Here the focus is upon one element of urbanization, the extraordinary rapid growth that cities sometimes experience. Of course, we are not the first to appreciate the importance of these urban events but we will deal with them in a specific theoretical manner. Jan de Vries' (1984) European Urbanization, 1500-1800 provides a contrast to our approach. He carries out careful statistical analyses of urban growth to find basic sources of change and shows that ‘the ten fastest growers' contribute disproportionately to total urban growth (p. 139). He develops this argument by identifying a category of ‘fastest growers' and identifies 70 such cities across three periods within his time frame (p. 140).¹

We carry out a similar demographic identification of cities of exceptional growth, extended forward to the present, but the key difference is that we engage with geohistorical theories to enhance our understanding of these phenomena. First, a theory of cities is used to account for a critical ‘economic' sub-group of the ‘fastest growers', and second we set this finding within a structural interpretation of the unfolding of modernity from 1500 as the modern world-system.² We are not so much interested in the demography of urbanization but in the nature of cities: demographics is used as a tool to begin an understanding of the economic change that generated modernity through cities.

The specific purpose of this paper is to provide a geohistorical empirical description and analysis of the modern world-system as a mega-process led by dynamic city-economies. Jane Jacobs (1969, 1984, 2000) and Immanuel Wallerstein (1974, 1979, 1983, 1984a, 2004) are our theoretical guides in this endeavour. Strange bedfellows to be sure, but in the first section below we describe the similarities and differences in their theoretical approaches and explain the way in which we bring them together for the purposes of this study. Their work imbues this research: it starts by directing the data collection and analysis and ends by informing interpretation of the findings. There are two empirical sections sandwiched between these contributions, a section on methodology sets out the mechanics of this research, and a section on results describes basic aspects of our inventory of extraordinary city-economies in the modern world-system. Finally, in a brief conclusion, we indicate how this paper is part of a larger study that takes the research down new, but related, avenues within the same overall theoretical perspective.

LOCATING DYNAMIC CITIES IN THE MODERN WORLD-SYSTEM

This research is based upon a unique combining of the immensely influential ideas of two scholars: Immanuel Wallerstein and Jane Jacobs. Although in developing their ideas in writings spanning the last four decades they have hardly cited each other, they do share one fundamental position: their critical intellectual edge comes from challenging and undermining the priority conventional social science accords to nation-states as actors and arenas. But they diverge in their respective replacements of the ‘national' geographical scale: for Wallerstein, understanding social change is sought through study of the modern world-system so that the world-economy replaces national economies as his focus; and for Jacobs, it is dynamic city-regions that require study in order to understand social change so that in her case the city-economy replaces the national economy as focus. This scale divergence offers the prospect of treating them as complementary theoretical theses. However, it takes more than placing their scale preferences together to combine their arguments since their ideas build from two very different social research traditions.

The basic premise of Wallerstein's world-systems analysis is that we live in a modern world-system that had taken shape in Europe by the end of Braudel's (1982) ‘long sixteenth century', and which was global in scope by about 1900. Historically, this is the only capitalist world-economy to have survived into self-sustaining reproduction. And in achieving this, it has been phenomenally successful: the economic development of the modern world-system has been historically unprecedented. However, this expansion has not been even through either time or space and from this induction he derives the basic structures of the system.

1. Historically, economic cycles describe high growth phases followed by low growth/stagnation phases. High growth phases are periods of innovation leading to new and more complex economic processes. Wallerstein (1984a) focuses on longer cycles, specifically Kondratieff cycles (c.50 years in length) and the longer hegemonic cycles. It is the latter where he has made a distinctive contribution in identify three hegemonic cycles throughout the modern world-system: a Dutch cycle centred on the seventeenth century, a British cycle centred on the late eighteenth and nineteenth centuries, and an American cycle centred on the twentieth century. Crucially, these three ‘hegemonic states' are more than merely ‘great powers' because (i) they each led international coalitions that defeated major military threats (in the Thirty Years War (1618-48), the Revolutionary and Napoleonic Wars (1792-1815), and the German Wars (1914-45) to survival of the system; and (ii) they have been the locales where fundamental economic reconstructions of the world-economy were invented (mercantilism, industrialization and mass consumerism respectively).

2. Geographically, the basic structure is a core-periphery pattern wherein the core zone is the locale through which economic expansion is generated. The core of the world-economy is created by relatively high-tech, high wage economic processes which is the obverse of the periphery with its low tech, low wage situation. Thus the core consists of a complex economic structure that is able to dominate and create simple peripheral economic structure so as to service its voracious needs. Wallerstein's (1979) contribution has been to argue that these structures are not stable; there is a third semi-peripheral category where core-making and periphery-making processes are relatively balanced. It is through the semi-periphery that regions or countries have risen (e.g. Japan) and fallen (e.g. Argentina) in the history of the world-economy.

The basic premise of Jacobs' city-economy analysis is that cities are the ‘engines' of economic expansion. Cities and their regions are true economic entities; she argues that there are no such things as ‘national economies', they are merely political amalgams of different city-regions. The key concepts from her analysis pertinent to this paper are ‘new work' and ‘import-replacement'.

1. A key starting point is to distinguish old work from new work. The former consists of producing more and more of what is already being produced. This repetitive ‘production work' leads to a form of economic growth that does not count as economic expansion. This is because although the economy may become larger, it remains essentially the same in content, structure and complexity. New work, on the other hand, is ‘development work': it creates new divisions of labour that change the content, structure and complexity of an economy as well as growing it. Jacobs argues that only in cities are the conditions found to generate new work because this is where the quantity, concentration and diversity of old work are found to act as parent. New work is not market-led, it is market-making. This also makes it dependent on cities for capital, producer goods and services, and suitable labour that are only available in a city. Not all cities at all times are locales of important new work, but when they are generating economic expansion through new work they constitute dynamic cities.

