Spatial Segregation in Global Cities The hybrid outcome of urban change

Ayat Ismail

Spatial Segregation in Global Cities The hybrid outcome of urban change

Ayat Ismail PhD Dissertation Vrije Universiteit Brussel

Prof. Dr. Bas van HEUR, Promoter Prof. Dr. Eric CORIJN, Promoter Prof. Dr. Nurhan ABUJIDI, Co-promoter

March 2015

Acknowledgement

First and foremost, a very special ‘thank you’ should be extended to my advisors, Professor Eric Corijn, Dr. Bas van Heur, and Dr. Nurhan Abujidi. Their exceptional knowledge, and our theoretical discussions have guided me to approach the topic of the research from different perspectives. I have gained unequalled inspiration and knowledge about globalization and cities throughout my journey over the past years. I am deeply grateful for the assistance of Professor David Wong – Department of Geography and Geoinformation Science, George Masson University – as he granted me the permission to use his analytical tool in developing the findings of this research. I also deeply acknowledge the great assistance of Professor Manal Nassar, since I could not have done the statistical analysis without her comments and suggestions. Many thanks go to Professor Ben Derudder for his insightful remarks on the analytical study. I also would like to thank Dr. Weiyang Zhang for providing the required data for the statistical analysis. A special expression of gratitude goes to Dr. Marwa Abdel Latif for making her extensive knowledge and experiences available to me. Last but not least, I am grateful to my family members and my friends for their unconditional support. Special thanks goes to my mother Mrs. Horeya Diab for her extreme dedication and care, my husband Mr. Karim Bayoumi, my sister Miss. Esraa Ismail, Mrs. Hayat Diab, Mrs. Sarah El- Shal, and Mrs. Hadil Nasr for believing in my ability to face all the challenges.

ii

iii

Samenvatting

Een brede waaier onderzoek ondersteunt de hypothese dat ongelijkheid, sociale polarisering, uitsluiting en ruimtelijke segregatie de laatste tientallen jaren toenemen in wereldsteden ten gevolge van een aantal mondiale processen, zoals de groei van de diensteneconomie, de daarmee gepaard gaande veranderingen in de arbeidsmarkt en de afname van de herverdeling via de welvaartsstaat. Die hypothese wordt echter tegengesproken door verschillende elementen zoals de sterke vereenvoudiging van de globale- lokale interacties, die geïllustreerd wordt door de algemene aanname van de hypothese voor verschillende steden als ze maar gelijkaardige economische functies vertonen. Dit onderzoek richt zich op de gevestigde associatie tussen mondiale economische functies van steden en de groeiende sociaal-ruimtelijke ongelijkheid binnen die steden. Dat gebeurt door een stevige empirische basis uit te werken om de algemeenheid van het model “wereldstad=verdeelde stad” te testen , alsook om de invloed van contextuele bijzonderheden van particuliere steden op de lokale ontwikkeling te evalueren. Tenslotte beoogt het onderzoek een heuristisch raamwerk op te leveren voor vergelijkende studie van de vastgestelde ruimtelijke veranderingen in wereldsteden. Het onderzoek geeft eerst een theoretisch overzicht van de veelvoudige herstructurering van steden. Vervolgens worden de macro- tendensen in de wereldeconomie verbonden aan hun micro-gevolgen (met een speciale aandacht voor de ruimtelijke segregatie patronen in de steden). Dat gebeurt door de effecten van de stedelijke economische functies op lokaal beleid en op de huizenmarkt te bekijken. Tenslotte worden de veranderingen in sociaal-economische en etnische segregatie voor een grote dataset van 91 wereldsteden berekend. De samengevoegde resultaten van de analyse tonen een neergang van het algemene “wereldstad=verdeelde stad” model, terwijl de gevalstudies het belang van contextuele bijzonderheden aanduiden die bijdragen tot een unieke en niet veralgemeenbare patronen van sociaal-ruimtelijke configuraties in verschillende wereldsteden. Dat leidt iv de studie tot het formuleren van een aantal onderzoeksvragen en een mogelijke agenda voor alternatief onderzoek over het complexe proces van herstructurering in wereldsteden in een poging de beperkingen van de wijdverspreide aannames in de wereldsteden literatuur te boven te komen. De voorgestelde onderzoeksvragen hebben de bedoeling de hybride natuur van zowel het proces als het resultaat van de herstructurering binnen elke wereldstad te exploreren. Het herstructureringsproces bevat immers verschillende economische, politieke, sociale en ruimtelijke processen, daarenboven met elkaar verbonden op verschillende schalen van de stedelijke hiërarchie op een wijze die als “structurele hybride” kan worden gekarakteriseerd. Daarenboven zijn de producten van de herstructurering niet alleen het resultaat van de recente veranderingen in stedelijk regimes, maar worden ze ook beïnvloed door de historische patronen uit de voorgaande cycli van kapitaal-accumulatie. Die historische gelaagdheid in de ruimtelijke ontwikkelingen kan als een “chronologische hybride” worden bestempeld.

v

Abstract

A wide range of studies supports the hypothesis that levels of inequality, social polarization, exclusion, and spatial segregation are rising in global cities over the past decades as a direct outcome of certain global processes, such as the rise of the service economy, its associated changes in division of labor, and declined redistributive power of the welfare state. However, this hypothesis is criticized based on several contentions, including the oversimplification of the global/local interplay, which in turn, is reflected in the general agreement on the applicability of the hypothesis on different cities around the world as long as they share similar global economic functions. This study aims to investigate the established association between the global economic functions of cities and the growing levels of socio-spatial divisions within them. Through providing a solid empirical ground for investigating the general applicability of the ‘global city/divided city’ model, as well as evaluating the influence of contextual particularities of individual cities on the outcomes of urban change within them. Finally, the study suggests a heuristic framework for comparative analysis of the resultant spatial changes in global cities. The research offers a theoretical review of the multifaceted restructuring of global cities. Then, the macro trends of global economy are linked to their micro outcomes (with a special focus on spatial segregation patterns within cities), through understanding the implications of cities’ economic functions on local urban policies and housing markets. Finally, the changes in socioeconomic and ethnic segregation over the past decades are calculated for a large dataset of 91 global cities. The collective results of the analysis show the downfalls of the generalized global/divided city model, while the discussion of individual cities highlights certain contextual particularities, which are contributing to the production of unique and non- generalizable socio-spatial configurations in different global cities. Finally, based on the analysis findings, the study attempts to overcome the limitations of the widely accepted assumptions in the global city literature vi through raising a number of research questions that constitute an agenda for a possible alternative research approach for the complex process of urban restructuring of global cities. The proposed questions aim to further explore the hybrid nature of both the process and the outcomes of restructuring within each global city, as the process of restructuring involves different economic, political, social, and spatial processes that mélanges across different scales of the urban hierarchy in a manner identified as ‘structural hybridity’, while the outcomes of restructuring are not solely a product of the recent change in the urban regime, but they are also influenced by patterns of historical developments in the preceding rounds of capital accumulation, this fusion of historical layers of spatial developments is referred to ‘chronological hybridity’.

Table of Contents

Acknowledgement ...... i

Abstract ...... v

Table of Contents ...... vii

Table of Figures ...... xi

Table of Tables ...... xiii

1. Globalization and urban change: an overview ...... 1

1.1
Introduction ...... 1
1.2
Research problem ...... 7
1.2.1
The general applicability of the ‘global/divided city’ model ...... 7
1.2.2
Challenging the prominent role of globalization ...... 9
1.2.3
The spatial dimension of inequality and polarization ...... 10
1.3
So what? One step towards breaking the vicious cycle ...... 12
1.4
Research objectives ...... 13
1.5
Research Questions ...... 15
1.6
Structure and methodology ...... 16

2.
Globalization and the production of inequality ...... 23

2.1
Approaching urban inequality: the three schools of thought ...... 26
2.1.1
Occupational polarization and urban dualism approach ...... 30
2.1.2
Occupational professionalization and inequality approach ...... 36
2.1.3
Contextual particularities approach ...... 42
2.1.4
Exacerbated inequality and polarization: other complementary views ………...... 47
2.2
Issues of generalization, scale, and space ...... 51
2.2.1
Justified generalization or an overstatement? ...... 52

viii

2.2.2
The global and the local: a question of scale ...... 54
2.2.3
Spatial segregation: automatic outcome of inequality? ...... 57

3.
Spaces of globalization and inequality ...... 59

3.1
Space, society, and capitalism ...... 61
3.1.1
Space as a social construct ...... 62
3.1.2
Spaces of capital accumulation ...... 63
3.2
Globalization on the ground ...... 65
3.2.1
Urban development and economic growth ...... 66
3.2.2
Inner city gentrification ...... 69
3.2.3
Issues of differential access ...... 71
3.1
Local contingencies and the individuality of cities ...... 79

4.
Testing the global/divided city model: an introduction to the analytical study ...... 85

4.1
Analytical framework ...... 86
4.2
Which city is a global city? – Dataset selection ...... 87
4.2.1
Primary dataset ...... 90
4.3
Measures of spatial segregation ...... 99
4.4
Data type ...... 104
4.5
Limitations and constrains ...... 105
4.5.1
Data availability ...... 105
4.5.2
Defining the boundaries of the study area ...... 108
4.5.3
Problematic cross-city comparisons of the results ...... 112
4.6
Final datasets ...... 113
4.7
Interpretation of the results ...... 119

5.
Socioeconomic segregation in global cities: Analysis and findings 129

5.1
Examining the correlation between the intensity of spatial change and cities’ global network connectivity...... 130
5.2
Examining the influence of location, history, and other factors . 135
5.3
The individuality of cities: a discussion ...... 139

ix

5.3.1
Latin American cities ...... 143
5.3.2
Australian cities ...... 152
5.3.3
African cities ...... 156
5.3.4
North American cities ...... 158
5.3.5
Western European cities ...... 161
Summary ...... 167

6.
Ethnic segregation in global cities: Analysis and findings ...... 171

6.1
Examining the correlation between the intensity of spatial change and cities’ global network connectivity...... 172
6.2
Examining the influence of location, history, and other factors . 177
6.3
The individuality of cities: a discussion ...... 180
6.3.1
Australian cities ...... 183
6.3.2
Southeast and East Asian cities ...... 187
6.3.3
North American cities ...... 192
6.3.4
Western European cities ...... 197
Summary ...... 201

7.
The color of poverty: the correlation between socioeconomic and ethnic segregation ...... 205

7.1
Intersection of socioeconomic and ethnic segregation: a discussion ……...... 206
7.2
Socioeconomic and ethnic segregation: parallel or divergent changes? ...... 209
7.2.1
Examining the strength of association between socioeconomic and ethnic segregation ...... 214
7.3
Regional and local patterns of association between socioeconomic and ethnic segregation ...... 218
7.3.1
Australian cities ...... 218
7.3.2
North American cities ...... 220
7.3.3
Western European cities ...... 223
7.3.4
Southeast Asian cities ...... 224

x

Summary ...... 228

8.
The hybrid outcomes of urban change: conclusions and discussion …… ...... 231

8.1
The assessment of the ‘global/divided’ city model ...... 232
8.2
The production of spatial change in global cities ...... 236
8.2.1
Urban change and the concept of ‘hybridity’ ...... 237
8.3
So what? A research agenda for the systemic study of spatial segregation in global cities ...... 244
8.3.1
Towards a heuristic model for the comparative studies of spatial segregation ...... 245

Appendices ...... 260
i.

Tables of results ...... 260
ii. Statistical hypothesis testing ...... 268
iii. Data catalogue ...... 274

References and bibliography ...... 321

xi

Table of Figures

Figure 3-1 the production of residential segregation in socially polarized global cities due to the commodification of urban space...... 78
Figure 4-1 Artificial dataset explaining the checkerboard problem and how spatial measures can overcome the deficiency of aspatial indices...... 101
Figure 4-2 Greater Melbourne Region map shows the boundaries between inner-city and suburban areas ...... 110
Figure 4-3 The built-up area within the Greater Melbourne Region ...... 111
Figure 4-4 check for a linear correlation between the dependent and the independent variables ...... 122
Figure 5-1 alpha – beta – gamma cities and their yearly percentage of change in socioeconomic segregation ...... 131
Figure 5-2 Summary scatterplot for the correlation between the standardized change in segregation index and the cities’ global network connectivity ...... 132
Figure 5-3 global cities categorized according to their geographical region and their yearly change in socioeconomic segregation index .. 142
Figure 5-4 distribution of workers earning the minimum wage or less in Mexico City municipalities in 2000 ...... 147
Figure 5-5 distribution of workers earning the minimum wage or less in Mexico City municipalities in 2010 ...... 147
Figure 5-6 distribution of population with monthly income less and up to half the minimum wage in Rio de Janeiro in 2000 ...... 150
Figure 5-7 distribution of population with monthly income less and up to half the minimum wage in Rio de Janeiro in 2010 ...... 150
Figure 5-8 distribution of population with monthly income less and up to half the minimum wage in Sao Paulo in 2000 ...... 151
Figure 5-9 distribution of population with monthly income less and up to half the minimum wage in Sao Paulo in 2010 ...... 151

xii

Figure 5-10 distribution of population with income less than 16$ per week in urbanized Melbourne in 2001 ...... 155
Figure 5-11 distribution of population with income less than 16$ per week in urbanized Melbourne in 2006 ...... 155
Figure 5-12 increase/decrease in socioeconomic segregation in both Sunbelt and Rustbelt cities in the USA ...... 160
Figure 6-1 alpha – beta – gamma cities and their yearly percentage of change in ethnic segregation index ...... 173
Figure 6-2 Summary scatterplot for the correlation between the standardized change in segregation index and the cities’ global network connectivity ...... 174
Figure 6-3 global cities categorized according to their geographical region and their yearly change in ethnic segregation index ...... 182
Figure 6-4 Brisbane in 2001– distribution of North African and Middle Eastern migrants ...... 185
Figure 6-5 Brisbane in 2006– distribution of North African and Middle Eastern migrants ...... 185
Figure 6-6 urbanized area of Adelaide in 2001, distribution of Middle Eastern migrants ...... 186
Figure 6-7 urbanized area of Adelaide in 2006, distribution of Middle Eastern migrants ...... 186
Figure 6-8 Singapore in 2000, the distribution of Indian population ...... 191
Figure 6-9 Singapore in 2010, the distribution of Indian population ...... 191
Figure 6-10 Houston urbanized area in 2000, Distribution of Mexican, Caribbean, and Central American ...... 194
Figure 6-11 Houston urbanized area in 2009, Distribution of Mexican, Caribbean, and Central American ...... 195
Figure 7-1 alpha – beta – gamma cities and their yearly percentage of change in both socioeconomic and ethnic segregation index ... 210

xiii

Figure 7-2 cities of the dataset categorized into four groups of possible directions of change in both socioeconomic and ethnic segregation ...... 212
Figure 7-3 Summary scatterplot for the correlation between the standardized change in socioeconomic segregation and ethnic segregation in global cities ...... 215
Figure 7-4 Global cities of the dataset categorized according to broad regions and the four modes of income/ethnicity association ... 219

Table of Tables

alpha cities according to the 2010 GaWC classification of world cities ...... 96
beta cities according to the 2010 GaWC classification of world cities ...... 97
alpha cities according to the 2010 GaWC classification of world cities ...... 98
final dataset for socioeconomic segregation analysis, by census year of the data ...... 115
final dataset for ethnic segregation analysis, by census year of the data ...... 116
Top ten cities with maximum increase and decrease in socioeconomic segregation and their respective ranks ...... 139
Top ten cities with maximum increase and decrease in socioeconomic segregation and their respective ranks ...... 181
A matrix shows cities of the dataset classified according to their global status and their direction of change in both socioeconomic and ethnic segregation: ...... 213

xiv

Chapter 1

Globalization and urban change: an overview

“The city, like no other place on earth, is the global site for economic development and power, and stark socio- economic division.” (Beaverstock et al., 2011: 189)

1.1 Introduction

The social and spatial configurations of global cities are repeatedly referred to as dual (Mollenkopf and Castells, 1991), divided (Fainstein et al., 1992), fragmented (Burgers, 2002), polarized (Sassen, 1991; Friedmann, 1986), and segregated (Massey and Denton, 1993; Madanipour, 2003). One of the many explanations for the growing dualization and polarization in cities - since the 1970s - announces the declined manufacturing industries, along with the expansion of advanced producer services to be the main motivation behind urban change. Where the shift in cities’ economic base contributes to other changes in their occupational and income structures, due to the resultant increase in supply of highly-skilled and highly-paid jobs on one hand, and low-skilled and low-paid jobs on the other hand (Friedmann, 1986; Sassen, 1991). Moreover, since professional workers are increasingly moving cross-borders as their firms operate globally, while low-skilled immigrants, illegal immigrants, and asylum seekers are also needed for low-

2 | Overview paid jobs1 (Sassen, 2001; Castles, 2002). Then, polarization is expected to not only refer to the increased gap between different socioeconomic groups, but the gap is also widening along lines of race and ethnicity (ibid.). The aforementioned process – known as the ‘polarization thesis’ (Kloosterman, 1996; Hamnett, 2001; Burgers and Musterd, 2002; Mingione, 2005) – is one of the attempts to theorize the intensifying inequalities within cities in the age of globalization. The thesis has emerged as a part of larger models that describes the common economic, political, social, and spatial characteristics shared by cities since the 1970s. Among numerous attempts to theorize such ‘generally applicable model’ (Harloe and Fainstein, 1992; Mercuse and van Kempen, 2000) for cities; Friedmann’s ‘world city’ (1986) and Sassen’s ‘global city’ (1991) were – and still are - highly influential on recent research on cities and globalization (Geniş, 2007). Briefly, each of the world/global city approaches were articulated in the form of seven theses model, which comprehensively portrays the changes in the functions of cities in the global economy, as well as, the resultant transformations in cities’ social and spatial structures. Coincidently, the sixth thesis in both world and global city models is dedicated to announce socio-spatial polarization as the outcome of the economic restructuring of major cities. Initially, as a part of the models, the polarization thesis meant to explain the transformations only in cities with the highest degree of integration into the global economy - such as New York, London, and Tokyo upon which Sassen’s (1991; 2001) global city model is based. Then, as a result of the increased popularity of the global city model among scholars and researchers, the model became a framework for studying other cities in different regions of the world with different degrees of integration into the global economy. Correspondingly, the use of the term ‘polarization thesis’ stretched to include the transformations in most of the deindustrializing2 cities of the ‘Western world’ (Musterd and Ostendorf, 2005). Eventually, studying patterns of polarization, segregation, and

Overview | 3 exclusion in cities such as Johannesburg (Beall et al., 2000), Faisalabad (Beall, 2002), and Guadalajara (Audirac et al., 2012) suggested that these cities are also facing similar polarization tendencies due to either their recent integration into the global economy, or as a result of certain policies adopted by local government to improve global competitiveness of the ‘wannabe’ (Stanley, 2003) global cities. Apparently, the development of wide-ranging studies - performed over cities around the world - has to a certain degree supported the general agreement on polarization thesis to be applicable on cities, not only on the top of the global network of cities, but also on cities with less attachment to the network. However, a closer look reveals that – beside the polarization thesis - the specialized literature offers different interpretations of both the micro socio-spatial outcomes of macro economic developments, and the dynamics in which these local outcomes are produced. The variations among the different approaches revolve mainly around several dimensions, including, the number of social and spatial divisions: is the city dual or is it quartered? The changing occupational structure: is it polarized or is it professionalized? The resultant social structure: is the middle class expanding or is it diminishing? The socioeconomic gap between the rich and the poor: does it resemble patterns of inequality or polarization (see chapter two for the differentiation of the two processes)? And other variations, such as the role of urban policies in tackling/intensifying the produced divisions, the ethnic dimension of inequality and/or polarization, and the influence of local contexts and history on the process of urban change. Overall, of the key studies addressing the socio-spatial transformations in global cities, three schools of thought are distinguishable. The first school, as discussed above, announces socio-spatial polarization as the outcome of the increasingly polarized occupational and income structures of cities. The process of social polarization is reflected spatially, where the expanding upper strata of workers are increasing the demand for attractive housing conditions, leading to the gentrification of

4 | Overview inner city neighborhoods, and spatial concentration of poverty in areas with less land and property value; this urban dualism is described by Castells and Mollenkopf (1991) in their analysis of New York City. The city’s social structure is characterized by the contrast between a cohesive core of a mostly white-male professional stratum, and a disorganized periphery of everyone else with different race, ethnicity, gender, and occupation. This dualistic or bilateral view of cities’ social and spatial structures appears also in the writing of Friedmann and Wolff (1982), Friedmann (1986; 1995), and Sassen (1991; 2001). Accordingly, the polarization thesis is often criticized for reducing the complexity of social and spatial structures of cities into a simple dichotomy without paying enough attention to the ‘remaining social strata’ in the middle between the high-income professionals and the low- paid poor (Fainstein et al., 1992: 255). Moreover, this approach also builds heavily on acknowledging economic globalization as the key driving force behind urban change (Castells, 2000; Taylor, 2000; Sassen, 1991; 2006). Which explains how different cities are expected to share similar socio- spatial transformations regardless of their local contexts and histories. As in this case, the global/local interplay merely refers to the impacts of powerful macro economic developments on the unproblematic local settings within cities. From this perspective, a second school of thought emerged to patch the shortcomings of the first. The second school opposes the dualistic views of cities’ socio-spatial structures. According to Marcuse, “cities today, at least in the advanced industrialized economies of the West, are not ‘dual’, but more like ‘quartered’ cities” (1993: 355). In this view, most cities of today contain multiple residential and business sub-cities that are separated based on class, race, ethnicity and lifestyle (Marcuse, 2003). However, the divided nature of cities is still explained by the macro economic development on the global scale as in the case of the first school. The difference here is that cities’ occupational structure is not necessarily becoming polarized; it is rather professionalized. In more detail, the supply of highly-skilled and highly-paid

Overview | 5 jobs is growing, yet there is no evidence that the supply of low-skill low-pay jobs is growing in response (Hamnett, 2003). As a result, the social structure is not polarized; instead, the middle class is expanding as people with the required skills are climbing the ladder of social mobility. Yet, earning inequality is still growing between the upper/middle class professionals, and those who lack the skills and education and therefore are excluded from the labor market altogether. This approach is clearly outlined by Fainstein et al. (1992), Fainstein and Harloe (2003), Hamnett (1994; 1996; 2003), and Wilson (1987; 1996). Finally, despite how this approach differs from the polarization thesis approach, they both agree that gentrification, segregation of the poor, and segregation of disadvantaged minorities are among the key spatial consequences of growing social inequality. Also, this approach pays more attention to the role of urban policies in explaining the process of urban change, as national and local policies are affected by macro economic developments, while at the same time, they affect the pace of gentrification, land and property market, public housing, and services subsidies in general. The third school of thought calls for an all-inclusive approach to theorize growing socio-spatial divisions in global cities. In order to reach such inclusiveness, two main aspects are taken into account. Firstly, Burgers and Musterd (2002: 411) suggest that different theories dealing with urban inequality should be viewed as “complementing, rather than competing”. They explain that the polarization thesis might be applicable on a certain city, yet patterns of inequality in another city can be explained by – for example - the mismatch theory (i.e. Wilson, 1987) (see chapter two for more details). Or even within the same city, different ethnic groups can be subjected to different mechanisms of generating inequality. Still, whether it is social polarization or inequality, they agree with the first two schools of thought in acknowledging the fact that growing social division is expected to activate processes of spatial segregation (Musterd and Ostendorf, 2005).

6 | Overview

Secondly, this all-inclusive approach challenges the importance of economic globalization as the key aspect in ‘explaining’ the rising inequalities and polarization within contemporary cities, based on the credible assumption that other existing political, social, cultural, and spatial structures are creating distinct local contexts for cities, within which the socio-spatial restructuring is materialized. Noticeably, they do not question the inevitability of rising inequalities, polarization, and eventually segregation. Instead, they acknowledge the complexity of the global/local interplay by identifying a set of variables (i.e. Burgers and Musterd, 2002) or ‘local contingent factors’ (i.e. Marcuse and Van Kempen, 2000; Van Kempen, 2007) that are relevant to the process of urban change, and important to comprehensively understand – and not altering - the production of socio-spatial divisions in cities of today. Recapitulating, the discussion shows that despite their differences, the three schools of thought agree on both the source and outcome of the multifaceted restructuring of global cities. On one hand, cities’ economic function in the global economy is a key source of a number of other social, cultural, and spatial transformations. On the other hand, they also agree that the outcome of restructuring is socially divided and spatially segregated cities. Yet, between the source and the outcome of this seemingly cause- effect relation, lies a sea of differences, different explanations and different aspects to be included or excluded from the analysis of cities’ restructuring. Furthermore, the aim here is not to determine which school is adequately capturing the dynamics of urban changes. Instead, the aim is to reach a better understanding of the theories, their shortcomings, and their implications on the stream of urban research, to finally contribute to the debate through offering a comprehensive interpretation of the complex dynamics.

Overview | 7

1.2 Research problem

“How could cities with as diverse history, culture, politics, and economy as New York, London, and Tokyo experience similar transformations concentrated in so brief a period of time?” (Sassen, 2001:4)

The logic behind this intriguing question suggests that similar transformations in different cities have to be of a common root, a source that can override historical and contextual variations among cities. The discussions of the three schools of thought reveals that there is a general agreement on that source, which is the economic functions of cities in the global economy, and its parallel change in the occupational and income structures of global cities. However, this widespread agreement can be challenged based on – at least – three problematic contentions that are described as follows.

1.2.1 The general applicability of the ‘global/divided city’ model

The association between economic restructuring and the intensifying socio-spatial divisions within cities is established, not only based on the key studies discussed above, but also based on a wide range of studies that aim to examine the transformation of individual cities around the world. One example is the polarization thesis discussed above. It is stretched from describing the transformations of London and New York, to cities like Johannesburg, Guadalajara, and Faisalabad. Yet, despite the wide-ranging studies dealing with this particular topic, investigating the validity of the presumed divided nature of global cities is proved to be problematic. The problem emerges from the scarcity of comprehensive empirical evidence available to prove or disprove the general applicability of the ‘divided city’ model on wide range of cities.