2. Dynamic cities grow in two ways: the export multiplier effect where new work creates new export markets (other cities), and the import-replacement multiplier effect where imports are replaced through imitation and improvisation creating new work to supply the city market (replacing suppliers from other cities). Whereas the former produces steady economic expansion, Jacobs shows the latter to create ‘explosive' growth: ‘a process of immense, even awesome, economic force' (Jacobs, 1969, p. 150). It is these ‘economic spurts' caused by import-replacement that are the prime subject matter of this paper.

Thus the central image of Jacobs' approach is value making, creativity released in cities to generate economic expansion through cities.

The differences between the two theorists are obviously quite profound: whereas Wallerstein emphasizes unequal exchange, Jacobs seeks out mutualities in exchange. But they are not nearly as far apart as first appears. Jacobs (1984), for instance, recognizes that trade relations can have both positive and negative effects on communities, with her ‘economic grotesques' equivalent to Wallerstein's periphery. Similarly, Wallerstein's definition of core processes centering on ‘high tech, high wage' processes reveals positive impacts on selected communities that will inevitable include Jacobs' dynamic cities. Such correspondences of ideas are the result of them both adopting a common inductive spirit in which theoretical ideas are carefully grounded as historical processes. But beyond this similarity, we argue further that there is an important complementarity between the two approaches. This is because we can recognise that each writer's work focuses critical attention upon a hiatus in the work of the other. In Wallerstein's case, there is an overt neglect of cities and their economic worth;³ in Jacobs' case, there is a complete lack of the formative structures through which economies operate.⁴

For the purposes of this study we combine the two approaches as follows. We use Wallerstein's identification of time and space structures for the overarching framework of the study. This is the modern world-system in our title. We use Jacobs' identification of city economic spurts in economic expansion as the critical mechanism in our study. This is the explosive city growth in our title. In short we add a basic structure to Jacobs' ideas and a key mechanism to Wallerstein's ideas. The result is a study that searches out cases of explosive city growth from 1500 to the present, starting with the European Atlantic economy and then extending the search as the modern world-system expands, first covering cities in India, the Ottoman Empire and Russia in the eighteenth century, and cities in the rest of the world from the nineteenth century. Although Jacobs' work on economic spurts has been lauded as initiating contemporary economic growth theory (Krugman, 1997, p. 5), it remains a hugely under-researched topic in historical social science. Her influential ideas have never been systematically assessed over long time and big space dimensions. Here this is attempted for the first time for the whole time-space that is the modern world-system.

METHODOLOGY: SEARCHING OUT EXPLOSIVE CITY GROWTH

Economic statistics for pre-1800 cities are scant and in many cases non-existent. According to Hohenberg and Lees (1985, p. 10), population size, while not ideal, provides a proxy for a city's economic base when dealing with pre-industrial European cities. De Long and Shleifer (1993) also use population as an index of economic activity but acknowledge that for pre-industrial cities, which were centres of consumption, seats of royal courts and administrative centres, the relationship between population and economic activity does not necessarily hold. In order to overcome the lack of economic data we follow the practice of using population as an index of economic activity and changes in the size of a city's population as an indicator of economic growth. We use rapid population increase as an indicator of economic growth for cities that were significant centres of trade and industry. Administrative centres and centres of consumption, such as Madrid, appeared in our initial data and, based on our reading of secondary sources in urban history, we have attempted to eliminate them from our sample.

The project deals with the cities of Wallerstein's modern world system. For the period 1500-1800, following Wallerstein, we examined only western European and American cities. We used the de Vries (1984) European data base which is based on a variety of archival records: church attendance lists, baptisms and burials, censuses and tax records. De Vries also makes use of some secondary sources such as the estimates provided by contemporary town historians. Our confidence in his estimates was confirmed by their general agreement, in the period under consideration, with a database constructed independently by Bairoch, Batou and Chèvre (1988). De Vries is critical of Bairoch's method and we found significant differences for smaller cities (5000-20 000) at the beginning of our period but Bairoch et al. (1988, p. xiii) report that for the sum of the population of 92 cities over 20 000 in 1700, the two databases differed by only 0.6 percent. We chose the de Vries database because our aim was to identify periods of rapid economic growth and, following Jacobs (1969, 1984), we search for economic spurts of no longer than 50 years. In the early modern period de Vries provides estimates at 50-year intervals whereas, before 1750, the Bairoch et al. data is given in 100-year intervals. Most cities we consider in the early data period are European but there are a small number of American cities included and for these we use Chandler's (1987) population estimates. Despite the criticisms of Chandler's method and sources (de Vries, 1984; Bairoch et al., 1988) his data are required for this small group of cities to ensure completeness in covering the initial European Atlantic economy.⁵

Reflecting the expansion of the modern world system we included cities in British India and the Russian and Ottoman Empires after 1750, and China and Japan after 1850. Russian cities are not included in the de Vries database and we again used Chandler (1987). The de Vries database, which we had used for our earlier estimates, ends at 1800 and deals only with European cities. Bairoch et al. (1988) cover a slightly wider geographic area than de Vries but their estimates do not cover the period after 1850. For city populations between 1850 and 1950 we used the estimates made by Chandler (1987); in the period when national censuses become available criticisms of his method and sources (de Vries, 1984; Bairoch et al., 1988) no longer obtain. For the final census point we decided to use the City Population website (Brinkhoff, 2007) because it specifically estimates populations for ‘urban agglomerations' (in 2007⁶) thereby ensuring that we avoid the worst cases of under-bounded cities and include twentieth century suburban growth. As will become apparent below, this decision was not a vital one because the demographics for the last change period, in the event, were not finally used.