8 | Overview

On one hand, most of the numerous studies - upon which the global/divided city model is generalized - are researching the socio-spatial transformations in a single case study, or in a limited dataset that consists of only few cities (for example, Sassen (1991) developed the polarization thesis based on the analysis of only three global cities). Thus, individual studies are not solely supporting the general applicability of the model on different cities. While at the same time, establishing a pattern of similar transformations in cities based on the collective results of different studies can be misinforming, due to different methodologies and different analysis’ parameters deployed by each study. Where in this case, the reliability of cross-study comparisons of the transformations in cities is compromised due to the particular scales and aspects each study is addressing. On the other hand, any attempt to deconstruct the validity of the model is also based on limited empirical evidence. In more detail, certain global cities that do not exactly fit into the theoretical description of – for example - the polarization thesis are the focus of numerous studies questioning the polarization thesis (i.e. Hamnett’s (2003) study of London). Yet, because these studies base their conclusions upon undersized datasets, as a result, the limited number of case studies does not eliminate the possibility that patterns of similar transformations among other global cities do exist. Which means that these studies are insufficient to completely disprove the applicability of the polarization thesis on a wide range of cities. Therefore, any comprehensive approach to the understanding of cities’ socio-spatial transformation has to overcome the limitations of the current research by avoiding the generalization tendencies in the literature, through examining the validity of the global/divided city model on a large number of cities, and providing a far-reaching analytical framework to explore the outcomes of cities’ restructuring in the advanced economies.

Overview | 9

1.2.2 Challenging the prominent role of globalization

As mentioned, of the many attempts to comprehensively theorize the intensifying socio-spatial divisions in global cities, Marcuse and Van Kempen (2000) and Burgers and Musterd (2002) challenged the importance of globalization as the main aspect in determining the outcomes of urban change. Marcuse and Van Kempen (2000) stated seven local contingent factors (including history, politics, race…etc.) that are highly relevant to the understanding of the resultant social and spatial divisions (see chapter two for more details). While, Burgers and Musterd (2002) built on the work of Sassen (1988; 1991; 1994) and Wilson (1987; 1996) then declared three variables that mediate the inequality within labor markets from the global to the local level, these variables are subcultural, national, and historical differences among individual cities. Despite acknowledging the importance of local contexts and histories, still, both approaches accept that inequalities are rising within cities regardless how influential local contexts are. For example, one of the conclusions highlighted by Marcuse and Van Kempen is that the spatial order of cities is characterized by the “strengthened structural spatial divisions with increased inequality among them and increasing walling between each” (2000: 249). In this sense, how are local contexts influencing the process of urban change if inequality is still its inevitable outcome? In other words, if the impacts of global processes are mediated through local contexts, then the outcome of this interplay should differ from context to context, and hence from city to city. Apparently, local contexts are perceived important only in terms of explaining the rising inequalities, instead of having an actual influence on the outcomes of change. If this is the case, then it can be argued that the global/local interplay is reduced to refer to the impact of powerful global processes on the unproblematic local settings within cities, while the impact of cities’ local contexts is minimal. This also explains the tendency in the specialized literature to generalize the increased

10 | Overview socio-spatial divisions as the main outcome of cities’ new functions in the global economy. Moreover, due to the aforementioned limitations of the available empirical studies, the local aspects that are assumed to be relevant and influential on the socio-spatial restructuring of cities are in fact drawn from analytical studies performed over individual cities. Since, any of these local factors might be relevant to the situation in one city and irrelevant to the situation in the other3, then the relevance of any given local factor to the process of urban change in general cannot possibly be confirmed based on the available limited number of case studies. Which also means that establishing a set of factors to be relevant to the explanation of the socio- spatial outcomes of different global cities is itself a form of debatable generalization. Therefore, the role of local contexts and histories in the production of urban change requires an in-depth examination, to determine whether they can have a significant impact on the resultant inequality (in terms of altering its direction and magnitude) or not. If they do have an impact, then how this global/local dynamics is taking place? What are the intermediate scales through which the interaction between a global process - such as economic restructuring - and local contexts is channeled? And to what extent this interaction can contribute to the production of various non-generalizable socio-spatial outcomes in individual global cities?

1.2.3 The spatial dimension of inequality and polarization

Massey’s metaphor of the ‘layered city’ (1984), and Harvey’s notion of ‘spatial fix’ (2001) are just few examples of how a city’s physical landscape is an essential dimension, especially when explaining changes in social relations among social entities, because these relations are manifested and only understood in space (and time) (Urry, 1985). However, so far in the discussion, the spatial dimension of inequality and polarization is not perceived as important as it is.

Overview | 11

On one hand, spatial change is implied as the natural and unproblematic side effect of social change. This automatic connection between social and spatial polarization is clearly affirmed by Friedmann (1986) in his key article ‘The World City Hypothesis’; “It is the familiar story of spatially segregating poor inner-city ghettos, suburban squatter housing and ethnic working-class enclaves. Spatial polarization arises from class polarization.” (1986: 76). One explanation of this direct connection is the commodification of residential space. As, access to housing opportunities is increasingly defined by household income (Wassmer, 2005). Eventually, social polarization is expected to lead to the exclusion of lower socioeconomic groups from the housing market – because they are unable to afford decent housing opportunities - to finally produce spatial polarization, that is, segregation. Apparently, this scenario is only possible based on the assumption that the different dynamics of local housing markets in different cities (such as home ownership rates, social mixing policies… etc.) have no real impact on the outcomes of restructuring. Then, by acknowledging the complexity of the global/local interplay, and the influence of local context and history on the production of inequality. How are patterns of spatial segregation determined away from the simplistic affordability scenario? On the other hand, despite the importance of urban space, the focus on social polarization and inequality led to the scarcity of comprehensive theoretical frameworks for systemic study of spatial segregation in global cities as an outcome of the global/local interplay. Similar to Burgers and Musterd (2002) model of inequalities in post-industrial labor markets, an all- encompassing scheme is required to study; patterns of spatial segregation, the factors involved in the production/or tackling of segregation in global cities, and the correlation of these patterns to other restructuring processes on global, regional, national, and local levels.

12 | Overview

1.3 So what? One step towards breaking the vicious cycle

“Segregation causes negative impacts on the cities and lives of their inhabitants. It imposes severe restrictions to certain population groups, such as the denial of basic infrastructure and public services, fewer job opportunities, intense prejudice and discrimination, and higher exposure to violence.” (Feitosa et al., 2007)

Spatial segregation is perceived as a result of social inequality, yet, there are growing concerns that spatial segregation of the poor and ethnic minorities are contributing to the reproduction of inequality (Wilson, 1996; Musterd, 2005). This is due to the spatial concentration of poverty, which is associated with other social problems including violence, crime, unwed child bearing, divorces, single parenting, low educational achievement… etc. (Wilson 1987; Soja, 1989; Massey and Denton, 1993; Andersen, 2002; Wassmer, 2005; Varady, 2005), while the isolation of ethnic minorities intensifies issues of estrangement, prejudice, and otherness (Sennet, 1970; Goldsmith, 2000; Peach, 1996). Overall, spatially segregated groups are most likely to be excluded from mainstream society; their limited educational opportunities also limit their employment opportunities, which diminish their chances of social upward mobility (Wilson, 1987; Kazepov, 2005; Boal, 2005; Massey and Denton, 1993; Madanipour et al., 1998). And once this cycle starts, excluded households become trapped in the poor neighborhoods, due to the vicious cycle (Jargowsky, 1997) of limited social options that leads to limited spatial options and so on (Madanipour et al., 1998).

Overview | 13

For that matter, in recent years, issue of spatial segregation has gained a particular attention in urban policies, especially in Europe and the US (Friedrichs et al., 2003; Musterd, 2005). These policies focus mainly on tackling spatial segregation by mixing neighborhood population and avoiding large concentrations of poor households, and/or immigrant households. However, evidence from cities around the world suggests that mixing neighborhoods policies are not always successful. Imposing ethnic quotas in Singaporean neighborhoods failed to prevent ethnic regrouping (Van Grunsven, 2000; Sin, 2003), while urban regenerating programs in Amsterdam resulted in ‘zero-sum’ outcome (Musterd and Andersson, 2005) (see chapters five and six for more details). In light of the above, spatial segregation is becoming an increasingly pressing issue, with serious implications on both the segregated groups and the city as a whole. The aim here is not to provide recommendations or alternative solutions for growing spatial segregation in cities. Instead, it can be argued that a better understanding of the production of segregation, its source, levels, and patterns, can assist in reaching a creative and more effective approaches for tackling segregation, and breaking the vicious cycle of persisting exclusion and segregation.

1.4 Research objectives

This study aims to contribute to the scholarly debate by overcoming certain limitations of the available research on the socio-spatial transformations in global cities. And due to the author’s background as an urban designer and planner, the study pays a particular attention to the spatial dimension of polarization and inequality, which is the change in spatial segregation patterns based on socioeconomic status and ethnic background of the inhabitants of global cities. The main objective of the study is to examine several widely accepted assumptions in the global

14 | Overview city literature. In order to reach this main objective, the research also aims to: • Investigate the complex process of restructuring based on the available research, while focusing on; the different mechanisms of producing socially and spatially divided cities, the suggested modes of interaction between global processes and local contexts, and the generalization tendencies in the specialized literature. • Overcome the shortcoming of available research by providing a solid empirical ground for investigating the general applicability of the socially and spatially divided city model, through an analytical study performed on unprecedentedly large dataset of 91 cities. • Evaluate the influence of local contexts of individual cities on the outcomes of spatial changes. To conclude whether local contexts are actually able to reverse, modify, or even intensify the outcomes of global restructuring process or not. • Based on the analysis results, the study aims to offer a research agenda for interpreting the socio-spatial transformations in global cities. The proposed future research aims to develop an alternative research approach that acknowledges the complexity of the global/local interplay, the influential role of local contexts, and the individuality of cities. • Finally, the study also aims to build on available theoretical frameworks to develop a scheme for studying the changes in socioeconomic and ethnic segregation within global cities, through identifying the relevant economic, political, social, cultural, and historical aspects that are impacting the direction and magnitude of change in spatial segregation level. The scheme does not only aim to capture the complexity of the global/local interplay, but also to facilitate cross-city comparisons of spatial change.

Overview | 15

1.5 Research Questions

In view of the above, in order to reach a fresh perspective on the process and outcomes of urban change, while at the same time, critically debate the long-standing association between processes of globalization and rising inequality, polarization, and segregation in global cities. The research attempts to reach an answer for this main question:

Is the global city a divided city?

More specifically, does the city’s high level of integration in the global economy contribute to socially and spatially divided internal structures of this particular city? As suggested by the different approaches for theorizing the socio-spatial transformations in global cities, the expected answer to this question is yes, due to a complex process of restructuring where the macro economic changes are impacting the political, social, cultural, and spatial structures of cities and leading to the fragmentation of cities on the micro level of the municipalities and neighborhoods. However, as explained earlier, this acknowledged association between cities’ global status and their level of social and spatial divisions requires verification. In the process of exploring the answer for this main research question, the research is faced by a number of secondary questions that contribute to a better understanding of the complex process of restructuring of global cities, and assist in reaching a comprehensive answer to the aforementioned question. The secondary questions includes: • What are the impacts of economic restructuring on other political, social, cultural, and spatial settings of global cities? • How is the multi-dimensional restructuring process mediated from the global to the local scales of urban development? And what are the in-between scales of structural organization between the global and local level?

16 | Overview

• How are the changes in economic, political, social, and cultural aspects manifested in space? • What are the main contextual differences expected to influence the process of urban change? • Do global cities share similar patterns of increase in spatial segregations as suggested by the global/divided city model? If not, what are the aspects leading to divergent patterns of change in global cities? • Does context matter? In other words, is the restructuring process producing a hybrid spatial outcome that carries properties from both the new economic functions of cities, as well as local and historical particularities of each city? Or can the process be reduced to powerful global process that are impacting unproblematic local context to produce a standardized increase and intensification of divisions within the global city?

1.6 Structure and methodology

In an attempt to offer an inclusive theoretical construct for the association between patterns of socio-spatial divisions within cities and their economic functions, the argument presented in this introductory chapter is further developed in the subsequent chapters as follows: In chapter two, the study aims to investigate what is new about the divided cities of today, it presents the possible sources and manners of the intensified socio-spatial divisions within cities under conditions of globalization. The chapter starts with a clarification of the basic concepts of inequality and polarization, followed by a brief description of a number of global processes of deindustrialization, despatialization of production, competitiveness, and the rise of network society, which are contributing to the mounting importance of cities as control centers of the global economy beyond national borders, as well as being responsible for raising levels of

Overview | 17 inequality among and within nations and cities. Then, the chapter discusses in more detail the three different approaches for explaining the growing inequality, social polarization, and segregation as a direct outcome of the changing economic functions of cities in the global economy. Finally, the discussion in chapter two outlines certain shortcomings of the available research including the presumed general applicability of the divided city model, the proposed dichotomy between the global and local scales, and the unproblematic views of space. In chapter three, the discussion focuses on the spatial dimension of polarization and inequality, which is identified in the previous chapter as socioeconomic and ethnic segregation. The chapter starts with highlighting the importance of space (and time) to the understanding of other economic and social changes that are producing, and in turn, reproduced in space. The importance of space is justified through a theoretical discussion of spatial structures of accumulation regimes. Then, in the light of the global city literature, the causes behind the increasing spatial segregation in global cities is explained through the discussion of the impact of global competitiveness on urban space in both business and residential areas. The chapter shows that the commodification and despatialization of urban space in already socially polarized global cities can lead to the exclusion of the disadvantaged groups from the competition over decent housing, and create issues of differential access to the housing market based on households’ ability to pay for desired housing opportunities. Finally, this affordability scenario for explaining spatial segregation is criticized based on the possible variations in local housing dynamics of individual cities. Chapter four introduces the empirical study, upon which the established association between the global economic functions of cities and intensified levels of spatial segregation within them is put to test. As noted earlier in this chapter, the aim of the study is to overcome the limitations of available research and examine the ‘global city/divided city’ paradigm based on an extensive analytical study.

18 | Overview

Therefore, on one hand, the study reviews several global/world cities classifications, and based on the advantages and limitations of each classification, the GaWC 2010 classification of world cities is chosen to be the primary dataset for the study (total of 178 cities). It is worth mentioning that the GaWC classification is criticized for narrowly focusing on cities’ economic functions as the basis for categorization of cities (see Robinson, 2005), yet this narrow focus contributed to the nomination of the GaWC classification to be highly suitable for this study. On the other hand, the chapter also reviews several indices for evaluating the change in socioeconomic and ethnic segregation levels in cities of the dataset. The findings of the analytical study are built on the values of ‘spatial multi-group dissimilarity index SD(m)’ created by Wong (2003), and the calculations of the index is generated using an ArcView tool provided also by Wong (2002; 2003). Moreover, the chapter shows the required data type for the analysis, the limitations of cross-city comparisons, and the criteria for defining the geographical boundaries for individual cities of the dataset. Finally, based on data availability, the final dataset is comprised of total of 91 cities, categorized into 32 alpha, 31 beta, and 28 gamma global cities. Also, the chapter provides a backdrop on the different statistical approaches that are deployed to interpret the analysis findings; these approaches include the linear regression analysis, and one-way analysis of variance ANOVA. Chapters five, six, and seven present the findings and general observations of the analytical study. Chapter five links level of change in socioeconomic segregation in any city over time, to its position as global city based on the GaWC classification. The collective results show no association between the agglomeration of producer services firms in one city and the intensified spatial segregation. On the contrary, cities of the dataset show divergent changes in socioeconomic segregation in terms of direction (increase or decrease) and magnitude of change. Moreover, cities that share regional or national contexts also show inconsistent changes. For that matter,

Overview | 19 individual cases are discussed to explain the lack of shared pattern of change among cities, and to evaluate the influence of the contextual particularities of each city on its corresponding change in spatial segregation. Similarly, chapter six shows parallel results for the change in ethnic segregation levels in cities of the dataset. Therefore, individual cases are also discussed to reveal the impact of historical migratory flows and discriminatory practices on current patterns of ethnic segregation. While chapter seven compares changes in socioeconomic segregation to changes in ethnic segregation within the same city over the same period of time. The aim of the comparison is to examine the influence of the contingency of ‘race/ethnicity’ on patterns of poverty concentration and vice versa. The chapter concludes that – in most cases - the increase (or decrease) in socioeconomic segregation in any city is associated with a corresponding change in ethnic segregation, which confirms the influential role of existing conditions (the ethnic background and/or socioeconomic status of the segregated groups) as important contingencies that affect the direction and magnitude of the expected change in spatial segregation levels. And finally, chapter eight presents concluding remarks and discussion of the analysis’ findings in an attempt to answer the questions that have been raised throughout the thesis. The chapter starts with a summary of the argumentation of the research, followed by a discussion of the key findings and their implications on the long-standing agreement on the ‘global city/divided city’ notion. On one hand, the divergent changes showed by cities of the dataset eliminates the possibility that cities’ economic functions in the global economy are associated with a common trend of growing socio-spatial divisions within them. On the other hand, the discussion of individual cases confirms that patterns of change in each global city are a product of global processes that mélanges with historical, political, cultural, and geographical variables across different levels of the urban hierarchy from global to local, to finally create specific and non-generalizable spatial outcome for each city. Which, not only confirms the influential role of local

20 | Overview contingencies on the process of urban change, but also suggests that it is no longer possible to perceive the global/local interplay as a unidirectional cause-effect relation, in which the macro global processes are expected to reshape local settings of cities. For that matter, based on the discussion of individual cities of the dataset, the study suggests a number of research questions, which can be considered the basis for approaching the socio- spatial transformations in global cities from a different perspective, which acknowledges the complexity of cities restructuring and the influence of local situations of cities on the outcomes of restructuring. An example for such prospects for future research is the development of a heuristic framework for analyzing and explaining the changing spatial segregation patterns in a particular local context based on the influential local ‘modifiers’ identified during the analytical study. Finally, the chapter acknowledges the constraints and the limitations of the study.

Overview | 21

Notes

1 Due to poor immigrants’ low bargaining power in the labor market (Waters, 1995). 2 Since polarization is strongly linked to the process of deindustrialization, polarization tendencies are most likely to materialize in western cities, as the manufacturing industries shifting spatially from those cities to other peripheral regions of the world. 3 Such as the provision of public housing, change in the public housing sector in Singapore is most likely influential on levels of socio-spatial polarization and segregation as 86% of Singapore’s population (in 1998) resided in public housing (Sin, 2003). While the aspect of public housing is not as strongly linked to levels of polarization in Brussels as it is in Singapore, where the role of the state in providing housing in Belgium have always been minimal (Kesteloot and Cortie 1998).

Chapter 2

2 Globalization and the production of inequality

“It is within the global cities that we can see most clearly how social polarization has exacerbated old inequalities and created new forms of inequality. The global city contains the cutting edge of the new order but also the worst of the old order […], now magnified and made permanent.” (Munck, 2005: 64)

Cities have always been divided (Massey, 1996; Caldeira, 2000; Marcuse, 2003; Beaverstock, 2011). Historical examples of the Jewish ghetto in medieval Venice (Madanipour et al., 1998), the imperial European island in Canton1 in 19th-century China (Marcuse and Van Kempen, 2002), and many other examples show that divisions are not unique characteristic of cities labeled as post-industrial, post-Fordist, world, or global cities. If this is the case, what is so different about the divided city of today? On one hand, according to Reichl (2007), the difference is mainly the intensity of the divisions, as the stark contrast between wealth and poverty has been exacerbated during the 1980s. This assumption is supported by the proliferation of different terms that describe how the socio-spatial divisions within cities are perceived to be different – in intensity and form - than those in previous decades, such as the new underclass (Musterd, 1994), advanced

24 | Theoretical background marginalization (Vacquant, 1996), and hyperghettoization (Marcuse 1996). On the other hand, for Badcock (1997) and Martinotti (2005), the main difference is not the nature of divisions themselves. Instead, it is the source of divisions, as cities are faced by deep structural changes, to the extent that the change in people’s life on community level cannot be understood separately from developments that take place on higher levels, on the national, regional, and global scales (Kazepov, 2005; Van Kempen, 2007). So, is it the nature of divisions or is it the source of these divisions that produces distinct socio-spatial changes in contemporary cities? According to Marcuse (2003: 271), since the 1980s, both the “source and manner” of divisions within cities are profoundly different than those in previous decades. This chapter investigates the different approaches for theorizing the source and manner of socio-spatial divisions in global cities. But before proceeding with the literature review, a clarification of the basic concepts is required. So far in the chapter, the use of the term ‘divisions’ was made for two reasons. Firstly, the term is general, yet suggestive and widely used in the world/global city discourse (see the divided city (Fainstein et al., 1992)). Secondly, the choice of the term was made to avoid the use of other terms, such as, inequality and polarization without a proper differentiation between them. In fact, there is a growing concern about the misperception and confusion between both terms in the specialized literature. The mix-up here is not derived from the lack of precise definition for inequality and polarization (for detailed discussion and definitions see Kloosterman, 1996; Hamnett, 2001; Marcuse, 2005). Instead, it is caused by the combined and – sometimes – the interchangeable (Wessel, 2000; Maloutas, 2007) use of the terms, which contribute to different perceptions of the correlation between the different processes they signify. In brief, the popularity of the polarization thesis (discussed in chapter one) has contributed to the widespread use of the term polarization, to the extent that the term became ‘an all-purpose general signifier’ of growing

Globalization and production of inequality | 25 urban inequality and social division in urban areas (Hamnett, 2001: 167). However, polarization and inequality are by no means synonyms. In fact, polarization shall be perceived as a sub-theme under the umbrella of inequality, or as Castells (1999: 7) puts it “polarization is a specific process of inequality”. This perception of polarization matches Kloosterman (1996) differentiation of the terms, where income inequality refers to the dispersal between certain levels of income, while polarization has to include a growth of the upper and lower ends of the income distribution on the expense of a shrinking middle. In this sense, income inequality can be pictured in several geometrical forms, such as, the increased gap between the top and the bottom of a pyramid (Pahl, 1988), an egg (Boter et al., 1988) - that is sometimes referred to as an onion (Bornschier, 2008), or an hourglass (Pahl, 1988; Boter et al., 1988; Marcuse, 1989; Hamnett, 2001). Yet, the hourglass is the only shape of inequality that can be referred to as polarization. Because literally, polarization means: “a movement toward the poles of a given distribution.” (Hamnett 2001: 169). Therefore, inequality can take place with or without polarization, as Fainstein and Harloe (2003) concluded that inequality is increasing in both New York and London, however, “polarization is not the most accurate description of this phenomenon in that, as a term, it fails to capture the improving position of a large proportion of the middle mass” (p: 165). Moreover, in order to avoid the problematic misuse of the term polarization, Hamnett (2001) stresses on the need to specify what exactly polarization refers to in different theoretical frameworks, is it class polarization, income, employment, or spatial polarization. This distinction between inequality and polarization is particularly important at this point, as the next section presents the three different approaches for explaining the association between cities’ economic functions in the global economy, and the growing socio-spatial divisions within them. As shown in the previous chapter, the approaches agree on the source of divisions, yet they differ in their interpretation of the manners of

26 | Theoretical background these divisions, a discussion in which the processes of polarization and inequality are central.