The first step was to designate rosters of cities over time. For this we used the five end-of-century dates 1600, 1700, 1800, 1900, and 2007. The initial task was to create a manageable and comparable roster of cities for each date. The data provide for long and variable lists of cities and we were faced with making pragmatic decisions on which cities to include in our analysis. We wanted to include enough cities to provide a coverage enabling results to be meaningful while at the same time ensuring our interpretations would not be overwhelmed by the quantity of cities. After some experimentation we developed a simple variable threshold method for inclusion: the average population for the top three cities⁷ of each year was computed and the threshold for city inclusion designated as one tenth of this value. This produced five rosters of cities with 68 cities in 1600, only 39 in 1700, rising again to 60 in 1800, 49 in 1900, and a massive 153 cities in 2007.⁸

For each roster population estimates were found for 50 years previously (1550, 1650, 1750, 1850, 1950) and for the initial year of 1500. Using all 11 census points, population changes were computed at 50-year intervals and annualised into change statistics.⁹ Until 1750 only cities from the initial ‘European-Atlantic world-system' were included in the rosters but, as previously indicated, after this date cities from other parts of the world were included as they were incorporated into the modern world-system. However, there is a potential problem with the demographic changes recorded in newly incorporated world-system zones: we cannot be sure whether change results are the result of processes before or after incorporation. This is considered further below.

We are now in a position to identify large demographic changes over each of the ten 50-year intervals. These were defined as cities with an annual growth of 1% or more over a 50-year period.¹⁰ In all 358 such demographic spurts were found but these were far from evenly spread: in the sixteenth century there were 13 and then 14; in the seventeenth century first only 10 and reducing to 6; in the eighteenth century 12 and 16; in the nineteenth century a jump to 29 cities and then 40; and, finally, the twentieth century has the highest numbers of demographic spurts – 91 and 127.¹¹ All of these city demographic spurts are potential candidates for Jacobs' ‘economic explosions'.¹²

The identification of economic dynamic cities is based upon using secondary sources (city histories) to understand why a city grew so quickly. Basically this came down to identifying three main groups of cities: first, those whose demographic expansion was largely due to political factors – these were mainly capital cities that prospered with the centralization of states in various phases of modernity; second, cities whose growth can be largely accounted for by economic factors – these were largely ports and new production centres; and thirdly there were several cities where both political and economic factors were important. In addition we have identified an ‘inherited' category to cover cities whose growth appears to be more related to their previous world-system than to the modern world-system. In practice this category was constituted by Indian cities between 1750 and 1800 when change appeared to be dominated by processes derived from the strong indigenous Indian urban tradition. It is the economic and mixed categories that constitute our inventory of dynamic cities experiencing Jacobs' process of ‘explosive growth'.

This categorization is inevitably subjective. It must be emphasized that we are dealing with relative degrees of economic and political processes generating growth: there are no cases of ‘pure' economic or political cities. In addition, of course, economic and political processes are rarely independent of each other. Some examples of difficult judgements will clarify. Dublin and Copenhagen in the seventeenth century are two political cases – capital cities – that had important economic functions through dominating their respective countries. Dublin provided vital port, financial and legal services for its agricultural hinterland (Dickson, 2001; Lennon, 2001); Copenhagen had important trading and added new industries related to its raw materials (forest and iron) hinterland (Kirby, 1990; Jespersen, 2004). However, their huge annual growth rates, 5.06% and 4.09%, in the throes of Europe's late seventeenth century economic crisis, does suggest the political process is dominating. There is no indication that these cities and their countries were immune to the economic travails of the time. Thus we conclude that they are best allocated to the political rather than the mixed category. In contrast, consideration of Seville and Cadiz either side of 1700 has a different outcome. Clearly the economic activity in each of the two cases is highly dependent of the Spanish state in terms of where it decides to receive its American silver: Cadiz replaced Seville and this is reflected in the sequence of these two cities' demographic spurts. However, while dependent on the state decision, the associated processes are essentially economic and are hugely implicated in the large change we have recorded. For this reason both cases are allocated to the mixed category and therefore qualify as a Jacobs' economic spurt.¹³

Before we turn to interpreting the inventory of economically dynamic cities we need a digression to consider the most recent lists of cities: the 2007 roster and the 1950-2007 change list. This is because in the twentieth century the relationship between population growth and economic expansion in cities broke down. This has been generally understood at least since Castells (1977), who noted that in the mid-twentieth century cities in the ‘third world' were growing demographically but without a concomitant ‘industrialization' (i.e. growth of (formal) jobs). This process has come to be identified as mega-city development and is directly reflected in the 2007 city roster and for 1950-2007 demographic change. The latter levels of change, and its overall quantity in the system, are historically unprecedented and can be identified as a final ‘de-peasantization' in the world-economy (Davis, 2006). Since the balance between push and pull factors in this rural-urban migration has moved decisively towards the push end of the spectrum, the process does not relate to Jacobs' dynamic cities as migration magnets. Therefore we need a different measure to identify leading cities in c.2000, and for measuring change c.1950-2000. Instead of mega-cities, we need to measure ‘global cities' or ‘world cities', defined by their economic prowess.