2.1 Approaching urban inequality: the three schools of thought

“The form and extent of a city’s integration with the world economy, and the functions assigned to the city in the new spatial division of labor, will be decisive for any structural changes occurring within it.” (Friedmann, 1986: 70)

The new structure of economic activity is only one facet of the complex economic, political, social, and physical restructuring of cities (Friedmann and Wolff, 1982; Mollenkopf and Castells, 1991), yet the above statement suggests that the economic factor is highly influential on the other dimensions of restructuring (Sassen, 1991). Hypothetically, The economic functions of cities are able to alter the structure of metropolitan labor markets, as well as the physical form of cities (Friedmann, 1986). But in what way have cities’ functions changed in the first place? In the light of the world/global city discourse, a set of macro economic developments - took place since the early 1970s - has contributed to not only the changing economic functions of cities, but it can be said that they produced a new type of cities (Beaverstock et al., 2011). Concisely, major cities witnessed a drastic decline in the manufacturing industries - such as steel production, textiles, heavy machinery, along with the rise of service industries - such as legal, accounting, and banking services (Beauregard and Deitrick, 1995; Sassen, 2001). Where this deindustrialization process was marked by the spatial shift of manufacturing industries from western cities to third world cities (Sassen, 1990; Beck, 2000). Then in the 1980s, the rise of the new communication technology has facilitated another global process

Globalization and production of inequality | 27 that is the ‘despatialization of production’, because the communication and information technologies opened the door for production sites to be located offshore, remotely from their main headquarters (Bluestone and Harrison, 1982, Friedman, 2005). Accordingly, the spatial dispersal of production sites was juxtaposed by the agglomeration of command and control functions in certain global cities (Soja, 1989; Sassen, 2001). Harvey (2005) argues that despatialization of production became a necessity, because certain locations generate higher profit rates since they offer low-cost raw materials and/or low-wage labor. However, in order to achieve such level of mobility, markets for both capital and commodities have to be “open across the world so that surplus capital in one territory can easily circulate into other territories” (2005: 94). Accordingly, the need for open markets and the boost in information technologies (connecting administrative centers, production sites, service activities, and their research centers) have led consequently to unprecedented level of mobility of capital beyond territorial or national borders (Soja, 1989). As a result, mobility of capital has intensified the geographical competition over international investment where every city and region has to compete with the others to attract and retain investment (Harvey, 1985b; Kesteloot, 2000; Soja, 1989; Fainstein, 2010). Both processes of deindustrialization and despatialization have marked the end of the Fordist regime of the golden sixties, and the rise of a new flexible economic regime dominated by financial corporations, multinational firms and information technologies, and characterized by flexible modes of production and consumption (Castells, 1999; Kesteloot, 2000; Kazepov, 2005), where the integrated Fordist assembly line has been replaced by segmented production processes taking place at many different and dispersed locations (Soja, 1989; Friedman, 2005). This spatial dispersal between production sites and their command centers has resulted in two interrelated outcomes. Firstly, major cities have gained a new strategic role beyond their long history as centers for international trade and banking (Sassen, 2001;

28 | Theoretical background

Haussermann and Haila, 2005). They became command and control centers that organize the global economy (Sassen, 2001; Haussermann and Haila, 2005). In this contexts, cities are not only ‘labor pools’, in which labor force is created and reproduced (Urry, 1985), but they are also particular sites of production of specialized services and financial innovations, as well as markets for the products and innovations produced (Sassen, 2001). Overall, cities and their regions are the true basis of the modern economy (Jacobs, 1992[1961]). Where, cities that are “instrument” for the control of production and markets (Friedmann and Wolff, 1982: 312) are hosting higher shares of corporate headquarters, international financial institutions, global transport and communications, and high-level business services (Friedmann, 1986). Secondly, the rise of transnational corporations resulted in an urban hierarchy based on networks and connectivity instead of territories, as Castells (1996) puts it, the shift in the world’s spatial order is a shift from ‘a space of places’ to a ‘space of flows’, a concept in which flows of capital, information, and organizational interactions are taking place through electronic circuits connecting nodes and hubs. In this context, the perception of cities has shifted from being basing points for global capital to nodal point in larger network (Beaverstock, 2011), where various types of global networks intersect (Castles, 2002). According to Sassen (2006: 477), “this type of city [the global city] is not simply one step in the ladder of the traditional scalar hierarchy that puts cities above the neighborhood and below the national, regional, and global levels. Rather, it is one of the spaces of the global, and it engages the global directly, often bypassing the national.” Consequently, the increasingly borderless economy and the heightened interregional competition over investment have contributed to an uneven economic growth of different economic sectors (Castells and Mollenkopf, 1991), and thus, growing inequalities on different scales among and within nations and cities (Badcock, 1997). Massey (2008) explains the increased

Globalization and production of inequality | 29 inequality to be a result of shifts in demand and supply in the global market. On one hand, the demand for financial capital, land, and energy is immense while supplies are limited, which give leverage to those who possess these resources. On the other hand, the demand for labor remains limited even in a global economy but the supply is immense, which exert downward pressure on physical labor. Similar logic of winners and losers of the global economy is stated by Castells (1998:162); “globalization proceeds selectively, including and excluding segments of economies and societies in and out of the networks of information, wealth and power that characterize the new dominant system”. Recapitulating, the rise of information and service economies, global dominance of transnational corporations, flexible production and assembly, accelerated capital mobility, and heightened intercity competitiveness have contributed to, on one hand, the concentration of specific economic functions in what is known as global cities, and on the other hand, growing inequality among and within cities. The rising concerns about inequality - as a side effect of economic restructuring - was reflected clearly in an over- expanding literature exploring the dynamics in which different forms of inequality are produced. This chapter categorizes the different concepts in the literature into three main approaches. Firstly, ‘the occupational polarization and urban dualism approach’, this approach is established based on the commonalities among the hypotheses of the “world city” (Friedmann and Wolff, 1982; Friedmann, 1986), the “global city” (Sassen, 1988; 1994; 2001; 2006), and the “dual city” (Mollenkopf and Castells, 1991). Secondly, ‘the occupational professionalization and inequality approach’, based upon the concepts of the “divided city” (Fainstein et al., 1992), the “unequal city” (Hamnett, 1994; 2001; 2003), the “mismatch theory” (Wilson, 1987; 1996), and the “quartered city” (Marcuse, 1989; 2003).

30 | Theoretical background

Finally, ‘the contextual particularities approach’, presents several attempts to provide an all-inclusive approach for understanding urban inequality, based on the writings of Burgers and Musterd (2002), Van Kempen (2007), and Marcuse and Van Kempen (2000). As highlighted earlier, the different approaches acknowledge the previously illustrated functions of cities in the global economy as the main source of inequality, yet they differ in explaining the impacts of the new functions on the supply and demand for labor in these cities, which led to the different assessments of the occupational, income, and social structures of the global city.

2.1.1 Occupational polarization and urban dualism approach

a. Changes in cities’ occupational structures The basic argument of the first school of thought is that cities’ economic functions have led their labor force to be dichotomized (Friedmann, 1986) or polarized (Sassen, 1991). The polarization of employment here refers to the increased number (or percentage) of the high- skilled and high-paid professionals running the service industries. While at the same time, a parallel change is the increased number (or percentage) of low-skilled and low-paid workers needed for the supporting personal services, tourism, and entertainment industries. Beyond this simple dichotomy, in fact, both the ‘world city’ and the ‘dual city’ acknowledge the complexity of the occupational structures of global cities. In more detail, Friedmann and Wolff (1982) in their introduction for the ‘world city’ hypothesis, they listed six possible clusters of employment in world cities. Yet, the dichotomy continued between, on one hand, a primary clusters of business professionals (the transnational elites), and on the other hand, all the remaining clusters including; a secondary cluster of workers serving the primary cluster in terms of construction, transportation, private police, domestic services and tourism (hotel services, restaurants, luxury shopping). Also, there is a third cluster of

Globalization and production of inequality | 31 manufacturing workers, followed by a cluster of government services workers, and the last two clusters are the informal/street economy workers and the unemployed. Similarly, Castells and Mollenkopf (1991) in the ‘dual city’ offer a parallel description of the most distinctive occupational strata in New York City, they listed six strata led also by an upper stratum of executives, managers, and professionals, then followed by an army of clerical workers, low-skilled workers, immigrant workers, public services workers (health, education… etc.), and finally adults outside the formal labor force. They agree with Friedmann and Wolff that despite the complexity of cities’ occupational structures, the employment polarization persisted between the upper stratum of professionals on one side, and all other strata of workers on the other side. Sassen (1991) - in the ‘global city’ - explains that the growing number of low-skilled and low-paid workers is not merely a side effect of the economic restructuring of the global cities. Instead, the agglomeration of producer services in these cities has led low-paid jobs to become essential. Directly, competitiveness among transnational firms put them under pressure to increase their profits by reducing the cost of their services through subcontracting or employing undocumented immigrants (Soja, 1989; Martin and Miller, 2000; Sassen, 2001). Also, low-skilled worker are needed for purposes of transportation, cleaning, construction, catering, and other supporting services (Sassen, 2001; Castles, 2002). Indirectly, producer services support high-income gentrification, which in turn requires large numbers of low-wage service workers to support restaurants, shopping and entertainment activities (Logan, 2000; Sassen, 2001). Moreover, since world/global cities are major destination for both domestic and foreign immigrants. The influx of immigrants has contributed to the immense growth of supply of low-skilled workers in these cities, which in turn, caused a downward pressure on wages (Friedmann, 1986), especially with the decline in manufacturing sector, as changing supply and

32 | Theoretical background demand of labor market has led to changes in the wages structures of relatively well-paid jobs to become increasingly low paid. One example of this change is the 3D jobs (dirty, demanding, and dangerous, or sometimes dirty, dangerous and difficult (Castles, 2002)), these jobs were originally offering higher wages to attract workers who do not want to be employed in this sector because of their dire and unhealthy working conditions (Lee, 1997). However, due to the increased supply of workers (immigrants, unemployed...) who are excluded from the labor market and unable to attract other kinds of work, consequently, the 3D-jobs became poorly paid despite its high physical and mental risks (ibid.). Also, the increased flows of immigrants have contributed to the expansion of ‘informal’, ‘floating’, or ‘street’ economic activities, such as sweatshop manufacturing, off-the-book childcare, and other unregistered, non-taxpaying activities (Castells and Mollenkopf, 1991; Sassen 1991). Friedmann and Wolff (1982) explain in more detail that a significant number of new comers to the city - looking for job opportunities - do not necessarily find legitimate employment, which lead to their exclusion from the formal and regulated labor market, and their engagement in informal or - even sometimes - criminal activities. Similarly, Sassen (2001) also ties the growth of informal economy to the flows of immigrants to the global city, she notes that informal work tend to be located in densely populated areas with very high share of immigrants, more than anywhere else in the city.

b. Socio-spatial outcomes of restructuring As a result of the changes in cities’ occupational structures; the world, global, and dual city approaches agree that the social and the spatial structures of cities are expected to reflect the growing gap between the high- skilled/ high-paid professionals and the low-skilled/ low-paid workers. Socially, Sassen (2001:343) concluded that ‘a new class alignment is being shaped’ in the global city, marked by declining middle class and growing both upper and lower stratum of workers. The same logic is introduced by the world city and the dual city but with using different

Globalization and production of inequality | 33 vocabulary, both approaches agree on the complexity of the occupational structures, as shown, economic restructuring generated at least 6 clusters of employment, yet the resultant social structure suggested by both approaches is bilateral, where society is dominated by two opposing forces. On one hand, there is the ‘upper circuit’ (Friedmann and Wolff, 1982) or the ‘upper social strata’ (Castells and Mollenkopf, 1991) of professional. While on the other hand, there are the ‘underclass’ (Friedmann and Wolff, 1982) or ‘the remaining social stratum’ (Castells and Mollenkopf, 1991). Spatially, since the upper stratum of professionals is the dominant class in the city, Friedmann argues that the city is arranged to serve their life styles and daily needs. In this sense, Sassen (1991) pins down the process of gentrification of inner city neighborhoods to be the most visible outcome of the growing number of professionals running the service industries, as the highly-skilled professional and the ‘transnational elite’ require attractive residential conditions (Kesteloot, 2000), as well as restaurants, shopping centers, etc. While the process of gentrification is also accompanied by spatially concentrated poverty and physical decay of poor neighborhoods (see chapter three for more detail about gentrification and displacement of the poor) Similar to Sassen’s concept of the gentrified neighborhood, Friedmann and Wolff (1982) provide another description for the ‘upper circuit’ neighborhoods. In fact, they argue that the dual social structure of the world city resulted in a contrast between spatially separated ‘citadel’ of the transnational elite, and ‘ghetto’ of the underclass. They picture both cities as symbols of inequality and class domination: “With its towers of steel and glass and its fanciful shopping malls, the citadel is the city’s most vulnerable symbol. Its smooth surfaces suggest the sleek impersonality of money power. Its interior spaces are ample, elegant, and plush. In appropriately secluded spaces, the transnational elites have built their residences and playgrounds: country clubs and bridle paths and private beaches. The overcrowded ghettos exist in the far shadows of the citadel, where it is further divided into

34 | Theoretical background racial and ethnic enclaves. Some areas are shanty towns. None are well provided with public services: garbage does not get collected, only the police in their squad cars are visibly in evidence. In many places, ghetto residents are allowed outside their zones only during working hours: their appearance in the citadel after dark creates a small panic.” (1982: 325). A similar image of the spatially separated cities within the city is also outlined by the Castells and Mollenkopf (1991: 402), their description of New York City reflected a widespread urban dualism resulted from occupational polarization and the growing income gap between the upper stratum and the remaining social strata. On one hand, there is a cohesive core of predominately white male professionals, a core where the advanced corporate services are concentrated, and the residential spaces are exclusively occupied by the new urban elite. While on the other hand, there is a disorganized periphery made up of diverse groups, who are differentiated by race, ethnicity, gender, economic sector, household composition, and the space they occupy.

c. Criticism The dualistic views of the city adopted by the world, global, and dual city approaches are criticized based on several contentions. Firstly, the dual city metaphor is considered an oversimplification of cities’ complex social and spatial structures (see Marcuse, 1989; 2003). However, Mollenkopf and Castells (1991) repeatedly acknowledge the limitations of the concept of dual city, they admit that New York is more than a prosperous White Manhattan surrounded by Black/Latino underclass boroughs. Yet, they defend the dual city concept because despite its limitations, it is useful as an ideological notion captures the contradictions of the global city, or a general idea describes inequality, exploitation, and oppression in cities, especially that the concept attracted the attention of policy makers and made popular through the media. That being said, Castells and Mollenkopf acknowledge the complexity of cities’ socio-spatial structures, yet they argue that this complexity does not contradict the dual nature of the city, they state: “what

Globalization and production of inequality | 35 those outside the core economic strata have in common is precisely their diversity, their heterogeneity, their externality to the well-structured social core directly connected to the strategic command centers of the corporate economy.” (1991:403). In this sense, the dichotomy is not simply between rich and poor, it reflects the contradiction between those who are included in and those excluded from the benefits of the economic growth of the global city. Secondly, this approach is also criticized for suggesting the applicability of the occupational, social, and spatial polarization thesis on different cities around the world (see Hamnett, 1994; 2001; 2003). The logic behind such general applicability of the thesis lies in acknowledging the economic restructuring as the key source of a direct and predefined socio- spatial outcome. Friedmann and Wolff (1982) argue that it is only a matter of time before the impacts of restructuring appear in different cities due to their deepening integration in the global economy. Once these cities started to serve as a headquarters location for transnational corporations, then their occupational structures will change accordingly, along with their social and spatial structures, towards more polarization. In this sense, local variations among cities including history, urban policies, subcultural differences, and other national and regional particularities are perceived irrelevant (at least in comparison to the economic factor) to the process of restructuring as well as its socio-spatial outcomes. Also, it can be argued that the impacts of the presence of transnational corporations in the city are overstretched. As the services industry may represent only a small part of urban employment Pretecille (1994), then in this case, socio-spatial polarization is not necessarily the expected outcome of economic restructuring. In the light of these shortcomings, a second school of thought appeared to offer another interpretation of the divided nature of the global city.

36 | Theoretical background

2.1.2 Occupational professionalization and inequality approach

The second school of thought agrees with the first school that economic restructuring is contributing to the divided nature of global cities, through also impacting cities’ occupational structures. However, the disagreement here is how these occupational structures are transformed, and how the changes are translated into social and spatial divisions among the city inhabitants. Concisely, the second school of thought rejects the bilateral perception of cities suggested by the first school, based on the assumption that the outcomes of economic restructuring are far more complex than a simple dichotomy between the rich and the poor.

a. Changes in cities’ occupational structures ‘The mismatch theory’ (Wilson, 1987; 1996), ‘the divided city’ (Fainstein et al., 1992), ‘the unequal city’ (Hamnett, 2003), and ‘the quartered city’ (Marcuse, 2003) have acknowledged the fact that the concentration of service industries in cities generates a demand for highly skilled professionals to operate the highly specialized service firms. However, unlike the first school, they argue that this ‘upgrading’ (Wilson, 1996; Van der Waal, 2010) or ‘professionalization’ (Hamnett, 2003) of the labor market is not necessarily accompanied by the increased supply of low- skill and low-wage jobs. Instead, the main argument here is that people with the required qualifications for highly skilled jobs are moving up the social ladder, leading to the expansion of the middle class, not its diminishing as suggested by the first school. Fainstein and Harloe (2003) argue that this middle class - which constitutes a large section of the metropolitan population - is repeatedly overlooked in the analysis of global cities, especially with the propagation of the polarization and duality approach in which workers are either highly-paid or under-paid. Another outcome of the professionalization of the labor market is the exclusion of people lacking the required qualifications – i.e. the poorly educated - from the labor market (Hamnett, 2003). As a result, the excluded

Globalization and production of inequality | 37 groups form an emerging underclass (Wilson, 1987; 1996), generated by economic restructuring, and combined of the unemployed and workers engaged in the informal economic activities. In this scenario, race, ethnicity, and gender play a major role in determining individuals’ positions in the labor market. According to Fainstein and Harloe (2003) study of London and New York, the upper stratum of the executive and professional is predominated by white males, as racial and gender discrimination is still restricting access of certain groups to the highly paid jobs even if they possess the required skills. Similarly, Wilson’s mismatch theory (1987; 1996) pays a special attention to the position of African Americans in the labor market across cities of the US. He concludes that black joblessness can only be explained through understanding a complex web of factors ranging from the historical discrimination against African American, their inadequate access to jobs or good quality schools, their lack of exposure to mainstream social networks, the lack of role models, to the shift in the American and the global economy. Similar observation is made by Hamnett (2003) in his study of London, he agrees with Fainstein et al. (1992) that levels of unemployment among Blacks and Bangladeshis are double or triple those among Whites. Yet Hamnett refuses to generalize that all ethnic minorities are of a weak position in the labor market, as many second-generation Indians and Chinese have done well at school and managed to succeed in several economic sectors, including IT, pharmacy, and accountancy. Wills et al. (2010) agree with Hamnett that particular ethnic groups are concentrated in specific parts of the economy. In their study of London, they found that one-third of London’s Filipino population is employed in health and social care, while Brazilians are more concentrated in cleaning services. In the light of the above, the upgrading of the labor market along with the emergence of the excluded underclass has contributed to exacerbated socio-spatial divisions within global cities. But unlike the first school of thought, the complexity of the changes in the occupational structures has not

38 | Theoretical background resulted in a dual and polarized socio-spatial structure of theses cities. Instead, as shown in the next sub-section, the outcomes are increased inequality, poverty, deprivation, and spatial segregation.

b. Socio-spatial outcomes of restructuring As noted in the beginning of this chapter, inequality and polarization are not synonyms, as inequality can refer to the widening income gap between the rich and the poor, while polarization has to include the growth of the number (or percentage) of people on the two ends of the spectrum at the expense of a shrinking middle. In this sense, the second school of thought argues that the expansion of the middle class in global cities – due to the professionalization of labor force – suggests that polarization is not an accurate description for the growing socio-spatial divisions in these cities (Fainstein et al., 1992; Hamnett, 1994; 2003; Fainstein and Harloe, 2003). Yet, evidence from London and New York suggests that their economic restructuring has contributed to a more complex outcome. On one hand, since the early 1980s, the income of the highly skilled labor in London has risen much faster than those in other segments of the labor market (Hamnett, 2003), leading to the exacerbated income inequality. Because apparently, the specialized services industries generate high profit – and high income – for those working in the sector, especially in comparison with other employment groups (ibid.). While on the other hand, the concentration of wealth is juxtaposed by a significant rise in the proportion of the population living in poverty. In 1997, this proportion reached 25 and 30 percent of the population of London and New York respectively, along with other indicators of growing deprivation such as large-scale begging and homelessness (Fainstein and Harloe, 2003), which matches Wilson’s (1987; 1996) hypothesis on the emergence of an underclass in American cities. Finally, unlike the dualistic views adopted by the first school of thought, the second school acknowledges that between the extreme wealth and poverty, there is a large section of the population in the middle whose situation is ambiguous, as they share – to a certain degree - the prosperity of economic

Globalization and production of inequality | 39 growth with the upper circuit, while at the same time; they still face difficulties in gaining access to desirable housing and education (Fainstein et al., 1992). Spatially, the added complexity to the occupational and social structures in this scenario contributed to the complexity of the parallel transformations in cities’ spatial structures. For that matter, the writers of the second school have criticized the over-simplistic dual metaphors, such as the citadel vs. the ghetto (Friedmann and Wolff, 1982), or the cohesive core vs. the disorganized periphery (Mollenkopf and Castells, 1991)

“The images of a dual or polarized city are seductive, they promise to encapsulate the outcome of a wide variety of complex processes in a single, neat, and easily comprehensible phrase. Yet the hard evidence for such a sweeping and general conclusion regarding the outcome of economic restructuring and urban change is, at best, patchy and ambiguous.” (Fainstein et al., 1992: 13)

Similar doubts about the dual city metaphor are also expressed by Marcuse (1989; 1993; 2003). He argues that the complexity of cities cannot be reduced to two types of neighborhood, one where wealth is concentrated, and the other where poverty is concentrated, because, there are other sections of each city that are neither very rich nor very poor (1993: 357). Therefore, he suggests that the complexity of cities’ social structure (in terms of class, socio-economic status, gender, race, and ethnicity) is reflected in what he calls the ‘quartered city’ (2003) or ‘partitioned city’ (2002), where the global city is increasingly divided into ‘several’ different – and in certain cases isolated – quarters. These quarters range from ‘the luxury city’ of residence for the very rich; ‘the gentrified city’ for professionals, managers, and technicians; ‘the suburban city’ of the lower middle class who are mostly blue and white-

40 | Theoretical background collar workers; ‘the tenement city’ for lower-paid workers who earn the minimum wage or little more, often with irregular employment, and no chance of advancement; to ‘the abandoned city’ (or the ghetto (1989)) of the very poor, the excluded, the permanently unemployed, and the homeless. While at the same time, these residential cities within the global city are overlapping (in occupancy and character, not necessarily in time or space) with corresponding quarters of economic activities, ranging from the controlling city, the city of advanced services, the city of direct production, the city of unskilled work, to the workless city (for more details, see Marcuse, 2003). In general, the concept of the ‘quartered city’ shares similar characteristics with other models such as the ‘divided city’ (Fainstein, 2010; Fainstein and Harloe, 2003; Fainstein et al., 1992), and the ‘unequal city’ (Hamnett, 2003; 2001; 1996; 1994). The three models agree that one of the most visible consequences of economic restructuring is the reinvestment in the built environment, including both the gentrification of the formerly working-class neighborhoods, and the new offices’ high-rise structures needed for the accommodation of the growing services industries. On the other face of the gentrification coin, spatial segregation of certain population groups (i.e. a disadvantaged ethnicity group, or people with low educational achievement) has increased, leading to the growing concentration of poverty and deprivation. Hamnett (2003:16) explains, “the impact of gentrification has if anything intensified the divisions between well-off homeowners and the economically inactive, unemployed or low-paid council tenants, at the local level.” Because access to housing became increasingly defined by households’ ability to pay, therefore, the highly paid groups ‘outbid’ other lower-paid groups (including disadvantaged ethnic minorities) in the competition over desirable housing (see chapter three for more details on the issue of deferential access to housing market). Finally, the second school of thought pays a special attention to the role of urban policies in explaining the growing socio-spatial divisions within

Globalization and production of inequality | 41

London and New York in particular, and within global cities in general. In more detail, economic restructuring was parallel to a neoliberal political manifesto that aimed to promote more liberal economic environment, and support the free operation of the market (Beall, 2002; Tickell and Peck, 2003; Held, 2004). Eventually, this shift in state role has affected its ability to provide essential services including health care, education, welfare benefits, and public housing. For example, the Thatcher government's “right to buy” program allowed the privatization of public housing units, leading to a decrease in the stock of subsidized housing in London over the past decades, i.e. less housing opportunities for the poor (Fainstein and Harloe, 2003). While at the same time, the very little investment in non-luxury housing in London since the 1980s was juxtaposed by state-facilitated gentrification projects, such as the transformation of the deprived East London areas of Docklands and the Isle of Dogs into an enterprise zone (Hamnett, 2003; Fainstein, 2010). Which is a clear indication of how urban development policies have shifted away from its social objective of improving residents’ wellbeing toward more entrepreneurial approaches that guarantee the city’s position in competitive economy.

c. Criticism Despite acknowledging the complex nature of the socio-spatial outcomes of economic restructuring, the second school of thought agrees with the first school that changing economic functions of global cities has a decisive role in defining the resultant social and spatial structures of these cities. They added that urban policies in each city (which are already influenced by economic restructuring) are also important as they mediate the production of inequality and segregation in different cities (Hamnett, 2001). In this sense, the second school of thought can be criticized for: firstly, overlooking other regional and national differences - such as historical and subcultural differences - that might also impact (or even reverse) the predefined socio-spatial outcomes of restructuring. Secondly, the second school rejects the suggestions posited by the first school that similar socio-

42 | Theoretical background spatial outcomes of socio-spatial polarization are shared by different cities around the world as long as they share similar trends of economic restructuring. However, since the focus of their studies is limited to London and New York (i.e. Fainstein et al., 1992; Hamnett, 2003), or the situation of Blacks in American cities (i.e. Wilson, 1987; 1996), accordingly, the second school does not offer far-reaching empirical evidence to totally disprove the general applicability of the polarization thesis on other cities. Instead, they merely challenge the over-simplicity of the bilateral outcomes of restructuring, while predicting other (also predefined) outcomes including growing inequality and spatial divisions in global cities.