This is a serious and difficult problem for this study and our solution is far from ideal. The only reliable information that approaches our demanding data needs is airline flights, a source commonly employed to measure global/world city importance (Smith and Timberlake, 2001; Derudder and Witlox, 2008). There is a worldwide time series of information on scheduled flights that can be used to identify Jacobs' dynamic cities. Here the assumption is that the more flights into and out of a city, the more economically vibrant it is. Widespread scheduled flights are not found in 1950 and 1960 and therefore we have to use 1970 as our start date by which time scheduled air flights are relatively common worldwide. Thus we draw on data for air flights from 1970 to 2005 that have been collected by Cándida Gago García, Universidad Complutense de Madrid, for 70 cities worldwide. We subjected these data to the same methods as the demographic data: (i) a roster was defined for 2005 by including all cities with 10% of the flights of the average of the top three; (ii) changes from 1970 to 2005 are computed and annualised; (iii) those with growth rates above 1% are identified as contemporary ‘spurt cities'. 35 spurt cities defined by the new method are identified.¹⁴ These replace the 1950-2007 demographic change results for our inventory of dynamic cities in the modern world-system.¹⁵ In comparison to the demographic data, these data are more direct measures of economic change and therefore there is no need to search out secondary materials to eliminate cities where non-economic processes dominate.

This re-specification of spurt cities after 1950 leads on to one other alteration in the constitution of an inventory of Jacobs' explosive growth cities. Mega-city development was a twentieth century phenomenon that came to a head in the second half of the century, but the process was operating in the first half also. This can be seen by comparing sizes of demographic change between 1900-1950 and 1950-2007. Although there are larger changes in the second period, the first period is also historically unusual in the sizes of change: the demographic data for the first half of the twentieth century are not so different from the later data we have replaced by air flight data. We cannot get equivalent replacement data for the first half of the century but it does seem necessary to trim the 1900-1950 long list of demographic spurts. This is done through a method of post-diction. Those cities from outside the core that are not in the new list of 1970-2005 cities are removed. The assumption is that ‘third world' cities did not peak in economic importance in the first half of the twentieth century (all evidence suggests the opposite) and therefore their position in the 1900-1950 list reflects early indications of the de-peasantization process discussed previously. This culling reduces the number of spurt cities for the demographic 1900-1950 inventory from 91 to 39.

The practical outcome of these methodological applications is an inventory of 184 explosive city growths in the modern world-system.

AN INVENTORY OF EXPLOSIVE CITY GROWTHS IN THE MODERN WORLD-SYSTEM

Table 1 provides a summary of the inventory of explosive city growths. The main message of this array is that both the frequency and magnitudes of the 184 economic spurts increase with the growth of the modern world-system, both geographically and economically, but with the major exception of the seventeenth century. The dearth of economic spurts at this time clearly reflects Europe's ‘crisis of the seventeenth century', a time of both demographic and economic decline. It is with the subsequent ‘industrial revolution' after 1700 that the frequency and magnitudes of city growths pick up and surpass the sixteenth century. That this trend continues through the nineteenth century but then comes to a halt in the twentieth century may reflect the change of metric for the final period, suggesting a possible under-estimation of economic spurts in the second half of the twentieth century. Data limitations mean that there is nothing further we can do about this, but, aside from this result, the findings do seem to track the general movement of the world-economy throughout the modern world-system quite well.

Table 1: Summary Distribution of 184 City Growth Explosions

^*Different metric, see text.

The inventory of all 184 economic spurts is given in Table 2. City spurts by time period are listed in rank order by size of change. The top row of cities, therefore, is the sequence of largest explosive city growths in the modern world-system: from Lisbon to Beijing via London, Amsterdam, Seville, Liverpool, Manchester, New York, Chicago, and Los Angeles. These are all reasonably predictable, except perhaps for ‘Shakespeare's London' in the second half of the sixteenth century (although this was an important city spurt identified in the original source for these phenomena (Jacobs, 1969, pp. 154-56)), and Seville a century later (the receiving port for ‘Spanish ‘ silver).

Table 2: Inventory of Explosive City Growths in the Modern World-System

The geography of the economic spurts shows two types of city. First, explosive city growths are found in world-regional concentrations initially in Western Europe, then North America, and finally Pacific Asia. Such concentrations of dynamic cities indicate the existence of core-making processes. Second, beyond this core zone there is a scatter of economic spurts within more isolated cities. These ‘islands' of core-making processes within predominantly peripheral regions define where the semi-periphery exists (Taylor, 2005). Examples are Moscow/Russia in 1750-1800, New York/USA in 1800-1850, Buenos Aires/Argentina in 1850-1900, and Shanghai/China in 1900-1950. The periphery is marked by the absence of economic spurts as typified by Africa, whose cities do not feature in Table 2.

There is a lot of detail in Table 2 and we will initially focus on the first and last time periods in this initial perusal. At the beginning of the modern world-system, we see the remnants of Braudel's (1984) ‘two poles' of medieval European economic development. Already the formerly dominant southern pole of northern Italy is in decline, represented by just three cities, whereas the northern pole of the (greater) Low Countries is represented by four cities led by Antwerp. However this pole is rather dispersed within a Hamburg-London-Rouen triangle. The important changes, however, are two fresh areas of important growth, two German centres, and Iberia housing the top two explosive city growths based upon new Atlantic trade. This mixture of old and new explosive city growths was to herald in a new world, the modern world-system.

The final column explicitly reflects contemporary globalization. Although cities of Western Europe – the only world region with constant core membership – continue to be well represented with nearly half the economic spurts, these are largely concentrated at the lower echelons of the list. In contrast, Pacific Asian cities have only half the number of Europe but their economic spurts are concentrated at the top of the table led by Beijing and Shanghai. It is the rise of this world region on the back of its leading cities that has made ‘globalization' a meaningful concept. Previously the core had stubbornly rested in the North Atlantic zone, but, led initially by Japanese cities, Pacific Asia has added a large new region to the core thereby moving the world-economy away from its focus on its American/European centre. Possibly this late mixture of old and new explosive city growths will herald the end of the modern world-system as a sustainable world-economy as Wallerstein (1999) foretells.¹⁶

Between these two end-columns of results there have been three hegemonic cycles and the remainder of the detail in Table 2 will be interpreted through this theoretical lens.

EXPLOSIVE CITY GROWTHS IN HEGEMONIC CYCLES

In Wallerstein's (1984) world-systems analysis, hegemonic states are a sort of ‘core within the core'; it is where the most creative ‘high tech, high wage' economic processes emanate to create such a strong economic situation that cultural and political dominance follows as other countries try to emulate the hegemon's conspicuous success. As previously mentioned, he identifies three such cycles; each centred on a distinct period of ‘high hegemony' of a few decades but the whole cycle, to and from high hegemony, is much longer. Although overlap is not ruled out between cycles, they are usually described as a sequence with the Dutch cycle the longest and the American cycle the shortest (the twentieth century as ‘American Century'). However, our view of the cycles through cities suggests a different way of sequencing. Since the creativity that generates and sustains high hegemony is to be derived from explosive city growths, this new look at the cycles is well-grounded theoretically, at least for the economic dimension of hegemony. A cursory examination of Table 2 suggests there are three cycles of equal length, two centuries each, with the British and American cycles sharing the nineteenth century. The sheer weight of American city spurts in the nineteenth century means that the economic antecedents of US high hegemony need to be rethought and made earlier. Thus the three cycles are a separate Dutch cycle in the sixteenth and seventeenth centuries, a British cycle in the eighteenth and nineteenth centuries overlapping with an American cycle in the nineteenth and twentieth centuries.

In Table 3 part of Table 2 is reproduced highlighting Dutch cities in Dutch hegemony. Four Dutch cities are featured that experienced eight examples of explosive city growth between them. Led by Amsterdam with three such spurts, this table confirms that the Dutch Republic was not simply ‘Amsterdam's city-state' as has been suggested but was a multi-nodal city-region of several vibrant cities (Taylor, 2005). Starting with Amsterdam's first explosive city growth, before the creation of the Dutch Republic, there are three explosive city growths that build up Dutch hegemony (Amsterdam, Haarlem and Leiden), three such spurts that continue during high hegemony (Amsterdam, Leiden and Rotterdam) with only the latter city continuing with a final Dutch city spurt in the down-side of the cycle. During this period the Dutch went largely without a serious economic rival; France came the closest with five cities (Rouen, Bordeaux, Paris, Lyon and Marseilles) and six spurts, but all are smaller than the Dutch city growths.

Table 3: Dutch Cities in the Dutch Hegemonic Cycle

In Table 4 part of Table 2 is reproduced highlighting British cities in British hegemony. In this case seven British cities are featured with 21 explosive city growths between them. The key feature is the dominance of the four great cities of northern Britain: Birmingham, Glasgow, Liverpool and Manchester. These four cities dominate the eighteenth century with 8 of the 13 economic spurts recorded plus an additional early spurt from Bristol. The big four cities continue with explosive growth in both nineteenth century periods, although gradually falling down the ranks. Their clustered position just below four US cities in 1800-1850 reflects the fact that the US cities were starting from a lower population base, that is to say, Liverpool, Manchester, Birmingham and Glasgow still dominated the world-economy. In both nineteenth century lists they are joined by Newcastle and London. London was conspicuous by its absence among eighteenth century city spurts and, although featuring in the nineteenth century, its economic spurts are lowly ranked. Newcastle is somewhat like Rotterdam in the Dutch cycle: it arrives late and has its largest spurt at the end of the hegemonic cycle. France is usually seen as Britain's main rival during its hegemony but its economic competition was severely weak: in Table 4 only two French cities are featured, Paris and Lyon, both with lowly ranked spurts in both nineteenth century time periods. Clearly the French were less of an economic rival to the British than they were to the Dutch in the previous cycle.

Table 4: British Cities in the British Hegemonic Cycle

In Table 5 part of Table 2 is reproduced highlighting American cities in American hegemony. In this case 15 US cities are featured with 25 explosive city growths between them. The surprise is the top four rankings in 1800-1850 for the leading east coast cities: New York, Baltimore, Philadelphia and Boston. These cities continue to feature in the second half of the nineteenth century, albeit with much lower rankings, but now Chicago is ranked first and Pittsburgh ranked fourth showing important inland explosive city growths. US dominance of spurts is greatest in the first half of the twentieth century with four of the top five places: explosive city growth has now reached the Pacific coast with Los Angeles ranked first, two southern cities ranked second and third, Houston and Dallas, with another inland industrial centre, Detroit, ranked fifth. In addition, San Francisco and Seattle add to the Pacific coast representation and Atlanta to southern representation. Washington also features for the first time and New York, Boston and Philadelphia, but not Baltimore, continue with economic spurts in the new century. In the 1970-2005 period the US returns to having just four cities in the list. Now it is Washington with the highest ranking (third),¹⁷ Los Angeles and Chicago continue to feature, and Miami makes a first appearance.¹⁸ In this hegemonic cycle the main rival is very clear: Germany has seven cities with 16 episodes of explosive city growth. The main challenge was in the second half of the nineteenth century, when there were five German cities near the top of the economic spurts: Leipzig (ranked third), Berlin (6^th), Dresden (8^th), Hamburg (11^th) and Munich (14^th).¹⁹

Table 5: American Cities in the American Hegemonic Cycle

Five general points can be made from comparing these hegemonies. First, although the USA features most cities in its hegemonic cycle, this is largely a feature of the overall growth of the modern world-system; therefore, despite having least explosive city growths, it is Dutch cities that most dominate their hegemonic cycle. This result argues against economic hegemony increasing as the world-economy grows.