2.1.3 Contextual particularities approach

The third approach to theorize the socio-spatial transformations in global cities vary greatly from the first two approaches especially when it comes to the understanding of the source and the nature of these transformations. Briefly, the first two schools agreed that the changes in cities occupational structures would lead to other changes in their social and spatial structures. Yet they disagreed on how this occupational structure is shifting (towards polarization or professionalization), as well as the social structure (towards polarization or inequality), which in either way are reflected spatially in growing gentrification and spatial segregation. Conversely, the third school of thought does not pay much attention to the nature of the outcome. Instead, it focuses mainly on the different aspects - other than economic restructuring and occupational shifts - that might be involved in defining the resultant social and spatial divisions within cities (see Musterd, 2005; Burgers and Musterd, 2002; Van Kempen, 2007; Marcuse and Van Kempen, 2000). The main argument of the third school of thought is that the global process of economic restructuring – as highly influential and important process – is still insufficient to solely explain the divided nature of global cities and the process of urban change in general (Van Kempen, 2007). For

Globalization and production of inequality | 43 example, economic restructuring alone fails to explain why European cities have evidently shown moderate change in income inequality and socio- spatial polarization when compared to the change in American cities over the last 20 years. The variation here suggests that a number of “moderating factors” (Van Kempen and Murie, 2009: 377) have contributed to different outcomes in the two contexts, such as type of national welfare states, and the ‘stronger interventionist traditions of European governments’ (ibid.). In this sense, the third school of thought calls for an all-inclusive approach that acknowledges place specific ‘variables’ (Burgers and Musterd, 2002) or ‘local contingencies’ (Marcuse and Van Kempen, 2000; Van Kempen, 2007) that are overshadowed by the exaggerated importance of economic restructuring in the global city literature. In one of the attempts to reach such all-inclusive model for explaining urban inequality, Burgers and Musterd (2002) discuss both the ‘polarization thesis’ (Sassen, 1988; 1991; 1994; 2001) and the ‘mismatch theory’ (Wilson, 1987; 1996). They conclude – based on empirical evidence – that both hypotheses should be viewed as complementing one another rather than competing to explain the social consequences of economic change. As, occupational polarization can be the accurate description for the situation in one city (i.e. Amsterdam), while occupational professionalization can be suitable to describe the outcome in another city (i.e. Rotterdam), leading to different patterns of social and spatial polarization and inequality. And even within the same city, different ethnic groups can be subjected to different mechanisms of generating inequality. For that matter, Burgers and Musterd (2002) suggest a three-layered explanatory model for analyzing inequality in cities. Concisely, the model assumes, on one hand, there is a layer of global change, which refers to economic restructuring, and the increasing mobility of capital, people, commodities and information. While on the other hand, there is a layer of local change, which refers to the change in labor market opportunities of individual people in a specific local context. At this point, unlike the popular notion in the global city literature that global changes are

44 | Theoretical background influencing the local context in unmediated and unidirectional manner, Burgers and Musterd argue “the dynamics of ‘the global’ do not impinge in a direct way on ‘the local’.” (2002: 406). Instead, they suggested a third layer of variables that mediate the change between the global and the local. These variables are – and not limited to – subcultural, institutional, and historical differences among cities. Firstly, subcultural differences among ethnic groups in an urban area are highly influential on the life chances of individuals who belong to certain group, where the poorly schooled members of one group may suffer from being excluded from the labor market (which matches Wilson (1987) description of the situation of African Americans in American cities). While at the same time, members of another ethnic group may be overrepresented in poorly paid service jobs or a certain informal economic activity (which matches Sassen (1991) and Friedmann (1986) description of the position of ethnic minority in the labor market). Therefore, patterns of income inequality and/or polarization in any city cannot be understood away from its specific ethnic composition, upon which levels of exclusion from education, employment, and even housing are dependent. Secondly, national institutional differences are expected to affect the resultant labor market structures, social exclusion, and inequality levels. In more detail, the corporatist and social democratic welfare states in Europe provided strong employment protection and benefit systems, which had a moderating impact on the outcomes of economic restructuring in the 1980s and early 90s (Musterd and Ostendorf, 1998). While at the same time, the liberal welfare state of the US showed more pronounced levels of inequality due to the loss of jobs and income. Burgers and Musterd (2002: 406) give an example of how the “multifold state intervention in various spheres of life” in the Netherlands – especially the regulations of housing sector - has contributed to a relatively mild extent of geographical concentration of disadvantaged population, particularly when compared to segregation levels of the poor and minorities in the US. However, in recent years, state-

Globalization and production of inequality | 45 spending cutbacks in Europe impacted the subsidy system leading to less income redistribution, growing income inequality, and less state-provided services including health care and housing (Musterd and Ostendorf, 1998). Overall, the third school asserts that minimal state intervention is strongly linked to greater income inequalities and vice versa. Therefore, the type of the welfare state is an important variable in defining the socio-spatial outcome of global economic change. Finally, specific social and economic history of individual cities creates different trajectories for the global changes to be mediated to the local level. They explain ““cities which had their momentum in the industrial revolution in most cases have been hit much harder by de-industrialization than cities which have a past which was more characterized by government and administration, services and trade.” (Burgers and Musterd, 2002: 407). In this sense, different cities are not expected to show comparable levels of growing inequality and/or polarization unless they share other characteristics developed over an extended period of time. Similarly, Marcuse and Van Kempen (2000) and Van Kempen (2007) offer a parallel argument to what Burgers and Musterd (2002) call ‘the mediating variables’ between the global and the local changes. Van Kempen (2007) acknowledges the fact that changes within cities can be traced back to developments that take place on higher spatial levels of the region, the nation, or even the world. Yet, he contends that other developments are also at play. In total, Marcuse and Van Kempen (2000) named seven local contingencies that are – in their view – defining the extent and form of the impact of global trends on different cities. They agree with Burgers and Musterd (2002) on the importance of the subcultural (contingency of race), institutional (contingency of politics), and historical (contingency of history) differences among cities, and they also added the contingency of geography (the pre-existing natural and built environment of the city), the contingency of economics (level of economic development), the contingency of globalization (cities’ position in the processes of globalization), and the

46 | Theoretical background contingency of inequality (existing levels of polarization in numbers and in shares of wealth) - see Van Kempen (2007) for detailed description of each contingency. Spatially, these variables and contingencies are expected to produce distinct patterns of socioeconomic and ethnic segregation in cities. For example, variations in housing policies can affect the allocation and size of public housing stocks in different cities, when public housing units are concentrated in specific areas within the city, then there is higher chance for low-income groups to be concentrated in such areas (Van Kempen and Murie, 2009; Marcuse and Van Kempen, 2000). Similarly, the contingency of race can have a noticeable effect on the changes of ethnic segregation patterns in cities, because the existing enclaves can have a role in determining where the new comers to the city are most likely to settle where their family or other ethnic ties are present (Fischer, 1976). Overall, social inequality and exclusion lead to diminished opportunities for the excluded population and limit their access to services including housing (find more details about differential access to the housing market in the subsequent chapter). However, Musterd (2005) and Van Kempen (2007) agree that the strong correlation between social inequality and spatial segregation is not an automatic one. Therefore, the authors of the third school of thought recommend that studying patterns of socio-spatial segregation in different cities require incorporating the previously stated variables and contingencies as a part of the framework of comparative studies.

Criticism Despite the fact that the third school of thought acknowledges the importance of local contexts in explaining the growing socio-spatial divisions within cities, still, they agree with the first two schools that growing social polarization, inequality, and spatial segregation are the main characteristics of global cities. Marcuse and Van Kempen concluded that there is no standard pattern of the global city, “[B]ut there is a set of common trends that, taken together, form a pattern, standing in some

Globalization and production of inequality | 47 orderly relationship to each other. Looking back at the alternate theories of the consequences of globalization, interdependent polarization vs exclusion, in effect both are correct: the rich get richer (and form citadel and exclusionary enclave) and the poor get poorer; most are needed (often forming immigrant enclaves), but some poor are left out (and confined to excluded ghetto).” (2000: 271, 272). In this sense, the influence that local contingencies have is confined to setting the pace of the inevitable change towards inequality and segregation (as in the difference between European and American cities), or fine-tuning the extent and form of divisions on neighborhood level. The problem with this assumption is while acknowledging the complexity of the global/local interplay, the influence of local contexts is still limited to explaining the unavoidable growing inequalities within cities, rather than producing remarkably different outcomes of socio-spatial change. In other words, as influential as suggested by the third school, local contingencies of policy, history, or race are not considered influential enough to ‘reverse’ the socio-spatial outcomes of economic restructuring. This discrepancy is further discussed in the subsequent chapters.

2.1.4 Exacerbated inequality and polarization: other complementary views

“Cities where these developments [post-Fordist economy] are most pronounced are also places where a broad social divide exists between the upper and lower segments of the labor force and this divide has actually tended to widen in recent years as neo- liberal ideology and practice have taken deeper and deeper hold in both the economic and political spheres.” (Scott, 2008: 769).

Beside the three main approaches for theorizing the exacerbated inequality and polarization in global cities, a wide range of studies have enriched the debate by highlighting a number of processes that are – to a

48 | Theoretical background certain degree – contributing to the social downward mobility of a large segment of the population of the global city, which in turn, contributes to the widening gap between different social groups in the city. In more detail, the economic restructuring discussed above has its toll on the role of the state in the global economy. Certain political shifts were pinned down as both an outcome of increased competitiveness, and at the same time, a ‘necessity’ for capital and labor mobility to take place. Consequently, political shifts have also contributed to the socio-spatial transformations of the global city as a part of the complex restructuring process leading to inequality and polarization. In general, state is not just one of the sources of regulation, but it is the regulative institution (Kazepov, 2005), which defines the role of all other institutions and their modes of interactions. Historically, with the rise of Fordism, modes of production were contained and controlled by the state (Soja, 1989). However, as highlighted earlier, the flexibility of the post- Fordist global economy and the unprecedented mobility of capital have led directly to an increased regional competition over international investments (Harvey, 1985b; Soja, 1989; Kesteloot, 2000; Marcuse and Van Kempen, 2000). Consequently, it is argued the central role of the state has declined, because national governments became ‘powerless’ in the face of cross- border competitiveness of globalization (Gray, 1996; Harvey, 1985a). The form and intensity of the declined role of the state sparked a debate in the specialized literature (Ray, 2007). Although the common notion of a powerless state is widely accepted, Panitch (1998) argues that the overall role of the state has not declined. Instead, the state is even more involved in facilitating business activity. Accordingly, the role of the state has shifted to be more market-oriented (Larbi, 1999). Tickell and Peck (2003) define this political shift as a Neoliberal political manifesto that is parallel to the economic trends of globalization, to promote more liberal economic environment, and support the free operation of the market. However, that shift in the state role implies and eventually produces a downsizing in the

Globalization and production of inequality | 49 nation-states ability in regulating the market, through different tendencies such as decentralization, liberalization, deregulation, privatization, diminished public spending ...etc. (Beall, 2002; Held, 2004). The deep structural changes of neoliberal globalization and their social implications can be summarized as follows: Firstly, dramatic decline in the redistributive power of the welfare state is caused by the shift in governance level and global competitive forces (Goldsmith, 2000; Marcuse and Van Kempen, 2000). Van Kempen (2007) summarizes the role of the welfare state in two main themes. On one hand, it aims to provide an elaborate system to support those who are in a weak position on the labor market, such as the unemployed, the ill, and the elderly. While on the other hand, it also aims to grant system of subsidies in all kinds of services like housing, recreation, and social work. Nonetheless, since the 1970s, governments have been forced to reduce state spending and interventions (Beall, 2002). Basically, economic globalization has drained off resources from the nation-states, and has reduced governments’ room to maneuver in promoting social protection (Gray, 1996; Beall, 2002; Mingione, 2005). Consequently, large sector of the population is exposed to higher financial risks especially those who are dependent on the state (Marcuse and Van Kempen, 2000). Secondly, state-spending cutbacks were parallel to a wave of privatization of business, enterprises, and even public services (Sullivan, 1987). As a result, privatization has affected both the labor force attached to privatized business and enterprises, and also affected the population accessing the privatized services. On the employment front, privatized businesses seek profit-maximizing organizational restructuring (Fernandez et al., 2005), which usually contributes to two severe drawbacks that affect public employees. On one hand, corporates adopt downsizing strategies (Burke and Cooper, 2000). Accordingly, the mass layoffs result in significant numbers of public employees losing their jobs (Fernandez et al., 2005). On the other hand,

50 | Theoretical background profit-oriented businesses focused on flexible forms of employment in order to reduce their costs, as part-time workers typically receive lower pay and fewer non-wage benefits than full-time workers (Mingione, 2005, Fernandez et al., 2005). Therefore, rates of unemployment are rising, while permanent stable jobs with standard contracts have begun to diminish. Unfortunately for workers, the deunionization and the weakening of labor organizations made the working class unable to resist the negative impacts of privatization (Marcuse and Van Kempen, 2000), as well as other drawbacks of deindustrialization such as the deterioration of wages (Sassen, 2001). On the services front, privatization of basic services - along with declining subsidies provided by the state - created difficulties in accessing crucial sectors such as health, education, transportation, and housing for a large sector of the population (Ritzer and Ryan, 2011; Mingione, 2005; Van Kempen, 2007), as privatization of services leads to the increase in services costs, and accordingly limit the access to crucial services only for people who can afford it. Finally, the pressure of the transnational corporations on central government to promote easier labor mobility has affected the migration restrictions put by the state, and has consequently reshaped the map of the migratory flows (Castles, 2002). The altered regulations aim to select immigrants who are expected to contribute to society and to prevent immigrants, who are expected to become a burden (Ritzer and Ryan, 2011). For example, governments of migration-receiving countries such as the United States, Canada, and Australia have set up privileged entry systems to attract entrepreneurs, executives, scientists, professionals and technical specialists. More recently, Western European and some East Asian countries have followed them (Findlay, 1995; Castles 2002). Furthermore, free trade agreements such as NAFTA or the GATS contain specific rule for easy cross border circulation of professional service workers (Sassen, 1988). At the same time, despite the notion that low-skilled migrants and asylum seekers can be socially harmful, they are considered an economic necessity, and as a

Globalization and production of inequality | 51 result, they are increasing in numbers while subtly welcomed and accepted by holders of economic and political power (Castles, 2002), to finally raise levels of ethnic heterogeneity in migration-receiving countries. Recapitulating, in addition to the altered occupational, income, and social structures of the global city as a direct outcome of the new functions of cities in the global economy; the gap between the rich and poor is expected to expanded indirectly due to different reasons; including, the fiscal crisis of nation-states which has placed the families of the lower portion of the income distribution in very unstable conditions (Martinotti, 2005; Van Kempen, 2007), as declining income and declining subsidies made them increasingly vulnerable to negative income shocks caused by illness or injury, the loss of a job, or any other family emergency (Massey, 2009). Also, increased levels of long-term unemployment developed by privatization along with increased supply of flexible low-wage jobs, have put large portion of the population so close to the edge of financial solvency (ibid.). Overall, the declined role of the state has originated new risk areas, which led polarization tendencies to persist (Martinotti, 2005). Finally, the surge of immigrant workers into global cities - which is partly facilitated by the state – contributed to the altered social and cultural mix of global cities, or as Madanipour (2004: 270) puts it, “arrival of new immigrants have created new tensions and challenges for social integration”. In turn, the increased cultural heterogeneity adds up to the pressure on local governments to focus on social cohesion issues in the political agenda.

2.2 Issues of generalization, scale, and space

This section aims to not only summarize the debate over the socio- spatial transformations in global cities, but also to raise several questions regarding the assumptions discussed above. Firstly, the section discusses the common trends among the different approaches for theorizing urban inequalities within cities, and then it raises question about the validity of the

52 | Theoretical background evident generalization tendencies in the available research. Secondly, the section questions the implied global/local dichotomy adopted by the global city literature, while clarifies the different scales through which the multifaceted process of restructuring is taking place. Finally, the section sheds more light on the spatial dimension of inequality and polarization, and underlines the shortcomings of available research regarding the unproblematic views of space.

2.2.1 Justified generalization or an overstatement?

“Distinct socio-spatial forms arising out of these processes (economic transformation of the last two decades) are high-income residential and commercial gentrification and sharp increases in spatially concentrated poverty and physical decay.” (Sassen, 2001: 257)

So far, this chapter presented a brief review of the different approaches for theorizing urban changes associated with the changing economic functions of global cities. As a main conclusion, the three approaches offer different interpretations of the complex dynamics of urban change; they vary in identifying the resultant occupational structures accompanying the rise of the service industries. They also vary in determining the situation of the middle class, and the impacts of education and ethnic background on the position of individuals in the labor market. Finally, they vary in acknowledging the impact of history, race, politics, and other contextual differences on the resultant patterns of urban change. However, despite their differences, the three approaches offer parallel conclusions, they strongly tie the global process of economic restructuring to a main outcome that is the intensified socio-spatial divisions within cities, and whether the social divisions are due to increased polarization or inequality, or whether urban policies or history are taken into account or not. Still, the three approaches

Globalization and production of inequality | 53 agree that gentrification and spatial segregation are the most visible characteristics of the global city.

“Social polarization was posited as an inevitable effect of global capitalist restructuring which was seen as an uneven process affecting cities and regions, and the people in them, in different ways, including some localities and groups while pushing others to the margins and spaces of exclusion.” (Bridge and Watson, 2003: 254)

Noticeably, the general agreement on gentrification and segregation as the main spatial consequences of economic restructuring is derived from a variety of empirical studies. London and New York were the main focus of these studies (as the leading global cities). Yet, the question is, to what extent is the global/divided city model applicable on other cities around the world? What makes a city a global city and therefore makes it more susceptible to intensifying socio-spatial divisions? As noted earlier in this chapter, certain economic functions and their supporting infrastructures determine the position of cities in the global economy (Beaverstock et al., 1999; Taylor, 2004). Although Robinson (2005) highlights the limitations of the narrow focus on the financial and business services in the global city literature (see chapter four for detailed discussion), yet according to the aforementioned description of the global city, no city is a non-global city (Taylor, 2004: 42), where cities around the world from Paris, Frankfurt, Amsterdam, Los Angeles, Sydney, Hong Kong, to Sao Paulo, Buenos Aires, Bangkok, Taipei, and Mexico City are declared to be – to different degrees - among the major international financial and business centers (Sassen, 2000). Congruently, the three approaches for theorizing the socio-spatially divided nature of global cities are building their conclusions, not only on London and New York (Mollenkopf and Castells, 1991; Sassen, 1991;

54 | Theoretical background

Fainstein et al., 1992; Hamnett, 2003), but the analysis included other cities such as Amsterdam and Rotterdam (Burgers and Musterd, 2002), as well as, Tokyo, Singapore, Rio de Janeiro, Brussels, Frankfurt, Sydney, and Calcutta (see contributions in Marcuse and Van Kempen, 2000). In this sense, the global/divided city model is not an exclusive description to the top ranked global cities. Instead, the model suggests common socio-spatial trends that are applicable on different cities as long as they share parallel transformations in their economic functions. The common trend is the growing socio-spatial divisions, while the pace of change and the course of change - through patterns of polarization or inequality - are marginally bounded by the contextual differences that vary from city to city. In a nutshell, this alleged applicability of the global/divided city model has several implications on the stream of urban research. On one hand, it supports the prominence of global processes over other scales of the urban hierarchy (see subsequent section). On the other hand, it renders a set of niches of cities that most scholars take for granted, as the model is the stepping-stone for a wide range of studies (for example, see McCarthy, 1999 and Badcock, 1997), while its fundamental idea is rarely challenged. For that matter, the validity of the generally applicable model is one of the biggest question marks that the study is attempting to investigate.

2.2.2 The global and the local: a question of scale

“The social polarization thesis has thus created a dominant way of seeing urban society extending far beyond the global city context. By producing an implicitly context-indiscriminate and seemingly unquestionable link between economic restructuring and segregation, this way of seeing reduces the need to explain segregation in specific contexts and terms, and

Globalization and production of inequality | 55

takes for granted the existence of common generating mechanisms and processes”. (Maloutas, 2007: 735)

The general applicability of the global/divided city model on a wide array of cities raises a question about scale, how is this generally applicable model considered an oversimplification of the global/local interplay? Geniş (2007) argues that although the world city (Friedmann, 1986) and global city (Sassen, 1991) approaches have ‘placed’ globalization in cities instead of the prevailing representations of global economy as a ‘placeless’ entity controlled by footloose capital. Yet, they approached the urban impacts of economic globalization by rendering the global/local interplay “as a ‘top down’ and unidirectional relationship from global to local” (Geniş, 2007: 61). The problem with this conceptualization of the global/local interplay is that it does not capture the complex reality; this conceptualization marginalizes the influence of the local and creates a binary dichotomization of the global vs. the local (Smith, 2005: 243), while the marginalization of the local explains why cities are expected to share common trends of urban change despite their local differences. Instead, scale theorists argue that the global and local scales are not opposites, but they are two sides of a coin (Ritzer and Ryan, 2011). Pacione (2009) explains that the discussion of different scales focuses on the different social processes that are taking place at these and in-between these scales, while the scales are not fixed; they are in a continuous process of rescaling. For that matter, when a social process does not fit neatly in one of the given territorial boundaries such as the nation-state, new prefixes appear to describe the social process that are taking place, for example, below (sub- ), above (supra-), or across (trans-) the national scale (Brenner, 2004). However, the contemporary conditions of globalization and the complex process of restructuring, requires not only more intermediate modes of organization: global, transnational, international, macro-regional, national,

56 | Theoretical background micro-regional, municipal, local scales to capture its complexity (Nederveen Pieterse, 1993), but also, the interwoven correlations among these scales is to be understood. Anderson (1996) and Brenner (2004) argue that the traditional metaphors of the ladder or the Russian doll that describe the vertical hierarchy of embedded territorial units are no longer viable, as these metaphors are inadequate to explain the interconnectedness and overlapping of the social processes taking place across different scales. “The contemporary world, is not a ladder up or down which processes move from one rung to the next in an orderly fashion […] not only are there now more rungs but qualitatively they are more heterogeneous; and direct movement between high and low levels, missing out or bypassing ‘intermediate’ rungs, are now a defining characteristic of contemporary life.” (Anderson, 1996: 155), and instead of the ladder metaphor, Anderson suggests that ‘adventure playground’ is a better metaphor, the movement across the complex construction of the playground (up, down, diagonally, from high to low, and from low to high simultaneously) represents a closer description of the correlation among different social processes and scales. As a result of perceiving the local scale as the other side of the coin instead of being contained and defined by the global (Sheppard, 2002), several approaches are developed to explain the global/local interplay beyond this prevailing oversimplification. ‘hybridization’ (Nederveen Pieterse, 1993; 2001) and ‘glocalization’ (Swyngedouw, 1997; 2000; Bauman, 1998; Kraidy, 1999) are just few examples. Although they are contested terms, yet they are highly indicative about how the global and the local are perceived mutually influential. One representative definition of glocalization states that it is "the integration of the global and the local resulting in unique outcomes in different geographic areas" (Ritzer, 2004: 73). In the light of this definition, are the contingencies of history, politics, and race contributing to unique spatial outcome in different cities? If yes, then the general applicability of the global/divided city model is challenged based on the credible assumption that existing political, social, cultural, and

Globalization and production of inequality | 57 spatial conditions are creating distinct local contexts for cities, which have significant impact on restructuring outcomes in individual cities. Therefore, this study aims to examine the influence of local contexts on the direction and extent of change in socio-spatial division within global cities.

2.2.3 Spatial segregation: automatic outcome of inequality?

Although the spatial dimension of economic restructuring is repeatedly acknowledged in the world/global city literature. Where on one hand, the spatial concentration of highly specialized service firms in global cities has contributed to growth of high-rise office districts, as well as shopping, hotels, and entertainment districts (Fainstein et al., 1992; Sassen, 2001). On the other hand, it is acknowledged that the spatial concentration of firms in a close proximity is still required to facilitate the face-to-face meetings and access to common facilities (Fainstein and Harloe, 2003; Robinson, 2005). Yet, as highlighted in the introductory chapter, the three schools of thought focus on explaining the growing income inequality and/or social polarization in cities, and then, spatial segregation appears as the implied and the unproblematic side effect of growing social divisions: “Spatial polarization arises from class polarization.” (Friedmann, 1986: 76). Spatial segregation in world/global city literature is usually explained based on the change in households’ ability to pay for desired housing, due to either their changing position in the labor market, or due to their exclusion from both decent jobs and housing opportunities. Yet, this affordability scenario does not take into account the impact of specific housing market dynamics including homeownership rates, public housing provision, and self-segregation preferences. The discussion of space and its relevance to the understanding of economic and social change is lengthy. Therefore, the next chapter provides a glimpse of the spatial structures of accumulation regimes. Then, the expected increase in spatial segregation in global cities is explained through the discussion of the impact of global competitiveness on urban space in

58 | Theoretical background both business and residential areas. Finally, the limitations of the aforementioned affordability scenario are discussed in more detail.

Notes

1 Known today as Guangzhou

Chapter 3

3 Spaces of globalization and inequality

“I argue that economic globalization is an inherently geographic phenomenon” (Yeung, 2002: 285)

In contrast to Yeung statement, the advancement of information and communication technology has led several globalization theorists to claim that distance and territorial borders are becoming obsolete, because the new technologies have altered the locational sense of boundaries and offered ‘proximity’ of social relations across large tracts of space and time 1 (Tomlinson, 1999; Rofe, 2009). Cairncross's ‘the death of distance’ (1997), Friedman’s ‘the world is flat’ (2005), and ‘the geography is dead’ thesis (see Sheppard, 2002; Morgan, 2004; Murray, 2006) are just a few example of how space and geography are perceived increasingly irrelevant when it comes to determining the cost of electronic communications and transactions, as well as other business decisions. As, money and information are diffused almost instantaneously across both borders and distance (Sheppard, 2002). The world is shrinking (Kirsch, 1995) and the footloose transnational corporations are spreading identical products across the world (Murray, 2006), a world that McLuhan (1962; 1964) calls a ‘global village’. However, the hypotheses declaring the death of geography are heavily disputed. On one hand, although it is acknowledged that the new technology

60 | Theoretical background has facilitated the time-space compression 2 (Harvey, 1989), yet the ‘geography is dead’ thesis tends to exaggerate the “distance-destroying capacity” of information and communication technologies (Morgan, 2004: 5), while inaccurately assuming that the digital space – in which electronic exchanges are taking place - is totally dissociated from physical space (Dodge and Kitchin, 2007). In fact, globalization does not imply a diminishing importance of physical space. Sassen (1999) explains, “Even the most digitalized, globalized and dematerialized sector, notably global finance, inhabits both physical and digital spaces. These firms' activities are simultaneously partly deterritorialized and partly deeply territorialized, they span the globe yet they are highly concentrated in very specific places.” (p. 8). In this sense, digital space is in no way a surrogate for physical space. Instead, the information technology has contributed to the increased importance of physical space to the global economy, through facilitating the concentration of power that controls capital investment in central locations, and through transforming how these locations are connected. On the other hand, other globalization related notions, such as, the rise of the ‘global scale’, and the shift from the ‘spaces of places’ to ‘spaces of flows’ do not imply that the importance of physical space is diminishing. Sheppard (2002) and Murray (2006) argue that these notions are in fact emphasizing the importance of localities. In order to explain this contradiction, Sheppard (2002) indicates that the use of spatial metaphors such as ‘scale’ and ‘networks’ to describe distinctive features of contemporary globalization suggests that space (and time) continues to matter. For example, the concept of networks focuses on horizontal relations among particular places or localities3, and among these localities, there is spaces of flows. In this sense, Murray (2006) argues that spaces of flows are not a substitute for spaces of places. Instead, “spaces of flows operate within and between spaces of places” (p. 55), and he continues, when the flows between local-to-local stretch across space, they become ‘global’ in extent. In view of the above, localities are not less important due to globalization,

Spaces of globalization and inequality | 61 yet localities continue to matter based on the assumptions that the global scale is locally constructed, where any global process is actually a stretched local–to-local process that unfolds in particular localities, while these localities are playing an important role as nodes in a larger network of flows.