Second, mention must be made again of the nineteenth century overlap of the British and American cycles; this is a unique feature of this city-centric analysis and needs to be taken seriously by those who prefer neat sequential cycles. This result argues against both the idea of simple sequential cycles and the notion that the cycles are becoming shorter as the world-economy grows.

Third, we must recognise the way in which the methodology affects the results. To make this exercise manageable, and respecting the limitations of the earlier demographic statistics, we adopted a demographic cut-off for our five century rosters of cities. Although this still produced almost 200 explosive city growths, there are inevitably other economic spurts by smaller cities that did not make the cut-off. Even though we identified 26 cities having 54 explosive city growths within the three hegemonic states during their respective cycles, this will not constitute the total economic spurts that created and sustained their hegemonic positions. For instance, each of the following pairs of cities respectively contributed to their country's hegemony although they do not appear in our results: Middelburg and Dordrecht, Bradford and Nottingham, and St Louis and Cleveland. What we have measured are the explosive city growths of the leading cities, which are just part of three tight clusters of economically vibrant cities – in Holland/Zeeland, in northern Britain, and in the US ‘manufacturing belt'/California – where hegemonic power was created within the world-economy.

Fourth, these results can be compared with other studies of cities within hegemonic cycles. Although Wallerstein (1984) does not refer to cities in his initial statement of the hegemonic model, others do refer to cities, particularly in relation to the financial function within hegemony. This is derived originally from Braudel (1984) and developed further by Arrighi (1994). Braudel traces a sequence of financial centres as the economic control centres of their eras that overlap with Wallerstein's cycles: after Venice, Antwerp and Genoa come Amsterdam, London and New York. The latter three are related to modern hegemonic cycles specifically as a source of financial power that stays with the hegemonic state late into its cycle. Thus Amsterdam only gave way to London as the leading financial centre during the Napoleonic War, and London to New York only during World War I. Making the link between these cities and the hegemonic model even starker, Lee and Pelizzon (1991) refer to these three cities as ‘hegemonic cities'. In the third point above we noted that our 26 cities in hegemonic states did not constitute the totality of cities' contributions to hegemony, but in the finance version, there is only one city per hegemon identified. This very narrow urban perspective ultimately derives from Braudel's (1984) list of leading financial centres that derives from hierarchical central place thinking. The limitations of this approach when considering non-local city processes has been discussed elsewhere (Taylor, 2009; Taylor et al., 2010). All that needs to be said here is that by thinking in hierarchical terms, focus has alighted on one top city per hegemon and the networking activity of multiple cities clustered within the hegemonic state that actually creates hegemony as described in this paper is simply missed.

Fifth, in all three hegemonic cycles, cases of explosive growth in leading cities are weighted towards the beginning of each cycle and trail off towards the end of the cycle. This empirical observation has critical theoretical importance. There are two hypotheses on how economic growth of cities relates to hegemonic status. For the majority of world-systems analysis that neglects cities and focuses on states, cities merely reflect the hegemonic status of their state so that their economic growth is an outcome of hegemony. In this case explosive growths should be relatively even in the distribution over the cycle. From a city-centric view, cities are the engines of economic growth and therefore they are where hegemony can be and is created. This is the position taken here: not outcomes, cities are the producers of hegemony. For this hypothesis, explosive economic spurts should be front-loaded on the cycle to get it started – which is exactly what we have found. There is a corollary of this that is of contemporary relevance. The fall-away of American spurts at the end of its cycle does mean that those who are expecting a second US era of dominance cannot expect the same economic basis that provided the foundation of the US hegemonic cycle.

CONCLUDING REMARKS: WHERE WE GO FROM HERE

This paper has been an exercise in theoretically-informed quantitative empirical analysis. In its use of a combination of Jacobs and Wallerstein concepts the research has made particular contributions to world-systems analysis. In short, a revisionist city-centric version of world-systems analysis has been presented that informs the following concepts:

1. core – core-making occurs in world-regional clusters of cities currently encompassing Western Europe, Northern America and Pacific Asia as the core zone;
   
2. semi-periphery – this mix of core-making and periphery-making is identified in states that have some core-making in one or a few cities but are surrounded by urban and rural periphery-making processes (e.g. major Latin American countries);
   
3. periphery - periphery-making is indexed by world-regions with no core-making cities (e.g. inter-tropical Africa);
   
4. hegemony – the state of hegemony is indexed by a concentration of explosive city growths;
   
5. hegemonic cycles – cycles are not simple sequences, they overlap: one country's cities (USA) can be rising during the high hegemony of another country (UK);
   
6. building hegemony - explosive city growths are front-loaded in hegemonic cycles strongly suggesting that they are the producers of hegemony.

For Jacobs' theory of cities as economic process, the research has illustrated her concepts in a big time, big space context but has not critically taken the concepts forward in any sense. From a critical realist perspective, the work reported here can only be a first step. Our analyses only describe outcomes of the mechanisms that constitute the Jacobs' city process. To balance the contribution made to world-systems analysis requires new and different methodologies. Therefore we envisage two further stages to this research. First, we need to understand the economic spurts in terms of the expansion of particular industries and services within the different cities. Focusing on selected cities with good secondary material (in the form of detailed economic history texts), we are comparing and contrasting explosive city growths across our time and space dimensions. Second, we have to carry out archival research on a limited number of explosive city spurts to chart how the division of labour in a city is being made more complex. This research will focus on cities especially well supplied with commercial directories, and will perforce be concentrated on nineteenth century cities. It is through these additional stages to this research project that we intend to round off our understanding of explosive city growths in the modern world-system.

ACKNOWLEDGEMENT

We acknowledge the support of The Leverhulme Trust for funding the research project of which this paper is a part.

REFERENCES

Abbott, C., 1999, Political Terrain: Washington, D.C. from Tidewater Town to Global Metropolis. Chapel Hill, NC: The University of North Carolina Press.

Arrighi, G., 1994, The Long Twentieth Century. London: Verso.

Bairoch, P., Batou, J., and Chèvre, P, 1988, La population des villes Européenes de 800-1850. Geneva: Libraire Droz.

Braudel, F., 1982, The Wheels of Commerce. London: Collins.

Braudel, F., 1984, The Perspective of the World. London: Collins.

Brinkhoff, T., 2007, The principal agglomerations of the world. Retrieved December 11, 2007 from the City Population web site at http://www.citypopulation.de

Castells, M., 1977, The Urban Question. Oxford: Blackwell.

Chandler, T., 1987, Four Thousand Years of Urban Growth: An Historical Census. Lewiston, NY: Edwin Mellen.

Davis, M., 2006, Planet of Slums. London: Verso.

De Long, J. B., and Shleifer, A. 1993. Princes and merchants: European city growth before the Industrial Revolution. Journal of Law and Economics, Vol. 36, 671-702.

Derudder, B. and Witlox, F., 2008, Mapping world city networks through airline flows: context, relevance, and problems. Journal of Transport Geography, Vol. 16, 305-312.

Dickson, D., 2001, Death of a capital: Dublin and the consequences of union. In P. Clark and R. Gillespie, editors, Two Capitals: London and Dublin 1500-1840. Oxford: Oxford University Press, 111-131.

Forstall, R. L., Greene, R. P., and Pick, J. B., 2009, Which are the largest? Why lists of major urban areas differ so greatly. Tijdschrift voor Economische en Sociale Geografie, Vol. 100, 277-297.

Hohenberg, P. M., and Lees, L. H. 1985. The Making of Urban Europe. Cambridge, Mass.: Harvard University Press.

Jackson, J. H., 1997, Migration and Urbanization in the Ruhr Valley, 1821-1914. Atlantic Highlands, NJ: Humanities Press.

Jacobs, J., 1969, The Economy of Cities. New York, NY: Random House.

Jacobs, J., 1984, Cities and the Wealth of Nations. New York, NY: Random House.

Jacobs, J., 2000, The Nature of Economies. New York, NY: Random House.

Jacobs, J., 2004, Dark Age Ahead. New York, NY: Random House.

Jespersen, K. J. V. 2004, A History of Denmark. Basingstoke: Palgrave Macmillan.

Kirby, D. 1990, Northern Europe in the Early Modern Period. The Baltic World 1492-1772. London: Longman.

Krugman, P., 1997, Development, Geography, and Economic Theory. Cambridge, MA: MIT Press.

Lawton, R. and Lee, R., editors 1989, Urban Population Development in Western Europe from the Late-Eighteenth to the Early-Twentieth Century. Liverpool: Liverpool University Press.

Lee, R. and Pelizzon, S., 1991, Hegemonic cities in the modern world-system. In R. Kasaba, editor, Cities in the World-System. New York, NY: Greenwood, 43-54.

Lennon, C. 2001, The changing face of Dublin, 1550-1750. In P. Clark and R. Gillespie, editors, Two Capitals: London and Dublin 1500-1840. Oxford: Oxford University Press, 39-52.

Smith, D. A. and Timberlake, M. F., 2001, World city networks and hierarchies, 1977-1997: an empirical analysis of global air travel links. American Behavioral Scientist, Vol. 44, 1656-1678.

Taylor, P. J., 2005, World-systems analysis and globalization: a Jacobsean exploration of pasts, presents and futures. Belgeo, 3/2005, 265-273.

Taylor, P. J., 2009, Urban economics in thrall to Christaller: a misguided search for city hierarchies in external urban relations. Environment and Planning A, Vol. 41, 2550-2555.

Taylor, P. J. and Aranya, R., 2008, A global ‘urban roller coaster'? Connectivity changes in the world city network, 2000-04. Regional Studies, Vol. 42, 1-16.

Taylor, P. J., Hoyler, M., and Verbruggen, R., 2010, External urban relational process: introducing central flow theory to complement central place theory. Urban Studies, Vol. 47, in press.

Vries, de J., 1984, European Urbanization, 1500-1800. London: Methuen.

Wallerstein, I., 1974, The Modern World-System. New York, NY: Academic Press.

Wallerstein, I., 1979, The Capitalist World-Economy. Cambridge: Cambridge University Press.

Wallerstein, I., 1983, Historical Capitalism. London: Verso.