3.1 Space, society, and capitalism

The above discussion highlights the contrasting views regarding the relevance and importance of space to the processes of contemporary globalization. Noticeably, the debate over the recognition of space and geography’s key role in the understanding of modern societies is not new. For decades, social sciences dealt with social forms as if “the world existed on the head of a pin” (Massey, 1985; 12), where 'geography' was underestimated in terms of distance, and in terms of local variation and uniqueness (Sayer, 1985; Massey, 1985). For instance, the classical approaches of Marx, Weber, Durkheim and Simmel are criticized for prioritizing time and history over space and geography (Harvey, 1985a), they dealt with space as merely a reflective mirror of society, where historical events simply take place in space in an unproblematic manner, as if space only represents a site or a stable context that is uninteresting and overlooked, while geographical variation rendered as 'unnecessary complication’ (Soja, 1989; Harvey, 1985a). Later in the 20th century, contributors including Lefebvre, Harvey, Castells, Massey, Soja and others offered diverse views for recognizing space as a prominent feature in the sociological understanding of the city. Since it is not possible to fully explore all these different views, this section focuses mainly on two perspectives; firstly, the conceptualization of space as a social construct, and secondly, the production of space under capitalism.

62 | Theoretical background

3.1.1 Space as a social construct

The discussion in the previous chapter concluded that cities are in a continuous process of change, where their economic, social, and spatial settings are being restructured. The spatial change occurs in accordance to other economic and social changes that are materializing in a spatial form (social polarization leads to spatial segregation is one example). In return, existing and historical spatial configurations are a contingency that can affect the social change. In order to understand this mutual correlation between the ‘social’ and the ‘spatial’, spatial patterns can no longer be perceived simply as a ground on which social life takes place (Massey, 1985). Instead, space is to be visualized as a medium (not a mirror) in which social relations are produced and reproduced, while at the same time, space itself is a social product that can be reinforced, reproduced, restructured, and reconstituted (Gregory and Urry, 1985). Similarly, Soja (1985) offers a parallel appraisal and describes this mutual tension between social relations and spatial structure as a ‘socio-spatial dialectic’, where “spatiality is both product and producer […] social life is both space-forming and space- contingent” (Soja, 1985: 98). He also declares that physical space is the realization and materialization of social life. Although spatial form of the city is not necessarily a precise and predictable imitation of its economic and social settings (Knox, 1991; Beaurgard and Haila, 2000), yet instead of the long-standing perception of space as static and timeless entity that is immune to historical change (Brenner, 2004), the conceptualization of space as a social construct helps explain; on one hand, how and why the transformations in a city’s economic uses, cultural composition, and variations in status and power are producing their different and distinctive spatial forms over the city’s history (Lefebvre, 1991; Marcuse, 2005). On the other hand, since globalization problematizes social relation - as they become increasingly interconnected on a global scale, it also problematizes the spatial patterns of these social relations (Brenner, 2004). Under these conditions, acknowledging the dynamic nature

Spaces of globalization and inequality | 63 of space (and its ability to be reorganized in accordance to globally stretched yet locally situated processes) helps explain the ways in which globalization is reworking and being reworked in cities and even in districts within cities.

3.1.2 Spaces of capital accumulation

Another perspective for understanding the basis of spatial restructuring of global cities is based on the definition of contemporary globalization as “nothing more than yet another round in the capitalist production and reconstruction of space” (Harvey, 2001; 24). In this discussion, concepts such as ‘spatial fix’, ‘creative destruction’ and ‘the layered city’ are central. Briefly, over an extensive series of publications, Harvey (2010; 2001; 1989; 1985; 1978) keeps emphasizing the concept of ‘spatial fix’, his main argument is that “[c]apital represents itself in the form of a physical landscape” (1978: 124). Because in order to produce, capitalists must invest in fixed capital such as plant and equipment, and in order to sustain flow of capital, they have to secure their supply of ‘materials’, their own spatial division of the labor force, and they finally need ‘market’ to put their output into circulation (Harvey, 2001; Walker, 1985; Kesteloot, 2005). At the same time, the concept of ‘spatial fix’ describes how capitalism overcomes its crisis tendencies through geographical expansion and restructuring. Suburbanization, for example, is one way to absorb the capital surplus, where the over-accumulation is ‘spatially fixed’ through the investment in fixed assets of infrastructures, roads, houses, and their maintenance. However, the ‘fixity’ of capital raises what Harvey (2010; 2001; 1978) calls one of the central contradictions of capital, whereas capital builds a physical landscape appropriate to its own condition at a particular moment in time, it also has to destroy4 it at a subsequent point in time in order to make way for a new “spatial fix”. Which creates a struggle for capitalist development to preserve the values of past capital investments in the built environment, and the unavoidable destruction of the value of these investments.

64 | Theoretical background

Similarly, Massey (1984) and Kesteloot (2000; 2005) portrayed eloquently the metaphor of the layered city, which also explains how in each round of accumulation the capital makes different locational decisions to achieve optimal conditions for development. As a result, each round has its qualitatively different economic sites and residential areas across geographical space. Yet, - similar to Harvey’s view - new locational decisions have to be made in a later point of history. These decisions, however, cannot be implemented unless parts of the past capital investments are demolished. From this perspective, the socio-spatial structure of the city is a product of historical process occurs in successive rounds of capital accumulation that are deposited in layers one upon another. Each layer represents distinct spatial arrangement of economic and residential functions that belong to a certain round of accumulation, where spatial features from previous layers may still be present in recent layers of spatial development. In view of the above, it is established that each accumulation regime creates a different spatial pattern (Kesteloot, 2005). For example, traceable structural changes appear when spatial patterns of industrial capitalism are compared to those of finance capitalism, due to not only the shift from manufacturing to service industries, but also due to the change in the division of labor coinciding with each accumulation regime. Because workers require a certain fixity for their reproduction, such as houses, schools and parks (Walker, 1985), then the changes in the division of labor and its associated changes in the social structure of the urban population will contribute to the distinctive spatial structure of each layer of accumulation (Kesteloot, 2005). In the industrial period of capitalism, “growing mass production found a market in growing mass consumption” (Kesteloot, 2000: 193), and with the support of the state as the key source of regulations, the city became a machine for such production and consumption (Soja, 1989). As a result, the economic growth of Fordism expanded the market for houses, cars and consumer durables (Harvey, 1985b), which was reflected spatially in the

Spaces of globalization and inequality | 65 form of massive suburbanization (Soja, 1989; Kesteloot, 2000), and the inner city became increasingly abandoned, filled with slums, and undesirable for those who could afford better (Marcuse and van Kempen, 2000). At a certain point, “the urban production machines surpassed their capacity to consume their product and triggered falling rates of profit and rising class conflict.” (Soja, 1989: 101). This crisis of the Fordist economy marked the beginning of another round of restructuring, where the flexible mode of accumulation is materialized in the form of new constructions that accommodate the service industries, the altered housing patterns, and the spatial arrangement of urban areas as a whole (Kesteloot, 2005; Fainstein, 2010). The formation of the post-Fordist spatial arrangement is discussed in more detail in the next section.

3.2 Globalization on the ground

“In the context of globalization, many of these [economic, political, and subjective] processes are not only operate at a global scale but also materialize in the concrete environments of cities.” (Sassen, 2006: 477)

Broadly, the discussion over the significance of space and its relevance to the economic restructuring emphasizes the fact that ‘global processes’ are produced in physical space and can have a tangible spatial dimension (Gregory and Urry, 1985; Goldsmith, 2000). The previous chapter focused on describing the outcomes of cities’ spatial restructuring, and highlighted distinct spatial features that are supposedly shared by global cities. Examples include, new high-rise constructions emerging in central cities to accommodate the growth of services industries (Fainstein, 2010), juxtaposed by vast areas of abandoned factories, empty lots, and environmentally polluted land (Beauregard and Haila, 2000) due to the deindustrialization of

66 | Theoretical background western societies since the 1970s, and finally, exacerbated spatial segregation of different population groups based on occupation, race, ethnicity, immigrant status, income, lifestyle, and other employment related variables (Soja, 1989). More specifically, this chapter focuses on how the production of these spatial outcomes is explained in the global city literature. Note that most of the studies dealing with spatial restructuring focus mainly on American and Western European cities. Yet, according to the three schools of thought discussed in chapter two, similar patterns of urban change appeared in cities in other regions of the world, leading to the consensus on a general ‘global city/divided city’ approach for addressing socio-spatial transformations in global cities. The rest of the chapter summarizes the agreed-upon general trends of urban development, and the production of spatial segregation in the global cities’ literature. The discussion revolves around the following questions; in what way are the increased mobility of capital and labor influencing urban development? To what extent is the pursuit of economic growth controlling urban policies? And how are the local housing markets of global cities restructured accordingly? Then, the chapter raises several questions about the general applicability of these trends on cities in different contexts, with a special focus on the impacts of different national and local policies, as well as the different local housing dynamics, on the outcomes of spatial restructuring in individual global cities.

3.2.1 Urban development and economic growth

As noted, the intensified regional competitiveness over international investments is a key characteristic of global economic integration, where every city and region has to compete with the others to attract and retain investments (Harvey, 1985b; Soja, 1989; Kesteloot, 2000; Fainstein, 2010). Accordingly, competitiveness has a great role in reshaping the built environment of global cities, by increasing pressures for infrastructural investments, improved housing, and services (Soja, 1989). In fact,

Spaces of globalization and inequality | 67 competitiveness made urban development under conditions of globalization different from those of previous rounds of restructuring, as the urban and regional planning process became increasingly controlled by private investors and profit-based organizations while the exclusive role of the governments as the leading policy-maker is retreating (Soja, 1989; Healey et al., 1995; Elander and Blanc, 2001; Andersen, 2002; Van Kempen, 2007). As a result, urban development shifted to focus on economic growth as its main objective, Susan Fainstein addressed this issue in more detail, she stated: “even the provision of amenities such as parks or cultural facilities is rationalized by their potential to raise property values and attract businesses and tourists. Decisions concerning where to locate facilities become warped by considerations of their economic, as opposed to their social, impacts. Thus, capital investments by city governments are intended to support development projects rather than improve the quality of peripheral neighborhoods, and rezoning for higher densities occurs in response to developer demands for more profitable investment opportunities.” (Fainstein, 2010: 1). A key example for the shift in urban policies in the UK - after the election of the government of Margaret Thatcher in 1979 – is the transformation of the deprived East London areas of Docklands and the Isle of Dogs into an enterprise zone (Hamnett, 2003; Fainstein, 2010), which indicates how urban development shifted away from its social objective of improving residents’ well-being toward more entrepreneurial approaches that guarantee the city’s position in competitive economy. Swyngedouw, Moulaert, and Rodriguez (2002) also agree with Fainstein that globalization and liberalization are articulated in new forms of urban governance, in which urban interventions are driven by the priorities of the elite, as the public budgets are increasingly redirected from social objectives to investment in large-scale urban development projects such as, waterfronts, exhibition halls, business centers, and international landmarks. Their conclusions are drawn from the study of thirteen different urban

68 | Theoretical background development projects in twelve European countries5, through which they highlight the characteristics and the consequences of such project-based urban development. Briefly, the main objective of these large-scale urban development projects is to enhance the competitive advantage of cities in a highly competitive economy, “[R]epositioning the city on the map of the competitive landscape meant reimagining and recreating urban space, not just in the eyes of the master planners and city fathers and mothers, but primarily for the outsider, the investor, developer, businesswoman or –man, or the money-packed tourist.” (Swyngedouw et al., 2002: 550-551). And in order to achieve its goal, the implementation of these projects is often associated with ‘exceptional’ regulations, including the bypassing or freezing of conventional planning tools, and the creation of project agencies with special or exceptional powers of intervention and decision-making. This ‘exceptionality’ reflects the primacy of project-based development over the holistic planning approaches, and eventually, it contributes to the increased spatial and social fragmentation in the city. Recapitulating, public-private partnership has replaced the tradition of regulatory town planning (Haussermann and Haila, 2005). And with the crisis of the welfare state, processes of decentralization6, deregulation, privatization7, and multi-actor policy-making (Van Kempen, 2007) have led the urban space to be delocalized and deeply commodified (Madanipour et al., 1998; Beauregard and Haila, 2000). These two qualities of urban space have impacted both business and residential areas in global cities, while also exacerbated the spatial manifestation of social polarization and income inequalities.

a. Delocalization of urban space

Delocalization (or despatialization) refers to a process started in the 1980s when real-estate market became global, as properties were both built and owned by transnational investors on the international market. Beauregard and Haila (2000) listed several examples for delocalization of

Spaces of globalization and inequality | 69 space in global cities in different regions of the world. In the United States during the 80s, cities of New York and Los Angeles had 21% and 46% of their office spaces owned by foreign investors respectively (Coldwell Banker’s survey, 1987) 8 . Albeit delocalization of space is caused by increased competitiveness and its consequent deregulation of the real-estate market, delocalization itself contributed to the increased global competition over limited local space (Fainstein, 2010), which eventually led to the commodification of the urban space.

b. Commodification of urban space

When urban spaces are perceived as commodities, they become under the control of market forces and the bidding power of money (XU et al., 2009). Accordingly, access to urban spaces is increasingly defined based on the ability to pay and the desirability for certain spaces that promote new opportunities for profitable investment. Noticeably, the implications of this commodification process are drastic. On one hand, it exacerbated the process of inner city gentrification. On the other hand, it created issues of exclusion and differential access to housing opportunities. Both are explained as follows:

3.2.2 Inner city gentrification

The process of gentrification has gained a particular attention in the global city’s literature. As shown previously, the different approaches for theorizing the growing urban inequality in global cities agree that gentrification is one of the most visible spatial outcomes of cities’ restructuring. Also, gentrification is considered the link between macro level processes of globalization and their micro level manifestations on neighborhood level (Hamnett, 2003; Butler, 2005) By definition, “Gentrification is a process involving a change in the population of land-users such that the new users are of a higher socio- economic status than the previous users, together with an associated change

70 | Theoretical background in the built environment through a reinvestment in fixed capital.” (Clark, 2005: 258). This process of gradual up-scaling of working class neighborhoods started after decades of urban decline affected most cities in industrial societies (Simon, 2005). Briefly, the suburban expansion during the industrial period of capitalism has contributed to the abandonment and property disinvestment in inner-city neighborhoods (Atkinson and Bridge, 2005; Smith, 1996), leading to the deterioration of housing condition as well as the depreciation of rent and property value. As a result these neighborhoods became affordable for the marginalized population including poor immigrants (Simon, 2005). All in all, the distressed neighborhoods of inner-city were characterized by high rates of poverty, concentration of ethnic minorities, deteriorated physical environment, and eventually, concentration of social problems such as high crime rates, violence, and drugs (MacLeod, 2002; Smith, 1996; Simon, 2005). With economic restructuring and the decline of manufacturing industries, many of the residents of the declining neighborhoods lost their jobs in the industrial sector, leading to high rates of unemployment and job insecurity in inner city (Simon, 2005). At the same time, decline of manufacturing contributed to the significant slowing down of suburbanization which was originally associated with the economic growth of industrialization (Kesteloot, 2000). With the middle-class back to the city, and with the formation of ‘transnational elite’ stirring the service industries, a potential confrontation between the rich and poor of inner cities occurred during the process of gentrification (Kesteloot, 2005). In more detail, due to the heightened demand for attractive residential conditions, corporate developers initiated the reinvestment in fixed capital in inner city’s declined neighborhoods (Clark, 2005; Butler, 2005; Ritzer and Ryan, 2011), a process which is facilitated by federal and local governments (Hackworth, 2002). More specifically, the reinvestment was confined to areas that are designated as ‘global’ and desirable as a profitable investment, while other ‘non-global’ spaces were left unnoticed (Rofe, 2009).’

Spaces of globalization and inequality | 71

“[W]hile the political invocation of an entrepreneurial urban agenda offers many inner-city spaces a spectacular makeover, it also risks deepening socioeconomic polarities along social cleavages like class, ethnicity, gender, age, and occupation.” (MacLeod, 2002; 606)

The reinvestment in inner city neighborhood contributed to a significant rise of property value in gentrified areas, and with the commodification and delocalization of urban space, workers who became unemployed - as a result of deindustrialization - and lower-income groups in general became no longer able to pay for dwellings in gentrified areas. As a result, gentrification generates a process of displacement and reconcentration of lower income groups (Atkinson and Bridge, 2005; Kesteloot, 2005). As the poor are displaced from their former residence, displacement is usually followed by their concentration in unattractive poorer neighborhoods with low-rent and most probably low quality (Marcuse and Van Kempen, 2000; Atkinson and Bridge, 2005; Kesteloot, 2005), which eventually contributes to the concentration of poverty in stark contrast to the gentrified areas of the highly paid professionals.

3.2.3 Issues of differential access

The unequal access to resources, services, or even urban space based on individual attributes can be aggravated due to privatization of these resources and services (Ritzer and Ryan, 2011; Mingione, 2005; Van Kempen, 2007). For example, the privatization of services such as health care, education, or housing leads to the increase in services costs - because private investors seek to maximize their profit – as a result, the added cost limit the access to crucial services only for people who can afford it. Similarly, patterns of differential access to urban space are experienced in cities in several ways. For example, access to public spaces can be limited for certain social groups either physically or symbolically or both, through

72 | Theoretical background management practices, or through deploying distinct design features such as gates and walls. In more detail, since the privatized urban projects (such as gated communities, shopping malls, … etc.) are increasingly valued as a financial asset (Haila, 1991; Beauregard and Haila, 2000). Thus, in order to protect their investments, private companies encourage the development of totally managed environment through heavy surveillance, and controlled access (Madanipour et al., 1998). Furthermore, a ‘symbolic’ restricted access can be generated by cues in the environment and design that are considered welcoming and inviting for certain target group, and at the same time the cues suggest an invisible no entry sign for the undesired population groups (Carr et al., 1992; Tiesdel and Oc, 1998). An example for such symbolic ‘no entry sign’ is a very expensive restaurant displaying the menu – outside the restaurant - including a price list of all its contents, the prices alone can exclusively attract the rich and prevent access to the poor. Since, exclusion and integration largely revolve around access (Madanipour, 2003). Then when a certain group of the population is denied access to certain resources, services, and public spaces based on their income, social class, or color, they become socially excluded (Peace, 2001). Based on this abstract definition of exclusion, it can be argued that commodification of space has elevated levels of social exclusion by promoting discriminatory access strategies to public spaces (that are organized or managed by the private sector). Equally, the process of commodification impacted levels of exclusion/inclusion from the housing market, as it led the access to housing opportunities to be increasingly defined by the household ability to pay for it, which has created new constraints/choices for people based on their socioeconomic status, family composition, and ethnic backgrounds. Based on this affordability scenario, and with the change in supply, demand, and regulations of local housing markets in global cities, two types of residential segregation are expected to emerge: (1) segregation of the poor and immigrants due the ‘lack of choice’ they have when it comes to decide

Spaces of globalization and inequality | 73 where to live, and (2) the self-segregation of the affluent population and ethnic minorities due to their preferences to maintain social status or cultural ties.

a. Segregation by ‘lack of choice’

The global city literature promotes the idea that lower income groups including poor immigrants are expected to be out of the competition for decent housing, because commodification of space implies a relatively free bidding for residential land and properties (Espino, 2005). Accordingly, disadvantaged groups who are excluded from the housing market are supposedly concentrated in the ‘residual’9 housing where they can afford (Kesteloot, 2000). In this sense, market-based factors are able to develop and intensify forms of spatial segregation in a city or even metropolitan area (Wassmer, 2005), or as Madanipour puts it “create enclaves for the rich and new ghetto for the poor” (2003: 185). The intensification of residential segregation of the poor is also explained by the decline in the redistributive power of the welfare state (mostly in Western European cities). On one hand, welfare cutbacks impact the income of those who are dependent on the state. Consequently, they are further excluded from the housing market and trapped in very limited housing choices. On the other hand, cutbacks also affected the supply of public housing units due to the diminished subsidies provided by the state, causing further shrinkage of the possibilities of affordable housing for disadvantaged group (Van Kempen, 2007). As highlighted earlier, another reason for the spatial concentration of poverty is the displacement and reconcentration of lower-income groups as a result of inner-city gentrification (Kesteloot, 2005; Atkinson and Bridge, 2005). Kesteloot (2005) points out the concept of ‘repressive city’ to describe the potential confrontation between rich and poor in inner cities during the process of gentrification, and the consequences of this confrontation especially on the poor. He explains, the social and spatial

74 | Theoretical background repression of the poor occurs on different stages. Socially, repression of the inner-city poor aims to reduce crime rates in order to attract potential investors to the targeted area. Spatially, repression takes the form of displacement of the poor to free space for the middle and upper classes. However, in case that displacement and concentration of the poor are not successfully completed, the confrontation of the rich and the poor in inner- city will remain. As a result, achieving security for the gentry class causes another round of social repression of the poor through “heavy police presence and zero tolerance” (Kesteloot, 2005: 143). This concept of ‘repressive city’ implies that disadvantaged population is perceived to be dangerous and threatening. Thus, ‘repressive city’ opens the discussion for another important issue that is ‘otherness’ or ‘strangeness’, which is highly relevant to the dynamics of residential segregation, especially ethnic based segregation. From social psychology perspective, estrangement, prejudice, and misunderstanding are most likely the implications of the socio-spatial dividing lines between population groups, black and white, rich and poor, gentry and working class (Marcuse and Van Kempen, 2000). Sennett (1970) explains the devastating impacts of such social disconnection, when different population groups live in separate neighborhoods with minimal interaction, they become misinformed about one another, and eventually they become unable to develop, to be tolerant, or to compromise. For example, white suburban residents of American cities used to perceive inner-city residents as the undesirable, alien, hostile, and dangerous ethnic "other" (Goldsmith, 2000; Peach, 1996; Boal, 2005). Furthermore, fear generated by poverty - and its associated crime rates and other social problems – is elevated when poverty is connected to certain ethnic groups, and the result is “double otherness”, socioeconomic and ethnic (Kesteloot, 2005: 142). Accordingly, exclusion of poor immigrants from the housing market can be explained based on a twofold assumption. On one hand, part of ethnic

Spaces of globalization and inequality | 75 segregation in cities is generated by prejudiced practices performed by individuals or institutions against ethnic minorities. In the USA for example, institutional and governmental practices such as redlining and exclusionary zoning have contributed to higher degree of spatial segregation of minorities (Marcuse, 2005), and those institutional conventions are further intensified by the discriminatory practices of the property owners, real estate brokers, lenders, insurers (Espino, 2005; Wassmer, 2005). On the other hand, with the total elimination of all bigoted views of race/ethnicity and the institutional practices that further them, market-based factors would still drive some forms of ethnicity-based spatial segregation in a metropolitan area (Wassmer, 2005). Because, if racial and ethnic minorities belong to lower income group, then – based on affordability – their concentration in poor neighborhood is also resulting in higher levels of spatial segregation by race/ethnicity (Vandell, 1995), and the result is that poor immigrants live in the worst places in town where nobody else lives (Peach, 1981; Marcuse and Van Kempen, 2000). Yet, it cannot be ignored that low socio-economic status of immigrant is itself a product of other forms of exclusion from societal resources such as decent education and job opportunities (Ritzer and Ryan, 2011). And with the increased pace of immigration due to globalization, the issue of exclusion becomes even more pressing (Madanipour, 2003).

b. Segregation by ‘choice’

Despite the fact that the highly paid professionals have no constraints that limits their access to the housing market. Still, their individual behavior, personal preferences, and selection characteristics can dramatically contribute to a higher degree of residential segregation (Van Kempen and Özüekren, 1998; Van Kempen, 2007; Vandell, 1995; Wassmer, 2005). Their choices are usually based on type of housing, local services, and other residents’ attributes (race, income, wealth, education and family composition) in the target neighborhood. For example, people who desire a

76 | Theoretical background certain service (such as K-12 public education in the US), and can afford to pay for it, will be concentrated where the service is provided (Wassmer, 2005). The result is a clustering of high-income households closer to spatially based amenities. Other reasons for self-segregation of the upper class in ‘exclusionary enclaves’, ‘citadels’, or ‘gated communities’ (for definitions see Marcuse, 2005) are maintaining social status, maintaining superiority of power and wealth, safety, security, and avoidance of the ‘other’ poor and ethnic minorities (Marcuse, 2005; Espino, 2005; Feitosa et al., 2011). Similarly, the voluntary segregation of ethnic minorities – especially non-poor immigrants – is caused by their desire to cluster together to seek protection and sense of belonging (Fischer, 1976). Eventually, the cluster may evolve into an ethnic enclave (Boal, 2005), and the larger the ethnic group, the larger their ability to support institutions that reinforce ethnicity, such as churches, stores, clubs, etc. (Fischer, 1976). For example, the voluntary concentration of the Jews in medieval towns is an early example of self-segregation described by Wirth (1928) in his book ‘the ghetto’, he elaborated: “to the Jews the geographically separated and socially isolated community seemed to offer the best opportunity for following their religious precepts, of preparing their food according to the established religious ritual, of following their dietary laws, of attending the synagogue for prayer three times a day, and of participating in the numerous functions of communal life which religious duty imposed upon every member of the community” (p. 19). Wirth’s description of the Jewish ghetto is, in fact, still applicable on several ethnic enclaves that exist today in Western Europe and North America, such as the Moroccan neighborhood in Brussels and the Indians in Manchester, where residents of the enclave invest in their neighborhoods so that buildings, streets, and shops would reflect their particular cultural identity. Finally, it is worth mentioning that many researchers perceive ethnic segregation to be positive (Varady, 2005), not only because the choice to live there is voluntary (Boal, 2005), but also because large ethnic enclaves

Spaces of globalization and inequality | 77 support "ethnic entrepreneurship". Where the ethnic space provides a context for the development of ethnic businesses, professional services (such as lawyers, teachers, doctors, and travel agents), and so on, all oriented to the specific needs of particular ethnic groups (Briggs, 2005). Accordingly, the enclave is even considered as an economic asset, because the development of ethnic businesses or services will eventually nourish the economy of the region (Qadeer, 2005). Yet, whether an enclave is good or bad, still, living in an enclave limits the intergroup social interaction and generates ‘avoidance’ relation between the dominant group and the segregated minority (Varady, 2005; Peach, 1996), which eventually intensifies residential segregation and its consequent social problems. Figure (3-1) sums up the theoretical assumptions, which correlate the commodification of space (in an already socially polarize city) to the spatial concentration of population with similar attributes in a relatively flexible housing market. Noticeably, the process discussed in the figure lacks several factors that could appear in individual cities due to their local contexts. For example, the theories assume that the differential access of social groups to the housing market will automatically lead to poverty concentration as a result of a rapid residential mobility in a globally deregulated housing market. Yet in fact, several global cities have strict housing allocation programs or social mixing policies (see chapters five and six) that can have a profound effect on the patterns of residential mobility, and consequently, the resultant patterns of poverty concentration. In other words, the process described in figure (3-1) could be generally applicable on the hypothetical global city that shows extreme cases of commodified housing market, high rates of residential mobility, intense gentrification, and declined public housing provision. Yet in case that one of these conditions is not fulfilled, then the characteristics of the resultant spatial change in global cities could not be predicted. In order to investigate in more detail the predictability of the spatial change in global cities, the subsequent section discusses the possible influence of local contexts in global cities on their spatial outcomes.