Wallerstein, I., 1984a, Politics of the World-Economy. Cambridge: Cambridge University Press.

Wallerstein, I., 1984b, Cities in socialist theory and capitalist praxis. International Journal of Urban and Regional Research, Vol. 8, 64-72.

Wallerstein, I., 1999, The End of the World as We Know It. Minneapolis, MN: University of Minnesota Press.

Wallerstein, I., 2004, World-Systems Analysis: An Introduction. Durham, NC: Duke University Press.

Weber, A. F., 1899, The Growth of Cities in the Nineteenth Century: A Study in Statistics. Ithaca, NY: Cornell University Press [1963 reprint].

NOTES

^* Correspondence concerning this article should be addressed to P. J. Taylor, Department of Geography, Loughborough University, Loughborough, UK; email: crogfam@yahoo.com

^** Ann Firth, School of Public Health, Curtin University of Technology, Perth, Australia

^*** Michael Hoyler, Department of Geography, Loughborough University, UK

^**** Dennis Smith, Department of Social Sciences, Loughborough University, UK

1. His results are problematic because his criterion is a doubling of population in a given time period when the latter vary in length (50 years, 100 years and 150 years). Thus it is not surprising that he finds by far the most ‘fastest growers' in his longest period.

2. De Vries (1984, pp. 141-2) does briefly ‘explain' these cases using a simple typology relating to capitals, ports and ‘others' but clearly he does not allocate the same theoretical importance to these extraordinary cities that we do in the argument developed in this paper.

3. This is despite Wallerstein drawing on the ideas of Fernand Braudel in whose work cities are prominent. Wallerstein (1984b) in his only paper on urbanization treats it as de-ruralization with no sense of the special nature of cities as attractions to migrants.

4. Jacobs presents her ideas as effectively ‘generic' with no structural constraints: explosive city growth is as old as cities. Therefore she has no notion of modernity and does not address the historically unique size of urbanization in the modern world (Jacobs, 2000).

5. Using Chandler (1987) is not ideal for comparative analysis – De Vries (1984) dismisses this data as unusable for European cities. However, we have no choice here but it does create a subsequent problem with estimates at 50-year intervals for American cities; this was relevant for later analysis and is noted below.

6. The problem with using a web site that is regularly updated is that prior results are lost: therefore, we have been unable to find 2000 population figures and have to use the population estimates for the year when we accessed the site (i.e. 2007). For the latest discussion of estimating current populations of large cities see Forstall et al. (2009).

7. Initially we used the population of the largest city but this left our roster lists dependent on the vagaries of just one single population estimate; using the average of the top three increases the chances of producing comparable lists over time.

8. The population estimates of the roster cities are available on the GaWC website at www.lboro.ac.uk/gawc/datasets/da25.html

9. The ten lists of city demographic changes are available on the GaWC website at www.lboro.ac.uk/gawc/datasets/da25.html

10. The choice of 1% is another pragmatic decision to provide a manageable set of results derived after some experimentation. As with defining the city rosters, there is no ‘theoretical' way of specifying the cut-off point for finding demographic spurts.

11. With set 50-year intervals, it is possible to miss high growth cities that straddle a census date. For example, a city that grew rapidly from 1680 to 1730 might not appear in our results because its growth is diluted into two of our periods. There is nothing we can do about this for the early periods, and regular official censuses at ten-year intervals only become widespread in the twentieth century. We could search out ‘straddle' growth statistics for some cities in the nineteenth century and for most cities in the twentieth century but this would break comparability with our first three centuries. Thus we stay with our 50-year intervals and accept that our listing of high demographic spurts is not necessarily comprehensive but is nevertheless a reasonable representation of urban demographic change across the whole history of the modern world-system.

12. It should be noted that for the few American cities before 1800, use of Chandler (1987) data means that estimates at 50-year intervals are not always possible. However looking at the change rates over 100-year intervals suggests a dearth of demographic spurts in this region. There are two probable exceptions: Potosi from 1550 to 1600 and Mexico City from 1750 to 1800. In each case no estimates are given for the earlier year because the cities fall below Chandler's thresholds for inclusion in his lists. Given their initial low populations, it is likely that both cities would have had annual growth rates above 1% but because their actual growth rates cannot be calculated they are not included in our demographic spurts.

13. The full results of this exercise are available on the GaWC website at www.lboro.ac.uk/gawc/datasets/da25.html

14. The results of these procedures are available on the GaWC website at www.lboro.ac.uk/gawc/datasets/da25.html

15. These results are far from satisfactory. There are well-known problems in using airline data to estimate a city's importance in this way, notably airline company hub policies that might lead to overestimating a city's economic importance (Derudder and Witlox, 2008). Thus these results might be considered indicative rather than authoritative; their inclusion – their usefulness – is so as to provide a full temporal coverage of the modern world-system.

16. Coming from her different direction, Jacobs (2004) makes a similar prophesy of crisis.

17. Although a political capital, we have allocated Washington to our mixed category following Abbott's (1999) argument in recognising the importance of economic processes to contemporary Washington.

18. However remember that the 1970-2005 results are derived from a different metric and therefore are not strictly comparable to the previous lists.

19. In addition, Chandler's data includes the Ruhr as a ‘city' in 1900 but not in 1850 and therefore change has not been calculated. In another source (Jackson, 1997, p. 7) the ‘Ruhr Valley Industrial Area' grows from 4 cities in 1850 to 14 cities in 1900 with population rising from 288,082 to 1,952,406, an annual change of 12.04%. This result would rank Ruhr 5^th in the 1850-1900 list.