78 | Theoretical background

Changes in social and cultural Competitiveness: delocalization of structures of global cities urban spaces

Inequality – Polarization – Commodification of space Ethnic heterogeneity

Differential access to the housing market Exclusion/Inclusion

Declining Preferences Constrains public housing Gentrification provision

Deregulated housing market – residential mobility

Self segregation by choice Lack of choice

Affluent Migrants Poor population Displacement & cluster Poor migrants population Gentrified areas isolate reconcentration together cluster in concentrated in inner city for themselves for, of the poor in seeking sense existing poor in residual middle and safety, security deprived of belonging enclaves housing upper class & maintaining neighborhoods and unity market social status

Ethnic Poverty segregation concentration Affluence concentration

Residential segregation based on socioeconomic status and ethnic background

Figure 3-1 the production of residential segregation in socially polarized global cities due to the commodification of urban space.

Spaces of globalization and inequality | 79

3.1 Local contingencies and the individuality of cities

“Cities are restructured to present attractive working and residential conditions, they are endowed with new economic and social infrastructure and try to host decisionmaking functions and to concentrate consumer power through tourism and cultural events. They build up an image reflecting their position as an attractive center for inward investment in the global economy.” (Kesteloot, 2000: 202)

In their attempt to build such attractive image, global cities are expected to share trends of rapid gentrification, market-oriented urban development, deregulated housing markets, and eventually intensified spatial segregation. The correlation between economic restructuring and spatial segregation in global cities is established in the literature based on; first, economic restructuring leads to changes in cities’ occupational structures, which in turn, contributes to growing levels of income inequality and/or social polarization in global cities. Second, under the conditions of market-oriented urban policies, the commodification of urban space defines the access of people - on the two ends of the polarized social structure – to housing opportunities based on their ability to pay, and the result is the concentration of excluded populations in certain geographical areas (Rodriguez et al., 2003), in a juxtaposition to the gentrified areas of the affluent population. However, in order for this affordability scenario to explain patterns of spatial segregation in wide range of global cities, a number of contextual differences among these cities are rendered invisible. In more detail, the concept of ‘local contingencies’ - discussed in chapter two – suggests that in different contexts there is a number of moderating factors (Van Kempen and Murie, 2009), which can affect the process of restructuring in individual cities, as well as, the outcomes of restructuring.

80 | Theoretical background

For instance, the affordability scenario suggest that low-income households get displaced, and high-income household voluntarily move to desired housing in rates of residential mobility that are relatively high enough to stimulate spatial segregation. Yet, in different contexts, residential mobility is not solely defined by market forces, but it is also dependent on other aspects such as the flexibility of local housing market and homeownership rates. In Latin American cities, neighborhoods are most likely to transform due to social upward (or downward) mobility of its residents instead of their physical mobility. As the financially successful households tend to expand and consolidate their homes rather than move from a neighborhood to another, in part due to inefficient housing markets and high transaction costs (Monkkonen, 2011). While in the US, social advancement of residents usually generates high levels of residential mobility among renters and even among homeowners (Espino, 2005). Accordingly, different patterns of residential mobility are expected to lead to different patterns of change in spatial segregation in cities of the two contexts. Similarly in the case of ethnic segregation, the ease of labor mobility contributed to the increase of cross-borders immigration flows (Sassen, 2001; Castles, 2002), yet patterns of concentration of these immigrants in individual global cities are also context-contingent. In western European cities, the existing post-war ethnic enclaves performed as nuclei where new comers to the city tend to settle around (Ritzer, 2004). While in the Arab Gulf states - such as United Arab Emirates (Keane and McGeehan, 2008), Bahrain (Gardner, 2010) and other recent migration-receiving countries of the region – patterns of concentration of foreign workers are not historically originated, instead they are defined by state policies and restrictions. From this perspective, what are the other ‘moderating factors’ that can also impact the process and outcomes of spatial restructuring of global cities? Are these factors influential enough to produce significantly different patterns of spatial segregation in individual cities? And finally, by assuming

Spaces of globalization and inequality | 81 that local contexts are able to alter the outcome of spatial restructuring in global cities, then to what extent the global/divided city model is expected to be applicable on cities undergoing similar processes of economic restructuring? In an attempt to answer these questions, the research examines both (1) the validity of the ‘global/divided city model’, and (2) the influence of local contexts on the outcomes of restructuring in individual global cities, this examination is based on an extensive analytical study introduced in the subsequent chapter.

82 | Theoretical background

Note

1 Also defined as the ‘time-space distanciation’ (Giddens, 1990) 2 ‘Time-space compression’ describes the reduction of distance experienced through the decrease in the time taken, either to cross a space physically by means of transportation, or symbolically by means of communication (Ritzer and Ryan, 2011). 3 Localities here refer to small-scale territories or places such as industrial districts and city-regions, and not necessarily referring to state-centric territories in the traditional sense. 4 The concept of ‘creative destruction’. Harvey (2010: 46) compares differences between Marx and Schumpeter's usage of the concept: "Both Karl Marx and Joseph Schumpeter wrote at length on the 'creative-destructive' tendencies inherent in capitalism. While Marx clearly admired capitalism's creativity he [...] strongly emphasized its self-destructiveness. The Schumpeterians have all along gloried in capitalism's endless creativity while treating the destructiveness as mostly a matter of the normal costs of doing business" 5 The analysis is developed by contributors of the URSPIC (urban restructuring and social polarization in the city) research project conducted in 1997-1999 and coordinated by Frank Moulaert. 6 In order to promote liberal atmosphere for global business activity, the formerly dominant nation-state level of governance have declined and shifted into two main parallel forms of governance. On one hand, decision-making process is now taking place on decentralized subnational levels like provinces, regions and cities due to the increased regional competition (Kazepov, 2005; Van Kempen, 2007). On the other hand, various governance functions shifted from governments to the corporate world (Sassen, 2000), where supranational institutions, transnational corporation, and international financial institutions such as World Bank, the International Monetary Fund IMF, and the World Trade Organization WTO are downscaling central governments in order to foster relatively easier mobility of capital, labor, and goods (Kazepov, 2005). Those institutions influence domestic policies by their ability to control funds and loans, while funds are tied to "structural adjustment" that guarantee powerless government (Goldsmith, 2000), leading to ‘deregulation’ which is the undermined ability of local government to impose regulative decisions on local markets (Hancher and Moran, 1989). Moreover, the corporate pressure on regional governments has led decision- making processes to be shared with other private organizations and individuals (Van Kempen, 2007), local governments are no longer playing an exclusive role as the leading policy-maker; they are merely one of the many actors involved in the multi- actor policy-making process (Healey et al., 1995; Elander and Blanc, 2001; Haussermann and Haila, 2005; Geniş, 2007).

Spaces of globalization and inequality | 83

7 The welfare crisis was parallel to a wave of privatization of business, enterprises and public services (Sullivan, 1987). On the employment front, privatized businesses seek profit-maximizing organizational restructuring (Fernandez et al., 2007), by adopting downsizing strategies (Burke and Cooper, 2000). Leading to mass layoffs of significant numbers of public employees. Also, profit-oriented businesses focused on flexible forms of employment in order to reduce their costs, such as part-time workers. On the services front, privatization of basic services created difficulties in accessing crucial sectors such as health, education, transportation, and housing for large sector of the population (Ritzer and Ryan, 2011; Mingione, 2005; Van Kempen, 2007), as privatization of services leads to the increase in services costs, and accordingly limit the access to crucial services only for people who can afford it. 8 no data for recent years available for comparison 9 Residual housing refers to old and private rented dwellings in inner city declined neighborhoods that are usually in bad condition and relatively cheap (Kesteloot et al., 1997)

84 | Theoretical background

Chapter 4

4 Testing the global/divided city model: an introduction to the analytical study

Based on the discussion in the preceding chapters, the widely asserted association between economic restructuring and intensifying socio-spatial divisions within global cities requires verification. Although chapter two concludes that the three schools of thought agree on the growing gentrification and spatial segregation to be basic characteristics of the ‘global city’. Yet, a number of issues raised in the discussion suggest that this association might be overstretched: • The possible influence of local contextual differences on the outcome of socio-spatial restructuring of individual global cities is marginalized. Whilst the third school of thought (see Marcuse and Van Kempen, 2000 for example) highlights the importance of such local contingent factors, still, they suggest that growing inequalities within cities is unavoidable, and no actual influence - of local factors on the predefined socio-spatial divisions - is stated. • The impacts of economic restructuring on other political, social, and spatial aspects of the global city are theorized based on the analysis of top-ranked global cities (especially New York and London). However, the possibility that such a global model is applicable on other cities - as long as they share parallel transformations in their economic functions – is not confirmed. Especially with the scarcity of

86 | Data and Method

comprehensive empirical evidence to support the general applicability of the global/divided city model on other cities beyond New York and London. • The presumed automatic correlation between the growing social divisions and its spatial dimension (socioeconomic and ethnic segregation), is based solely on assuming a free operation of land and property markets in global cities under conditions of global competitiveness and market-oriented urban development. While in fact, other factors such as homeownership rates, residential mobility, and public housing provisions are defining the situation of the housing market in different cities, and therefore, assuming a free operation of the market regardless these factors does not capture the complexity of the housing market dynamics in global cities.

4.1 Analytical framework

In view of the above, the analytical study introduced in this chapter aims to examine the association between economic restructuring and the changes in socioeconomic and ethnic segregation in global cities over time. Also, the study attempts to evaluate the influence of local contextual differences on the resultant patterns of segregation in individual cities. In order to reach this goal, the study intends to; first, identify a wide-range of cities that match the established description of the ‘global city’ in the global city discourse. Second, calculate the values of spatial segregation among different population groups within each city in two different time frames over the past two decades (time frames are defined based on data availability). Accordingly, each global city in the dataset will have two calculated values that indicate the magnitude and direction (increase or decrease) of change in spatial segregation level over a certain period of time. In order to interpret the resultant values of change in segregation level in global cities, several statistical approaches, linear regression analysis and

Data and method | 87 one-way analysis of variance ANOVA, are performed on the results to evaluate the validity of the widely accepted assumptions in the global city literature (see a detailed discussion of these statistical approaches and the different stages of analysis in the end of this chapter). The findings of the analytical study are expected as follows: • In terms of the direction and the magnitude of change in segregation level, the regression analysis assists in finding the correlation between the spatial change in cities and their level of global network connectivity. Also, the analysis gives an indication of to what degree the spatial change in global city is influenced and can be explained by its global status. • In case that the aggregate results show no shared pattern of change in segregation among global cities, i.e. the assumptions in the literature are rejected according to the significance probability, then an alternate explanation for the resultant patterns of change in segregation in global city is required. Therefore, a detailed discussion is provided to identify the local aspects involved in the process of spatial change, and to assess the influence of these aspects in an attempt to understand the complex dynamics in which socio-spatial patterns of global cities are produced.

4.2 Which city is a global city? – Dataset selection

Many attempts are made to classify and categorize cities of the world according to their global position, their size, their main economic or cultural activity, and their connectivity to other cities in the world. For example, Foreign Policy with A.T. Kearney and The Chicago Council on Global Affairs have created the Global Cities Index, which comprehensively ranks 60 cities’ metro areas across five dimensions: business, human capital, information exchange, cultural experience, and political engagement (Foreign Policy, 2010). Similarly, other classifications such as Global Power

88 | Data and Method

City Index (MMF, 2011) and Mercer’s Quality of Living Survey (Parakatil, 2011) are also giving an insight about cities’ position through multi- dimensional analysis including factors of business, environment, safety, livability, cultural interaction…etc. Finally, there is the ‘Globalization and World Cities’ (GaWC) classification of world cities (Taylor, 2001; Taylor et al., 2002; Taylor et al., 2011), which categorizes cities according to their share of advanced producer services firms, as well as the potential inter-city flows generated by these firms. Since the subject in question (the global/divided city model) is constructed theoretically based on ‘the global city’ (Sassen, 1991), ‘the dual city’ (Mollenkopf and Castells, 1991), ‘the divided city’ (Fainstein et al., 1992), and other hypotheses discussed in chapter two. Then, this study has to develop its finding based upon a dataset that fulfills the criteria sketched by the leading theorists of the global city discourse; criteria that define the main characteristics of the ‘global city’, and distinguish which city is qualified for world or global city status. For example, Fainstein and Harloe (2003) state six aspects of global economic restructuring that are relevant to the understanding of what is meant by “global city status” (p. 159). The aspects range from the decline in manufacturing, the rise of advanced producer services, the concentration of financial services, to the growth of consumer services to meet the needs of high income workers and the growing demand from international tourism. Noticeably, these aspects are very similar to Sassen’s (1991; 2001; 2006) specification of the global city. They agree on the presence of higher shares of advanced producer services and financial institutions in cities that are the ‘command and control centers’ of the global economy (Sassen, 2001; Haussermann and Haila, 2005). Where basically, global cities host management and coordination centers of a dispersed transnational network of factories, branch offices, and service outlets (Sassen, 2000). Therefore, the function of global cities exceeds being a location or production site for the services industries, according to Castells (1996); global cities are key

Data and method | 89 nodes and hubs in a global ‘space of flows’ through which the information economy is operating; in his view, the global city is a phenomenon or “a process that connects advanced services, producer centers, and markets in a global network” (p. 380), or as Sassen (2002: 17) describes them; global cities “are a function of cross-border networks”. From this perspective, discerning a global city’s status has to incorporate both the location of advanced producer services (Sassen, 2001), as well as the connections and global flows among cities containing them. As, these connections are essential to understand the role of global cities as key nodes in a ‘global network’ (Castells, 1996). As noted, there is a broad consensus on few cities (especially London and New York) to be at the top of the global city hierarchy (Beaverstock et al., 1999), while below the top- ranked cities, there are still attempts to empirically assess other cities of the world and classify them according to their global status. Yet in this regard, several limitations have faced the available research to arrange cities in relation to one another in a global economy. On one hand, it is proved problematic to rank cities solely based on the theoretical specification of the global city, Fainstein and Harloe (2003) highlights this issue after a detailed discussion of the six aforementioned aspects that define the ‘global city status’, they concluded, “there is, therefore, no simple division between these two global cities (London and New York) and other cities. All of the six aspects of global economic restructuring discussed exist elsewhere” (p. 161). In this case, what differentiates between cities that share similar economic functions is the number and size of the advanced producer services hosted by a city compared to other cities, which is considered one of the standard measure for global city ranking. However, according to Sassen (2001), the presence of top-ranked firms alone is not an adequate measure to understand what makes a city a global city. As, this measure provides information about the internal restructuring of cities, while gives no or limited insight on the inter-

90 | Data and Method city flows generated by the function of cities as nodes where various global networks intersect. On the other hand, it is also proved problematic to measure and compare actual business and information flows between global cities. Due to the lack of easily available data that indicate the size and directions of these flows, such as the data on the business-related communications between head offices and other branches or other service firms around the world (Short and Kim, 1999). Therefore, according to Taylor et al. (2011: 4), there is no ‘simple’ or ‘direct’ measure of interurban information flows. As, the lack of data leads to a gap between the theoretical understanding of the global city status and the empirical evidence upon which the global city rankings are based. According to this brief discussion of the factors determining the global city status in the global city literature, as well as the limitations facing the available rankings of cities based on their global status. This study aims to adopt one of the aforementioned cities’ classifications to constitute its primary dataset, given the condition that the selected classification matches the requirement of the study, through providing robust indications on the size of advanced producer services firms in global cities, and their potential inter-city work flows. Thus, when these indications are compared to the correspondent level of change in spatial segregation, the association between economic globalization and socio-spatial divisions within cities can be verified.

4.2.1 Primary dataset

Since identifying the ‘global city status’ is narrowly focusing on the changes in the economic functions of cities rather than incorporating other diverse forms of globalization (Robinson, 2005). Accordingly, the broad scope of classifications – such as ‘Global Cities Index’ (Foreign Policy, 2010) and ‘Global Power City Index’ (MMF, 2011) - is adding to the complexity of the analysis by incorporating several variables (culture,

Data and method | 91 safety, livability…) while giving a narrow perspective about the economic dimension, which is the main theme in the global city literature. Therefore, to overcome the complexity generated by the multi-dimensional classifications, the study deploys one of the well-known, yet debated, classifications of world cities that is actually criticized for its limited focus on financial and business services as criteria for defining cities’ global status (Robinson, 2005), this classification is developed by the research network GaWC (Globalization and World Cities), centered in the Geography Department at Loughborough University, UK. The choice of the 2010 GaWC classification is made for several reasons, the first and foremost is that the GaWC research is building heavily on Sassen’s (1991) emphasis on advanced producer services, and Castells’ (1996) idea of cities as nodes in a ‘space of flows’ (Taylor et al., 2011; Beaverstock et al., 2011). Secondly, the GaWC analysis offers a comprehensive attempt to “cut through the conceptual confusion” (Taylor et al., 2011: 4) of the global city discourse by specifying an empirical model for the ‘world city network’, which is not only derived from Friedmann- Castells-Sassen seminal work, but the model also offers a quantitative approach to measure and assess the magnitude of potential flows and relations between world cities. Finally, the classification includes reasonably large number of cities, and therefore it overcomes the limitations of the available research that focuses mainly on a number of major cities while overlooks the rest of cities through which the global network of firms is operating (ibid.).

a. GaWC classification methodology

As noted, basically, advanced producer service firms (e.g. accountancy, advertising, banking/finance, insurance, law, and management consultancy) are concentrated in global cities while these firms play crucial role in the formation of a world city network. Because in order for these firms to provide a 'seamless' service for every client regardless of his location

92 | Data and Method

(Taylor, 2001), each firm is typically operating through multi-city office network (Taylor et al., 2011). Accordingly, the generated flows of information among offices located in different cities are what constitute the basis of the world city network. Yet, since data on actual business and telecommunication flows is not available (Short and Kim, 1999), then the classification had to be conducted based upon indirect measure of flows that is “the potential working flows” among firm’s offices (Taylor et al., 2011: 4). In more detail, the assumption here is that, the larger and more important the office in a city (headquarter, regional office, …, to minor office), the more flows of information and knowledge it will generate from and to the city (ibid.). For example, the potential flows between two cities hosting major offices of a firm are expected to be much larger than flows between two cities hosting minor offices of the same firm. Accordingly, potential flows between each pair of cities can be estimated based on the size of office the two cities host, as the importance of any office in any city is given scores from 0 to 5, where 0 (a firm having no office in a city) to 5 (a city housing the headquarters of a firm). Finally, aggregating the scores for all firms in one city gives a ‘service value’ of the city, which is an indicator of how important a city is to the advanced service firm’s network. While at the same time, the total potential flows generated by the high (or low) service value of the city provides an estimate of the ‘network connectivity’ of that city, which is an indicator of the degree of integration of the city into the world city network (ibid.). In the year 2000, the first GaWC classification of cities was based on the number (and not the size) of offices of 100 global services firms in accountancy, advertising, banking/finance, insurance, law, and management consultancy, across 315 cities worldwide (Taylor, 2001). In later years, Collaboration between the Global Urban Competitiveness Project (GUCP) at the Chinese Academy of Social Sciences and GaWC researchers at Loughborough and Ghent Universities made possible a much larger and

Data and method | 93 complete data collection of advanced producer services firms. The 2008 ranking is based upon analyzing the office networks of 175 advanced producer service firms in 525 cities, and the firms were chosen by their size instead of where their offices are located (Taylor et al., 2011). From 2008 onwards, network connectivity became the main measure of importance of cities (Taylor, 2011), which relates any given city to all other cities within the network through its share of work flows among advanced producer services firms. The connectivity measures are used to classify cities into five levels of world city network integration. These levels are explained later in this chapter.

b. GaWC criticism

Evidently, the GaWC classification of cities is based upon a comprehensive data collection, and according to Beaverstock (2011: 219), the project “should receive many commendations and plaudits from both the academy and policy-arena”. Yet their empirical approach is criticized based on several contentions. Including, Robinson (2005) argument that globalization involves more than financial and business services, and categorizing cities solely based on their connection to the network of advanced services firms leads to the overlooking of other world cities that are embedded in other networks of globalization. And although the GaWC team recognizes this fact, according to Taylor (2004), the GaWC team gathered evidence on other forms of globalization, including NGO location in different cities, and Nairobi emerged top of the list of global cities defined along these criteria. Still, Robinson (2005) criticizes Taylor (2004) for insisting on the narrow focus on financial and business services as the dominant network of globalization. In fact, this narrow focus (if any) on financial and business services as a criterion for categorizing cities made the classification highly suitable for the requirement of this research. On one hand, the scope of the classification eliminates the complexity generated by other environmental, cultural, or political variables involved in the multi-

94 | Data and Method dimensional classifications. While on the other hand, the GaWC team provides an assessment for both the location and the potential inter-office work flows between advanced producer services firms located in cities. And according to the later point, no further data about the economic functions of global cities is required to be collected or assessed by the author. Further, the discussion of the criticism of the GaWC methodology is lengthy, as one of many examples, GaWC critics focus on the shortcomings of estimating the magnitude of potential flows among firms, which in Thierstein et al. (2008) view, does not tell the whole story of the nature and quality of business activities between firms located in different cities. Still, the presence of producer services and the estimated levels of integration into the world city network might not be a comprehensive perspective for ranking global cities. However, these criteria, as shown, are highly relevant to the analytical framework – and the scope of this research in general.

c. Cities of the primary dataset

The 2010 GaWC classification list contains total of 525 cities categorized based on network connectivity measures into five categories (alpha, beta, gamma, high sufficiency, and sufficiency). According to the GaWC official website1, the five categories are interpreted as follows: • alpha++ cities: in all analyses, London and New York stand out as clearly more integrated than all other cities and constitute their own high level of integration. • alpha+ cities: other highly integrated cities that complement London and New York , largely filling in advanced services needs for the Pacific Asia. • alpha & alpha- cities: very important world cities that link major economic regions and states into the world economy. • All beta level cities: these are important world cities that are instrumental in linking their region or state into the world economy.

Data and method | 95

• All gamma level cities: these can be world cities linking smaller regions or states into the world economy. • Cities with high sufficiency and sufficiency of services: these are two groups of cities that are not world cities as defined here but they have sufficient services so as not to be overtly dependent on world cities.

According to the description of cities in each category, cities in the last two categories (high sufficiency and sufficiency) are excluded from the analysis, as they do not match the criteria of global cities discussed in the literature. Instead, these levels are comprised mostly of smaller capital cities, and traditional centers of manufacturing regions. Therefore, the presence of advanced producer services firms in cities of the last two categories is not comparable to those in other alpha, beta, and gamma cities, and accordingly, cities with high sufficiency and sufficiency of services are not expected to show significant changes in their occupational or income structures as a result of their insignificant level of integration into the global economy. The remaining alpha, beta, and gamma categories hold 178 cities, which constitute the primary dataset of this research, tables (4-1), (4-2), and (4-3) present cities of the dataset as well as their assigned ranks in the world city network. In order to perform the analysis on the alpha, beta and gamma cities identified by the GaWC research team, the study aims to explore different measures for spatial segregation, and then define the suitable method for evaluating the changes in residential segregation levels in the selected global cities. The selected spatial segregation measure has to fulfill the requirement of the research by being able to incorporate the characteristics of the physical space into the analysis, as well as being able to evaluate the segregation levels among several population groups simultaneously. The next section provides a review of number of indices of segregation, their parameters, and shortcomings.

96 | Data and Method

alpha cities according to the 2010 GaWC classification of world cities alpha cities

alpha++ alpha+ alpha alpha- 1 London 3 Hong Kong 11 Milan 29 Miami 2 New York 4 Paris 12 Beijing 30 Dublin 5 Singapore 13 Toronto 31 Melbourne 6 Tokyo 14 Sao Paulo 32 Zurich 7 Shanghai 15 Madrid 33 New Delhi 8 Chicago 16 Mumbai 34 Munich 9 Dubai 17 Los Angeles 35 Istanbul 10 Sydney 18 Moscow 36 Boston 19 Frankfurt 37 Warsaw 20 Mexico city 38 Dallas 21 Amsterdam 39 Vienna 22 Buenos Aires 40 Atlanta 23 Kuala Lumpur 41 Barcelona 24 Seoul 42 Bangkok 25 Brussels 43 Taipei 26 Jakarta 44 Santiago 27 San Francisco 45 Lisbon 28 Washington 46 Philadelphia 47 Johannesburg

Data and method | 97

beta cities according to the 2010 GaWC classification of world cities beta Cities

beta+ beta beta- 48 Dusseldorf 64 Budapest 83 Abu Dhabi 49 Stockholm 65 Beirut 84 Nicosia 50 Prague 66 Luxembourg 85 Birmingham UK 51 Montreal 67 Guangzhou 86 Rio de Janeiro 52 Rome 68 Seattle 87 Brisbane 53 Hamburg 69 Caracas 88 Geneva 54 Manila 70 Ho Chi minh city 89 Calcutta 55 Houston 71 Auckland 90 Detroit 56 Berlin 72 Oslo 91 Denver 57 Athens 73 Kiev 92 Monterrey 58 Tel Aviv 74 Chennai 93 Bratislava 59 Bangalore 75 Bucharest 94 Port Louis 60 Copenhagen 76 Manchester 95 Casablanca 61 Cairo 77 Karachi 96 Manama 62 Bogota 78 Lima 97 Stuttgart 63 Vancouver 79 Cape Town 98 Sophia 80 Riyadh 99 Cologne 81 Montevideo 100 St Louis 82 Minneapolis 101 Helsinki 102 Panama city 103 San Diego 104 Lagos 105 Perth 106 Shenzhen 107 Cleveland 108 San Juan 109 Calgary 110 Guatemala city 111 Osaka

98 | Data and Method

alpha cities according to the 2010 GaWC classification of world cities gamma Cities

gamma+ gamma gamma- 112 Glasgow 135 Valencia 153 Tallinn 113 Nairobi 136 Kansas city 154 Pune 114 Bristol 137 Phoenix 155 Porto 115 Hanoi 138 Almaty 156 Porto Alegre 116 Cincinnati 139 Guadalajara 157 Orlando 117 Charlotte 140 Lyon 158 Gothenburg 118 Antwerp 141 Quito 159 Marseille 119 Doha 142 St Petersburg 160 Ottawa 120 Lahore 143 Leeds 161 Colombo 121 Baltimore 144 Santo Domingo 162 Ljubljana 122 Jeddah 145 San Salvador 163 Tegucigalpa 123 Edinburgh 146 Vilnius 164 Richmond 124 Amman 147 Rotterdam 165 Islamabad 125 Hyderabad 148 Tampa 166 Muscat 126 Zagreb 149 Columbus 167 Durban 127 Adelaide 150 Indianapolis 168 Austin 128 Kuwait 151 Pittsburgh 169 Belfast 129 Portland 152 Edmonton 170 Guayaquil 130 Belgrade 171 Nagoya 131 San Jose CR 172 Turin 132 Tunis 173 Southampton 133 San Jose us 174 Milwaukee 134 Riga 175 Wellington 176 Curitiba 177 Accra 178 Georgetown cl

Data and method | 99

4.3 Measures of spatial segregation

A wide range of indices and measures for segregation was developed since the beginning of the 20th century, these indices differ not only in mathematical formula (Peach, 1981), but also in the type of the segregation they can measure, the number of population groups they can handle simultaneously, and how they incorporate ‘space’ as an important variable in the equation. This section gives a brief backdrop on the different indices available, and then provides criteria for choosing the index suitable for the requirement of this research. One of the very popular measure for segregation is the dissimilarity index D (Feitosa et al., 2004; Wong, 2005), which was initially introduced by Duncan and Duncan in (1955). The dissimilarity index D is defined as the proportion of population from a specific social group that would have to relocate within the city so that each area would have the same social composition of the city as a whole. This index was originally limited to measure segregation between only two population groups (mostly segregation of African Americans – referred to as Blacks - and Whites in the US). The value of D ranges from 0 to 1 where 0 is the maximum integration between population groups and 1 is the complete segregation (ibid.), and is defined as:

where bi and wi are black and white population counts in areal unit i, and B and W are the total black and white population counts of the entire study region, respectively. Dissimilarity index along with other indices such as P* exposure index (Bell, 1954), the Atkinson index (Atkinson, 1970), and many others, have

100 | Data and Method been criticized for being limited to two groups analysis. Therefore, a second generation of indices was developed to address multi-group issues (Feitosa et al., 2004). However, the multi-group measures have been also criticized for being aspatial (Wong, 1993), and unable to incorporate the spatial arrangement of population among areal units, which is an essential aspect in segregation studies (Reardon and Firebaugh, 2002; Feitosa et al., 2004). Accordingly, several studies were developed to overcome the spatial deficiency of the original D index and other multi-group indices (e.g., Wong, 2002; Reardon and O'Sullivan, 2004; Feitosa et al., 2004). The dˇ(m) index is a spatial version of the dissimilarity index D developed by Feitosa et al. (2004), dˇ(m) aims to measure how the population of each locality differs, on average, from the population composition as a whole, and tries to overcome the ‘checkerboard problem’ associated with the original D index. The checkerboard problem is easily explained by an example of a city divided into several census tracts, which are distributed in white-black checkerboard form, with assuming that white population reside mostly in the white tracts and black population reside mostly in black tracts. In this setting, calculating dissimilarity index will give a certain value, then by shuffling the white and black squares on different sides of the board and calculating dissimilarity index once again. The problem emerges when the value of D stays the same despite the added spatial concentration of census tracts populated by each group (Behr, 2004). As shown in figure (4-1), Feitosa et al. (2004) have created three artificial cities populated by 4 population groups and divided into 144 areal units with equal dimension. In case A, population groups are distributed in extreme segregation, where each areal unit has only individuals of one group and the units characterized by the same group are clustered. In case B, each areal unit has individuals of one group, but units are distributed evenly on the board. In case C, population groups are distributed in extreme integration, where each areal unit has the four population groups in equal shares.

Data and method | 101

D=1

dˇ(m)=0.86

D=1

dˇ(m)=0.05

D=0

dˇ(m)=0

Figure 4-1 Artificial dataset explaining the checkerboard problem and how spatial measures can overcome the deficiency of aspatial indices. Source: (Feitosa et al., 2004).

102 | Data and Method

By calculating D and dˇ(m) for the three hypothetical cases, although case A has a much more segregated distribution than case B, the D index value in both cases is the same, indicating maximum segregation. While the dˇ(m) showed a higher sensitivity to the spatial distribution of areal units. For case C, both measures indicated maximum integration, because unless all population groups are present in a single areal unit, the value of D and other aspatial indices will show a case of extreme segregation (ibid.). Wong (2002; 2003) developed another elaborate spatial index D(s), which is a two-group index based on the original aspatial dissimilarity index D. In general, the D(s) overcame the spatial deficiency of D by incorporating the geometric characteristics of areal units into the evaluation of segregation (Wong, 2003), where the ratio between perimeter and area of different neighborhoods is a part of the D(s) calculations. It is defined as: XX 1=2 ðÞþP =A P =A ðÞ¼ 1 i i j j Ds D wij zi zj MAXðÞ= 2 i j P A

where D is the original dissimilarity index, as defined earlier; zi and zj are the proportions of the minority population group between areal units i and j;

Pi/Ai is the perimeter-area ratio for areal unit i; and wij is defined as

Pdij wij ¼ dij j

where dij is the length of the shared boundary between areal units i and j.

The length of shared boundary is used as an indication for interaction between areal units, where the longer the shared boundary the higher the chance of interaction between different population groups across different neighborhoods. The problem with this index is the limited number of population groups it can handle simultaneously. Therefore, Wong (1998) developed another multi-group spatial index SD(m) based on the multi- group version D(m) - of the original dissimilarity index D - introduced by

Data and method | 103

Morgan (1975) and Sakoda (1981). The D(m) can accommodate more than two population groups, but the measure is aspatial and rearranging populations among areal units will not change the overall level of segregation (Wong, 2003) where:




















where:


and

Nijis the population count of the j population group in areal unit i

Ni is the total population in areal unit i

Nj is the total population of group j in the entire study region N is the total population in the entire region P is the proportion of population in group j

The formulation of the spatial version of D(m) - proposed by (Wong, 1998) - is based upon the concept of composite population counts. The composite population count of areal unit i for group j is defined as








where d( ) is a function defining the neighborhood of i, and based upon the premise that within the neighborhood of i, people belong to different population groups can interact as if they are in unit i. After the composite population counts for all areal units are computed, they are used to calculate D(m) as if they are the original population counts. Therefore, the spatial version of D(m) is SD(m) and defined as:





 
















104 | Data and Method

Based on the detailed review of segregation indices, the suitable index to be utilized in this research has to fulfill twofold criteria. Firstly, the chosen index must incorporate the spatial aspect into the analysis, as the aim of the study is relating spatial changes to the economic restructuring of cities (and also to avoid the checkerboard problem). This first condition has eliminated all the aspatial measures, leaving us with the spatial two-groups dˇ(m) index by Feitosa et al. (2004), two-groups D(s) index, and multi-group SD(m) index by Wong (1998, 2003). Secondly, another important condition is that the index must capture the diversity of contemporary cities and measure the spatial segregation among several population group simultaneously, and since dˇ(m) and D(s) are suitable only for two-groups comparisons, thus they were excluded. Accordingly, this study and its conclusions are built on the values of ‘spatial multi-group dissimilarity index SD(m)’ created by Wong (1998), and the calculations of the index is generated using an ArcView tool provided also by Wong (2002, 2003).

4.4 Data type

In order to calculate the SD(m) index, each city in the GaWC list of world cities requires census data on local level such as division, neighborhood, census tract...etc. in two different years within the past 20 years. For the calculations of socioeconomic segregation, count of population/households in different segments of monthly or yearly income is required, as well as their spatial distribution over different neighborhood/tracts in the city. For the calculation of ethnic segregation, count of population from different ethnicities or different national origins (depending on the type of official data presented in the different national censuses) is required, also with their spatial distribution over the city. For example, the data used in the analysis of the city-state of Singapore is as follows: the city is divided into 35 populated planning zones defined by

Data and method | 105

Urban Redevelopment Authority. Population is categorized into 4 ethnic groups (Chinese, Malay, Indians, and others), and 9 income groups according to their monthly income (below 1000$ - 1000$ to 1499$ - 1500$ to 1999$ - 2000$ to 2999$ - 3000$ to 3999$ - 4000$ to 4999$ -5000$ to 5999$ - 6000$ and over). Also, the index is calculated for the city on the years 2000 and 2010 based on the availability of data.

4.5 Limitations and constrains

Several challenges have faced the analysis process throughout the data- gathering phase, unifying data formats, calculating segregation indices, and interpreting the results. These challenges are summarized as follows:

4.5.1 Data availability

One of the biggest challenges faced this research during the data- gathering phase, which affected the whole analytical process, was the availability of the required official census data in different global cities of the primary dataset. The scarcity of the data has consequently led to:

a. The final dataset is geographically unbalanced. Since the analysis is based on official data collected by the author from available resources including, statistics yearbooks, national and regional census, and governmental reports. Therefore, all cities of the Middle East and some African, Asian, and Latin American cities showed a huge difference in data quality and availability compared to North American, European, and Australian cities. Briggs (2005) explains the gap in data quality by the scarcity of statistical data in many developing nations and the traditions of data collection that refuse collecting certain data such as race and wages as a matter of policy in some countries. For example, when it comes to economic characteristics of population, the Egyptian census of population avoids all data about wages, income, or taxes, the only data available is describing the economic activities performed by

106 | Data and Method

population in workforce over 16 years of age. Similarly, censuses of other developing countries lack essential data, which caused the dataset to be – to a certain degree - geographically unbalanced.

b. Since all regions of the world are not equally represented in the final dataset, this raises the issue of sampling bias. Briefly, sampling bias refers to the selection of non-random sample of a population causing some members of the population to be less likely to be included than others, resulting in a biased sample, in which all participants are not equally balanced or objectively represented (Kirkwood, 1988). The problem with sampling bias is that it undermines the ability of the test results to be generalized to the rest of the population (ibid.). In other words, if the primary dataset contains 178 global cities, and due to data availability this number is reduced to 91 cities, then the behavior of the 91 cities under the test could not be generalized to the 178 cities in case that sampling bias is suspected. Sampling bias is very common in real- life samples because it is practically impossible to ensure a perfectly random sample, and there are several techniques to correct the effect of the bias (see Cortes et al., 2008; Little & Rubin, 1986; Heckman, 1979). However in our case, it can be argued that sampling bias does not affect the quality of the analytical study and its outcomes for the following reasons: firstly, the final dataset might not be geographically random, but in terms of the global status, the final dataset (see section 4.6) contains 32 alpha, 31 beta and 28 gamma cities, and since the main parameters under test are globalization and its associated spatial change, then the sample can be treated as a reasonable approximation to a random sample as it contains similar shares of city from each global category. Secondly, the selection of cities is based solely on data availability, which means that sampling bias is not a result of a deliberate exclusion of subjects, pre-screening of subjects, or other misleading selection criteria by the aim of controlling the statistics computed from the test.

Data and method | 107 c. The scarcity of the available data also caused the analysis to be performed on the scale of the administrative divisions within cities where the suitable data is available. For example, in a certain city the distribution of the socioeconomic groups is available on the municipality level, while on the neighborhood level, the only data available is the median income of the population; in this case, the analysis is performed over the scale of the municipality due to data availability. Accordingly, it was extremely difficult to compute the variation in segregation between population groups at various scales within the city, which affected the comparability of the results in later stages of the analysis (see section 4.5.3), and also added to the difficulty of dealing with the modifiable areal unit problem (MAUP). Briefly, MAUP refers to the possible variation in correlation values when different boundaries, such as administrative boundaries, political districts and census tracts, are used in the spatial analysis (Wong, 2009). As, the strength of a relationship between specific variables at one scale may vary or even disappear at another scale (Marceau, 1999). Therefore, the availability of data on a single geographic scale made it difficult to examine the correlation between the different variables at various scale, and thus there is not much to be done to deal with the modifiable areal unit problem in our analysis. d. Similarly, the categorization of population groups in official census data is defined by the traditions of data collection in each country. For example, income groups in Latin American cities are defined by the number of minimum wages earned by each group into; people earning up to half the minimum wage, people earning the minimum wage, people earning double the minimum wage and so on. In other cases such as in Australian or American cities, the groups are categorized based on their earning below or above fixed values into; people earning up to 10000 dollars, or people earning above 100000 dollars and so on. The problem with these given categories is that it undermines the

108 | Data and Method

analysis ability to sensitively trace the changes in groups’ distributions beyond the fixed categories. In more detail, the over-expanding literature on income inequality stresses on the fact that the income gap is growing due to the rise of the super-rich; Thomas Piketty and Emmanuel Saez (2003) - for example - have reached the conclusion that “the composition of income varies substantially by income level within the top decile [of the income distribution]. Therefore, it is critical to divide the top decile into smaller fractiles.” (p. 5). Therefore, the given categories of income in census data do not capture the variations among the rich and the super-rich, it simply combines people earning more than – for example – 100000 dollars a year in a single category, which affect the sensitivity of the test to the variations in income within the top decile of the income distribution.

e. Another difficulty faced the researcher is the language of census data, because not all cities have their data in English, some data forms are easily translated such as spreadsheets and word reports, while other pdf formats were difficult to be translated.

f. Finally, several cities have available census data but they were eliminated from the final dataset for one of two reasons: data is available on a single year, which means that change in segregation cannot be evaluated over a period of time, or divisions of areal units have changed between the two years of the available data, which also affect the value of the index, due to the modifiable areal unit problem and, made the comparison unreliable.

4.5.2 Defining the boundaries of the study area

Another challenge was to define the extent of the study area for each city, as the boundary chosen in each case (inner city, city with inner/outer suburbs, or metropolitan region) can have a profound effect on the degree of segregation observed (Boal, 2005). In more detail, confining the calculations

Data and method | 109 to inner-city neighborhoods can lead to a misperception of the actual situation of segregation in the urban area as a whole, as it eliminates a large and diverse portion of the population residing in the inner and outer suburbs. Similarly, calculating segregation on the metropolitan region level will incorporate rural areas that are disproportional in both size and population density into the analysis, also leading to miscalculations of the spatial segregation index2. For example, in order to determine the study area in the case of Melbourne, figure (4-2) shows a map of Greater Melbourne Region divided into Local Council Areas and highlighted into four zones: inner city, inner suburban, middle suburban, and outer suburban, while figure (4-3) shows the built-up area within the Greater Melbourne Region. A comparison of the two maps shows that the inner city is very small part of the built up area, while at the same time, the outer suburbs are mostly vacant areas with several scattered communities and town that are not a part of the urban core extending from the inner city to the middle suburban areas. Accordingly, in order to eliminate the miscalculations triggered by the low density of rural areas within the metropolitan regions, and at the same time, insure the integration of inner suburbs into the results of analysis, the calculations of segregation indices are performed over the divisions situated within the cohesive built-up area shown in figure (4-3). The same logic is applied to all other cases in the dataset (when possible), because defining the study area for cities in different contexts is mainly based on the available data and its type. For example, in the case of Melbourne and all other Australian cities, the study area is divided into Local Government Areas (LGA), where each LGA defined in the census by its location within the Greater Region (e.g. LGAs of the inner suburbs and LGAs of the outer suburbs). Based on their description, LGAs that are situated outside the built-up areas were excluded from the study area. While in the case of all American cities, the study area is divided into census tracts (CT), where the census provides a possibility to acquire data for CTs of the inner city, CTs of the metropolitan region, or CTs of the in-between level of

110 | Data and Method

frastructure,State Government of Victoria, Australia

areas suburban and inner-city between boundaries shows the map Region Melbourne 4 – Figure 2 Greater In Local and Planning of Transport, Department Source:

Figure 4-2 Greater Melbourne Region map shows the boundaries between inner-city and suburban areas

Data and method | 111

Region bourne

Source: Department of Transport, Planning and Local Infrastructure, State Government of Victoria, Australia of Victoria, Government State Infrastructure, Local and Planning of Transport, Department Source: Figure 4 – 3 The built-up area within the Greater Mel Greater the within area 4 – 3 The built-up Figure

Figure 4-3 The built-up area within the Greater Melbourne Region

112 | Data and Method the urbanized area, in this case the urbanized area is the boundary of the study area as it is equivalent to the built-up area in Australian cities. Defining the boundaries of other cases such as Brussels and the city-state of Singapore was different. In case of Brussels, the study area contains the 19 municipalities within the administratively defined Brussels-Capital Region. While in the case of Singapore, the study area is defined based on the information provided by the Urban Redevelopment Authority which classifies Singapore’s planning zones into populated and non-populated areas, in this case the non-populated areas are excluded from the analysis.

4.5.3 Problematic cross-city comparisons of the results

The next challenge was the interpretation of the results, as the actual value of the calculated index of segregation for any city was not comparable to the actual value of indices calculated in other cities for the following reasons. • Population groups in different census data are not unified, for instance, in one city the population groups are categorized according to their monthly income into nine categories, while in another city the groups are categorized according to their yearly earning into four categories (see Hamnett, 2003 for the differentiation between income and earning), accordingly, the result will be unreliable if value of segregation among nine population groups is compared to value of segregation among four population groups. • The different scales of the areal units within cities are not comparable, and the issue of scale can highly influence the results as the spatial segregation indices are scale dependent, if the areal units are smaller the indices tend to be with higher value and vice versa (Pendall, 2005; Boal, 2005). Therefore, an index for city divided into census tracts cannot be compared to an index for city divided into wards or municipalities.

Data and method | 113

In order to overcome the difficulties of inter-city comparison and to avoid standardization of data for all cities in the dataset, each city has two values of SD(m) in two different years, each city can be compared to itself over time to finally evaluate the percentage of change in the value of segregation within this particular city. After evaluating the change occurred in all cities over the years, the direction and magnitude of change in segregation level in all cities can be compared. In this case, issues of scale and unified data generated by inter-city comparisons are avoided, because the actual values of the indices are not used in the comparison. Instead, only the percentages of change in indices value over the same interval of time are compared to each other and to the position of cities in the global network as well. In view of the above, the primary dataset contains total of 178 cities, then due to data availability (and comparability), this number is reduced in the final dataset to a total of 91 cities. Of them, only 56 cities have data about both the socioeconomic status and ethnic background of the population, while the rest of cities have available data about either the socioeconomic status or the ethnic background of the population.

4.6 Final datasets

Tables (4-4) and (4-5) present the final datasets upon which the analytical study is conducted. In case of the evaluation of socioeconomic segregation levels, the analysis is performed on total of 66 global cities, categorized into 20 alpha cities, 23 beta cities, and 23 gamma cities. While in case of the evaluation of ethnic segregation in global cities, the analysis is performed over a total of 81 global cities, categorized into 29 alpha, 26 beta, and 26 gamma cities. Also, since the year of release of official census data differ from country to country, tables 4-4 and 4-5 show the census years on which the data for each city is collected. Noticeably, data regarding socioeconomic status of the population is available during the period3 from 1996 to 2010.

114 | Data and Method

And data regarding the ethnic composition of cities is available during the period from 1991 to 2012. The difference in release dates of the data generated other challenges for the analytical study. On one hand, comparing the changes in segregation level among different cities cannot be performed if the change in one city is calculated over the course of 9 year (as in the case of the American cities), while in other city the change is calculated over a course of 5 years (as in the case of the Australian cities). Therefore, the percentage of the average ‘yearly’ change in segregation is calculated for each city in the dataset to assure that the changes in segregation are compared within the same interval of time. On the other hand, the change in segregation in a city during the early 1990s is not comparable to the change in another city during the late 2000s. Therefore, according to table 4-4 and 4-5, cities with data in previous or later periods of time outside the average time range of all cities are excluded from the results. For example, cities of Lisbon and Porto are excluded from the ethnic segregation dataset, because their most recent data is in 2001, and their 2011 census was not officially released until the analysis took place (i.e. the available data for these two cities falls before the time span of data available for other cities in the dataset). Similarly, several cities, such as Tokyo, Brussels, Geneva, and Rotterdam, have available data on three different census years; the census year falling out of range of all other cities is also excluded.

Data and method | 115

final dataset for socioeconomic segregation analysis, by census year of the data

No. City Country

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Rank 1 2 ++ New York US * * 2 3 Hong Kong China * * 3 5 Singapore Singapore * * 4 8

alpha+ Chicago US * * 5 10 Sydney Australia * * 6 13 Toronto Canada * * 7 14 Sao Paulo Brazil * * 8 17 Los Angeles US * * 9 20 Mexico city Mexico * *

10 21 alpha Amsterdam Netherlands * * 11 25 Brussels Belgium * * * 12 27 San Francisco US * * 13 28 Washington US * * 14 29 Miami US * * 15 31 Melbourne Australia * * 16 36 Boston US * * 17 38 Dallas US * * 18 40 alpha- Atlanta US * * 19 46 Philadelphia US * * 20 47 Johannesburg South Africa * * 21 49 Stockholm Sweden * * 22 51 Montreal Canada * * * 23 55 Houston US * * 24 56 Berlin Germany * * 25 60 beta+ Copenhagen Denmark * * 26 62 Bogota Colombia * * 27 63 Vancouver Canada * * 28 68 Seattle US * * 29 71 Auckland N. Zealand * * * 30 72 Oslo Norway * * beta 31 79 Cape Town South Africa * * 32 82 Minneapolis US * * 33 86 Rio de Brazil * * 34 87 Brisbane Australia * * 35 90 Detroit US * * 36 91 Denver US * * 37 92 Monterrey Mexico * * beta- 38 100 St Louis US * * 39 102 Panama city Panama * * 40 103 San Diego US * * 41 105 Perth Australia * *

116 | Data and Method

Continue

final dataset for socioeconomic segregation analysis, by census year of the data

No. City Country 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Rank 42 107 Cleveland US * * 43 109 Calgary Canada * * 44 116 Cincinnati US * * 45 117 Charlotte US * * 46 121 Baltimore US * * 47 127 Adelaide Australia * * 48 129 gamma+ Portland US * * 49 133 San Jose us US * * 50 136 Kansas city US * * 51 137 Phoenix US * * 52 139 Guadalajara Mexico * * 53 147 Rotterdam Netherlands * * 54 148 Tampa US * *

55 149 gamma Columbus US * * 56 150 Indianapolis US * * 57 151 Pittsburgh US * * 58 152 Edmonton Canada * * 59 157 Orlando US * * 60 158 Gothenburg Sweden * * 61 160 Ottawa Canada * * 62 164 Richmond US * * 63 167 Durban S. Africa * * gamma- 64 168 Austin US * * 65 174 Milwaukee US * * 66 175 Wellington N. Zealand * *

final dataset for ethnic segregation analysis, by census year of the data

No. City Country 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Rank

1 1 London UK * * * ++ 2 2 New York US * *

3 3 Hong Kong China * *

4 4 Paris France * *

5 5 Singapore Singapore * *

6 6 Tokyo Japan * * * alpha+ 7 7 Shanghai China * *

8 8 Chicago US * *

9 10 Sydney Australia * *

Data and method | 117

Continue

final dataset for ethnic segregation analysis, by census year of the data

No. City Country 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Rank

10 11 Milan Italy * *

11 12 Beijing China * *

12 13 Toronto Canada * *

13 15 Madrid Spain * *

14 17 Los Angeles US * *

15 19 Frankfurt Germany * * alpha 16 21 Amsterdam Netherlands * *

17 24 Seoul S Korea * *

18 25 Brussels Belgium * * *

19 27 San Francisco US * *

20 28 Washington US * *

21 29 Miami US * *

22 31 Melbourne Australia * *

23 32 Zurich Switzerland * *

24 34 Munich Germany * *

25 36 Boston US * *

26 38 Dallas US * * alpha- 27 40 Atlanta US * *

28 41 Barcelona Spain * *

29 45 Lisbon Portugal * *

30 46 Philadelphia US * *

31 47 Johannesburg South Africa * *

32 48 Dusseldorf Germany * * *

33 49 Stockholm Sweden * *

34 51 Montreal Canada * * *

35 52 Rome Italy * *

36 53 Hamburg Germany * * * beta+ 37 55 Houston US * *

38 56 Berlin Germany * *

39 60 Copenhagen Denmark * *

40 63 Vancouver Canada * * 41 68 Seattle US * * 42 71 Auckland N. Zealand * * * 43 72 Oslo Norway * *

44 76 Beta Manchester UK * * 45 79 Cape Town S. Africa * * 46 82 Minneapolis US * * 47 87 Brisbane Australia * * 48 88 Geneva Switzerland * * * 49 90 beta- Detroit US * *

118 | Data and Method

Continue

final dataset for ethnic segregation analysis, by census year of the data

No. City Country 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Rank 50 91 Denver US * * 51 93 Bratislava Slovakia * * 52 97 Stuttgart Germany * * 53 99 Cologne Germany * * 54 100 beta- St Louis US * * 55 103 San Diego US * * 56 105 Perth Australia * * 57 107 Cleveland US * * 58 109 Calgary Canada * * * 59 116 Cincinnati US * * 60 117 Charlotte US * * 61 118 Antwerp Belgium * * * 62 121 Baltimore US * *

63 127 gamma+ Adelaide Australia * * 64 129 Portland US * * 65 133 San Jose us US * * 66 135 Valencia Spain * * 67 136 Kansas city US * * 68 137 Phoenix US * * 69 140 Lyon France * * 70 147 Rotterdam Netherlands * * * 71 148 Tampa US * * gamma 72 149 Columbus US * * 73 150 Indianapolis US * * 74 151 Pittsburgh US * * 75 152 Edmonton Canada * * 76 153 Tallinn Estonia * * 77 155 Porto Portugal * * 78 157 Orlando US * * 79 158 Gothenburg Sweden * * 80 159 Marseille France * * 81 160 Ottawa Canada * *

82 164 gamma- Richmond US * * 83 167 Durban S. Africa * * 84 168 Austin US * * 85 174 Milwaukee US * * 86 175 Wellington N. Zealand * * *

Data and method | 119

4.7 Interpretation of the results

One of the main objectives of this research is to evaluate the validity of several widely accepted assumptions in the global city literature. These assumptions include; firstly, global cities tend to witness intensification in spatial segregation level based on socioeconomic status, and ethnic background of population as a result of the new economic functions of cities in the global economy. Secondly, the intensity of spatial change is expected to be proportional to cities’ position in the global network of cities, as the more advanced producer service firms the city host, the more stark the changes in the city’s occupational structure, income structure, pace and intensity of gentrification, and eventually intensified socioeconomic and ethnic segregation. In order to put these assumptions to test, the study adopts several statistical hypothesis-testing approaches to determine whether to reject or accept each of the aforementioned assumptions. After calculating the change in ‘spatial multi-group dissimilarity index SD(m)’ over time for all global cities of the dataset, both the direction and the value of these changes are utilized to explore whether there is a correlation between cities’ global status and their change in socioeconomic and ethnic segregation or not. The interpretation of the results takes place over several stages: In the first stage, the study presents the general observations and attempts to find a statistically significant tendency among global cities to share similar directions of change towards intensified spatial segregation. In the second stage, the study examines the correlation between the magnitude of change in spatial segregation level in cities and their global network connectivity. The third stage aims to incorporate other variables to the analysis in a way that captures the influence of region, history, or global status on the resultant change in segregation level in global cities. Finally, stage four is dedicated to discuss in more detail the situation in several individual cities in an attempt to explain any unexpected results or

120 | Data and Method discrepancies appeared during the first three stages of the analysis. These stages are discussed in more detail as follows.

• Stage one: do global cities share patterns of intensified segregation?

In this stage, the researcher inspects – both visually and statistically - the common trends of change in segregation levels among global cities, through a series of tables and graphs that compare the value of the spatial segregation index calculated for each global city. (see also appendix I and II) It is worth mentioning that before proceeding with the following stages of the analysis; the standardized z-scores of the actual SD(m) values are calculated for all cities of the dataset. The aim of calculating the z-scores is to facilitate the comparison between different samples distributions (i.e. compare levels of change in socioeconomic segregation to those in ethnic segregation). Briefly, a z-score is an indication of how many standard deviations each sample deviates from the mean. For example, a z-score equal to 1 represents a sample that is 1 standard deviation greater than the mean; similarly, a z-score equal to -2, represents a sample that is 2 standard deviations less than the mean, and so on (Cohen et al., 2013). It can be calculated from the following: z = (X - μ) / σ where z is the z-score, X is the value of the element, μ is the population mean, and σ is the standard deviation.

• Stage two: examining the correlation between intensity of spatial change and cities’ global network connectivity.

In this stage, we aim to test the correlation between the position of a city in the global network and its corresponding change in spatial segregation. In order to examine this correlation, a bivariate linear regression analysis is performed to test the dependence of the ‘response’ (that is the change in spatial segregation level) on the ‘predictor’ (that is the global

Data and method | 121 network connectivity of the city) (Weisberg, 2014). Note that the response and the predictor are referred to in the text as the dependent and the independent variables respectively. Before going on with the linear regression analysis, the distribution of the observed data has to fulfill several assumptions. Including the absence of significant outliers (Freund et al., 2006), and a confirmation that the independent variable x and the dependent variable y have a linear relationship (Hoffmann, 2010). As explained the potential outliers are excluded from the analysis based on their calculated z-scores (cities scored more than 2 standard deviations from the sample mean are considered outliers). Also, in order to check whether a linear relationship exists between the two variables, a summary scatterplot is created to visually inspect the correlation between variables for linearity. According to Kahane (2008), when dealing with a large collection of data, the correlation between x and y may not be clear. Therefore, it is possible to group the data point on the x-axis (in our case the x-axis represents the cities’ global network connectivity, where cities can be grouped in alpha, beta, and gamma categories). And for each group, the mean value of y is calculated and represented on the y-axis (in our case the y-axis represents the standardized change in spatial segregation index). Given this graph, the dots represent the mean value of SD(m) values for each of the alpha, beta, and gamma categories. And by drawing the trendline, we can verify the following assumption: the mean values of y, for given values of x, are a linear function of x. y = a + bx, where x is the independent variable and y is the dependent variable. The slope of the line is b, and a is the intercept (the value of y when x = 0).

1221 | DataData andand MethodMethod

Ethnic segregation - means - means segregation Ethnic

Socioeconomic segregation - means means - segregation Socioeconomic gamma beta alpha gamma beta alpha

Figure 4-4 check for a linear correlation between the dependent and the independent variables

It should be noted that in case that the relationship between the mean values of y and x is nonlinear, then fitting a linear regression model to the data probably will not provide a useful model, and other alternative models are to be considered. Figure (4-4) confirms that the dependent and independent variables in our case have a linear relationship and therefore the linear regression model is fit for the data. The output of the bivariate regression analysis provides several indications on the strength of the association between the two variables, which can be used as tool for hypothesis testing: For example, the analysis provides the R and R2 values. The R value represents the simple correlation between the dependent and the independent variables, where the value of R ranges from –1 and +1, with positive numbers indicating a positive association and negative numbers indicating a negative association, also, a correlation of zero means there is no statistical association (Hoffmann, 2010). While the R2 value indicates how much of the total variation in the dependent variable – segregation index - can be explained by the independent variable – cities’ global status. And by obtaining these values, the aforementioned assertion is accepted or rejected depending on the strength of the association between the dependent and independent variables.

Data and method | 123

• Stage three – examining the influence of location, history, and other factors

In case that there are more than one independent variable that have influence on the variation in the dependent variables, a multiple linear regression is used to obtain the multiple correlations R (Weinberg and Abramowitz, 2002), and the relative contribution of each of the independent variables to the total variance explained. In our case, there are several variables that may affect the level of spatial segregation in global cities. For instance, cities located in Western Europe may show different trends in those in the US or Australia. Also, cities that recently joined the classification of world cities may show different trends than those appeared in previous classifications, as it can be argued that cities with longer history as global cities will show distinct spatial change proliferated over long period of time compared to other newly added global cities. Similarly, variables such as the different urban policies, the pace of gentrification, the condition of the public housing and so on, can be added to the model. However, not all the aforementioned variables are quantitatively measured. While the conversion of qualitative data into quantitative data is possible (Loehnert, 2010), still it is proved problematic to convert nominal non-scalar or non-ordinal data into numeric values without compromising the accuracy of the results. In more detail, a city’s status as a recent global city or a city with a longer history as a global city can be given the numerical values of 0 and 1, where 0 indicates that the city didn’t appear in global city classifications prior to 2010, and 1 indicates that the city already appeared in the 2008 classification. In this case the city’s global history is a dummy variable that can have an indicative numeric value without compromising the analysis results. Similarly, if the location of a city in a certain region can affect its level of segregation, and we are faced with 6 different regions (North America, South America, Western Europe, Australia, Southeast Asia, and North Africa/ Middle East), then giving each region a code from 1 to 6 is not a proper way for quantizing the given data.

124 | Data and Method

Because the values from 1 to 6 are only indicative if the given categories are scalar or in a certain order (such as the satisfaction scale in survey data which can range from 0: no satisfaction to 5: 100% satisfaction), yet in our case, in what order can the different regions be assigned to the different values on the scale from 1 to 6. In other words, which region deserves to take the numeric value of 6, which is a higher value than 1 and on what ground. In order to avoid the quantizing of nominal data, each region can be presented in the model as a separate dummy variable. If the city is located in the region then the variable takes the value of 1, if not, the variable takes the value of 0. The problem with this approach is the possible multicollinearity between two or more dummy variable, which leads to problems with understanding which variable contributes to the variance explained and technical issues in calculating a multiple regression model. Similarly, other variables such as urban policies and public housing are not dummy variables with 0 or 1 values; rather they incorporate higher level of complexity. The case of public housing for instance requires taking into account not only the presence or absence of public housing projects, but also other factors such as the allocation of the housing units, the rate of public housing privatization, the quality of housing units, the dominant ethnic groups in the public housing projects and so on. For that matter, in order to avoid the dummy variable trap and the miscalculations associated with the inaccurate quantizing of qualitative data. Another statistical approach is deployed to compare the influence of different locations, histories, and other variables on the resultant change in spatial segregation without the need to assign numerical values for each group of cities. Instead, the one-way analysis of variance ANOVA compares means that are drawn from different groups that are part of a larger sample. Using multiple comparison procedures, we may also determine if one of the means is significantly different from one of the others (Hoffmann, 2010). For example, the one-way ANOVA can give an indication if cities located in North America are significantly different than cities of Western Europe in

Data and method | 125 terms of changes in spatial segregation. Similarly, it can indicate whether alpha cities are behaving differently from beta and gamma cities. In order to use the one-way ANOVA in testing the hypotheses postulated in the literature, a p-value is calculated and compared to significance level of 0.05 (or 0.01) to determine whether to accept or reject the hypothesis under the test. For example, the hypothesis states that the rank of a city as alpha, beta, or gamma city will have a significant effect on the changes in spatial segregation within the city. In this case, we have three groups of cities (alpha, beta, and gamma), where each group contains 20 global cities. According to Kirkman, (1996), each group probably include different values of y and different sample mean. But can we take this difference in the average of y as evidence that the groups in fact are different (and perhaps that the rank causes that difference)? The likely range of variation of the averages if there is no significant difference between groups is given by the standard deviation of the estimated means: σ/N½ where σ is the standard deviation of all the values of y and N (is the number of cities in each group: in this case N=20). Thus if we treat the collection of the 3 group means as data and find the standard deviation of those means and it is "significantly" larger than the above, we have evidence that groups are significantly different. This is to say that if some (or several) group's average change in segregation is "unusually" large or small, it is unlikely to be just "chance" (ibid.). The comparison between the actual variation of the group averages and that expected from the above formula is expressed in terms of the F ratio:

F=(found variation of the group averages)/(expected variation of the group averages)

After obtaining the F ratio, and by using the degree of freedom (in our case, it is the number of groups - 1, and the number of cities in each group - 1: (2,9)); a value x is obtained from the ‘table of critical values for the F

126 | Data and Method distribution’, which indicates how large the F needs to be for the groups to be significantly different. In other words, if the calculated F is greater than x, then the groups are significantly different and vice versa. The same principle can be applied on cities located in different regions or cities with other shared characteristics.

• Stage four – the individuality of cities: a discussion Both the regression analysis and the one-way ANOVA are expected to give robust indications on how the cities’ global status and their spatial changes may or may not be associated. Still, the detailed discussion of individual cases is required for several reasons: Firstly, the discussion of the outliers, which is useful in explaining why these particular cities have witnessed above (or below) average change in their spatial segregation level. The discussion aims to uncover the variables – other than global network connectivity – that contributed to the observed spatial change. Secondly, based on the results of the one-way ANOVA, any significant influence of one of the examined factors on the resultant segregation, e.g. if the analysis revealed that the location of a city in a certain geographic region is an influential factor, then a detailed discussion of cities of the region can assist in reaching a systemic explanation for the observed phenomenon. Also, discussion of individual cases aims to address any inconsistencies within the results. An example for such inconsistency is the case of two cities that share several characteristics including the global status or the national contexts, yet both cities show inexplicable variations in their spatial changes. In this case, the discussion of the situations in the two cities can point out certain contextual particularities to be as influential on the process of spatial restructuring as other macro developments propagated in the global city literature. The subsequent chapters present the findings of the analytical study. Chapter five deals with the change in socioeconomic segregation in global

Data and method | 127 cities. Chapter six provide the same analysis for the change in ethnic segregation. And in chapter seven, the study attempts to investigate the correlation between the change in both socioeconomic and ethnic segregation within the same global city.

128 | Data and Method

Notes

1 http://www.lboro.ac.uk/gawc/gawcworlds.html 2 Since the calculations of the spatial index incorporate both area and length of shared boundaries among division, then large rural tracts will have long shared boundaries (implying a high possibility of interaction between population groups in different divisions), which can falsely decrease the value of segregation index. While at the same time, these areas have very low population density compared to neighborhoods in inner city and suburbs, then excluding these areas from the analysis can give a better perception of the situation in the urban area with no miscalculations added by incorporating the adjacent rural areas within the metropolitan region. 3 Earlier data for the 1970s and 80s are not available in digital form on the official census bureau websites and could not be obtained by the researcher.

Chapter 5

5 Socioeconomic segregation in global cities: Analysis and findings

This chapter presents the findings and general observations that links the level of change in socioeconomic segregation in any city over time, to its position as a global city based on the GaWC classification of world cities. The collective results show if there is an association between the agglomeration of producer services firms in one city (the city’s service value) and the intensified socioeconomic segregation within this city. Also, individual cases are discussed to evaluate the influence of the contextual particularities of each city on its corresponding change in spatial segregation. As noted in the preceding chapter, the final dataset contains a total of 66 cities categorized into 20 alpha, 23 beta, and 23 gamma cities. This section presents the preliminary results and the general observations drawn from the calculations of the ‘spatial multi-group dissimilarity index SD(m)’ over time for the 66 global cities of the dataset, followed by the interpretation of the results, which takes place over the three stages of the statistical hypothesis testing discussed in chapter four. Figure (5-1) is a graphical representation of the average yearly change in SD(m) value of all cities of the dataset (see appendix i for more details about the calculated SD(m) values for all cities of the dataset). As a general observation, the figure shows that the three global ranks (alpha – beta – gamma) contain cities with both increase and decrease in levels of

130 | Analysis and findings socioeconomic segregation. Also, cities do not only differ in the direction of change towards more or less segregation, but also the intensity of the change varies greatly across cities of the dataset. In total, 36 cities (54.5% of the dataset) have witnessed an average increase in socioeconomic segregation over the past years, compared to 30 cities (45.5%) with decrease in socioeconomic segregation. The values range from the maximum increase in segregation, that is 3.0% of average yearly change in SD(m) value scored by the gamma city of Guadalajara (the city scored total of 30.3% increase in socioeconomic segregation from 2000 to 2010), to the other end of the spectrum, the maximum decrease in segregation, that is -3.7% of average yearly change in SD(m) value scored by the alpha city of Brussels (a total of -26.4% of change from 2001 to 2008). As a general observation from figure (5-1), there is no statistically significant tendency among global cities to share an increase in socioeconomic segregation levels (see appendix ii for more details regarding the significance probability). Similarly, there is no obvious correlation between the spatial change in one city and its position as an alpha, beta, or gamma city. Therefore, these results are further explored as follows:

5.1 Examining the correlation between the intensity of spatial change and cities’ global network connectivity.

This section presents the results of the linear regression analysis introduced in chapter four. The calculations are generated using IBM – SPSS statistics software, and the aim of the analysis is to test the strength of the association between the cities rank in the global network of cities and their corresponding change in socioeconomic segregation. The expected result - if the null hypothesis is correct – is a strong correlation between the two variables indicated by a value of R that is greater than 0.5. Also, the

Socioeconomic segregation in global cities | 131

New York Hong Kong Singapour Chicago Sydney Toronto Sao Paulo Los Angeles Mexico city Amsterdam Brussels San Fransisco alpha Washington Miami Melbourne Boston Dallas Atlanta Philadelphia Johannesburg Stockholm Montreal Houston Berlin Copenhagen Bogota Vancouver Seattle Auckland Oslo Cape Town Minneapolis Rio de Janeiro beta Brisbane Detroit Denver Monterrey St Louis Panama city San Diego Perth Cleveland Calgary Cincinnati Charlotte Baltimore Adelaide Portland San Jose us Kansas city Phoenix Guadalajara Rotterdam Tampa Columbus indianapolis Pittsburgh gamma Edmonton Orlando Gothenburg Ottawa Richmond Durban Austin Milwakee Wellington

4.00 3.00 2.00 1.00 0.00 -1.00 -2.00 -3.00 -4.00

FigureFi 5-15 1 alphalh – betabt – gamma citiesiti and d their th i yearly l percentage t of f change h in i socioeconomic i i segregation ti index segregation socioeconomic in change of percentage yearly their and cities – gamma – beta 5-1 alpha Figure

132 | Analysis and findings








y = -3E-06x + 0.281 R² = 0.00197









Standardized change in SD(m) index value 

 







Global Network Connectivity

Figure 5-2 Summary scatterplot for the correlation between the standardized change in segregation index and the cities’ global network connectivity summary scatterplot in figure (5-2) represents the correlation between the dependent variable, which is the standardized change in the spatial dissimilarity index SDI on the y-axis, and the independent variable, which is the cities’ global network connectivity GNC on the x-axis. The SPSS generates several tables for the linear regression. In this section, we show the three main tables required to understand the results of the linear regression procedure: the model summary table, the ANOVA table, and the coefficients table.

Socioeconomic segregation in global cities | 133

Table 5-1 Regression output – model summary, ANOVA, and Coefficient tables

Model Summary Adjusted R Std. Error of the Model R R Square Square Estimate 1 .044a .002 -.014 1.00679900 a. Predictors: (Constant), AbsoluteGNC

ANOVAa Sum of Model Squares df Mean Square F Sig. 1 Regression .127 1 .127 .125 .725b

Residual 64.873 64 1.014 Total 65.000 65 a. Dependent Variable: zscoreINCOME b. Predictors: (Constant), AbsoluteGNC

Coefficientsa Unstandardized Standardized Coefficients Coefficients

Model B Std. Error Beta t Sig. 1 (Constant) .109 .333 .328 .744

AbsoluteGNC .000 .000 -.044 -.354 .725 a. Dependent Variable: zscoreINCOME

Firstly, the regression output table (5-3) provides the information needed to determine how well the regression model fits the data. In the model summary section, the column ‘R’ is the absolute value of the Pearson correlation coefficient between the dependent and the independent variables.

134 | Analysis and findings

It simply indicates the strength of the association between the two variables. In our case, R = 0.044 < 0.5, which indicates a weak correlation. The R2 value in the R2 column represents the proportion of variance in the dependent variable that can be explained by the independent variable. In our case, R2 = 0.002, which means that the independent variable, global network connectivity, explains only 0.2% of the variability of the dependent variable, change in segregation index. Please note that the R2 is calculated for the sample (66 global cities), SPSS generates another value called ‘adjusted R2 that represents the proportion of variance if the analysis is performed over the entire population (i.e. the full list of global cities). Table (5-3) shows that the adjusted R2 = -0.014, which indicates a very low dependence between variables. The next table is the ANOVA table, which reports how well the regression equation fits the data (i.e., how accurately the model can predicts the dependent variable SDI based on the changes in the independent variable GNC). The number in the "Regression" row and the "Sig." column indicates the statistical significance of the regression model that was run. Here, the regression model is statistically insignificant, F (1, 64) = 0.125, p = 0.725, which is greater than 0.05, and indicates that, overall, the regression model can not statistically predicts the outcome variable, which conform with the low dependence between variables observed from the model summary table. The last table is the coefficients table, which provides the coefficients of the regression equation that is used to predict the dependent variable from the independent variable. y = b0 + (b1 x) where y is the change in segregation, x is global network connectivity, b0 is the intercept and b1 is the coefficient, both can be found under the B column in the table, b0 = 0.109, and b1 = 0.000. In our case, it is already established that the model poorly fits the data due to the weak dependence between the variables. Recapitulating, the linear regression established that global network connectivity for a city could not statistically significantly predict the change

Socioeconomic segregation in global cities | 135 in spatial segregation level, as the two variables have a weak correlation. The global network connectivity of cities is accounted for only 0.2% of the explained variability in spatial segregation index.

5.2 Examining the influence of location, history, and other factors

So far, the analysis confirmed that cities’ position in the global network of cities is not associated with the direction or the intensity of change in socioeconomic segregation within these cities, which also suggests that there are other independent variables that can explain the variation in the dependent variable. This section aims to assess the influence of several factors that are assumed to be relevant to the outcomes of spatial restructuring in global cities. As explained in chapter four, the influence of these factors is evaluated by the one-way analysis of variance ANOVA. For example, the sample of cities – upon which the regression findings are drawn – contains two distinguishable groups of cities; cities that were already ranked as global cities in the GaWC classification prior to 20101, and cities that are recently added to the classification in 2010. This distinction is assumed to be relevant to the explanation of the variation in spatial segregation, as the longer the period of time the city is considered a global city, the more visible the spatial outcomes of restructuring are expected to be. One can argue that this factor (referred to as global history during data entry phase) is causing the regression results to be inaccurate as the sample is not homogeneous. Therefore, to examine the influence of a city’s global history on the resultant change in spatial segregation, the sample is divided into two groups: cities newly added to 2010 classification, and cities in the classification since 2008. Table (5- 4) reports the output of the one-way ANOVA generated by IBM- SPSS.

136 | Analysis and findings

Table 5-2 ANOVA output when cities are grouped according to their global history ANOVA Zscore: INCOME Sum of Mean Squares df Square F Sig. Between Groups .082 1 .082 .169 .683 Within Groups 28.246 58 .487 Total 28.328 59

ONEWAY zscore SD(m) BY global history

The table provides the F ratio, as explained in chapter four, the calculated value of F can be compared with x that is the critical value below which there is no statistically significantly difference between groups’ means. In other words, when F

Socioeconomic segregation in global cities | 137 respective regions but they are ranked below alpha cities due to their less significant ‘service value’. In this sense, if the global/divided city model is valid, then alpha cities are expected to show higher tendency for intensified socioeconomic segregation than beta and gamma cities2. Therefore, the sample is once again divided into three groups; table (5-5) reports the output of the one-way ANOVA when cities are categorized in alpha, beta, and gamma groups.

Table 5-3 ANOVA output when cities are grouped according to their global rank ANOVA Zscore: INCOME Sum of Mean Squares df Square F Sig. Between Groups .181 2 .090 .183 .833 Within Groups 28.147 57 .494 Total 28.328 59 ONEWAY zscore SD(m) BY global rank

The calculated F ratio from the table equals 0.183 which is once again less than the critical value of x = 3.16. Accordingly, it can be concluded that changes in segregation witnessed by alpha cities is not statistically significantly different from those witnessed by beta or gamma cities. The p- value from the table also confirm this conclusion, where p= 0.833, which is greater than a significance level of 0.05. Another factor to be tested in this section is the location of a global city in a certain geographic region. The cities of the sample are divided according to their broad geographic region into 6 groups: Western Europe, North America, Latin America, Australia, Asia, and Africa. Table (5-6 – above) presents the ANOVA output, which reveals that – unlike the previous factors of global history and rank – the F ratio equals 3.025, which is greater than x= 2.32. In this case, the analysis results suggest that cities located in

138 | Analysis and findings certain region do share trends of change in spatial segregation, and these trends are significantly different from cities in other regions. This conclusion is confirmed by the p-value = 0.018 which is less than 0.05.

Table 5-4 ANOVA output when cities are grouped according to their geographic region ANOVA Zscore: INCOME Sum of Mean Squares df Square F Sig. Between Groups 6.198 5 1.240 3.025 .018 Within Groups 22.130 54 .410 Total 28.328 59 ONEWAY zscore SD(m) BY region

ANOVA Zscore: INCOME Sum of Mean Squares df Square F Sig. Between Groups 5.849 3 1.950 4.634 .006 Within Groups 21.454 51 .421 Total 27.302 54 ONEWAY zscore SD(m) BY region except Asia and Africa

In order to enhance the accuracy of the analysis and to avoid any misread conclusions, table (5-6 - below) shows a re-run of the ANOVA test when cities of Asia and Africa are excluded from the sample. Groups are excluded due to the small number of cities in each, Asia contained only 2 cities and Africa contained 3 South African cities. Accordingly the analysis is performed again among 55 cities grouped into four regions: Western Europe, North America, Latin America, and Australia. Noticeably, the F ratio = 4.63 is still greater than x = 2.79, and the p-value has improved to

Socioeconomic segregation in global cities | 139 reach 0.006, which indicates a high significance level of the influence of the geographic region on the changes of segregation in global cities. Similarly, cities can be grouped according to other characteristics to ultimately determine the influence of these characteristics on the outcomes of spatial restructuring in global cities. For example, cities can be divided into capital cities and non-capital city, which is an indication of the concentration of political power in the city and the associated change in segregation. Other factors concerning the pace of gentrification, the status of public housing, and other urban policies in general can be examined. Yet, since the collection of this data proved problematic (for instance, there is no system for ranking cities according to their gentrification level), while other factors inhibit a high level of complexity that problematize the quantizing of qualitative data (as in the case of the public housing – see chapter four for details), then the analysis will go on to stage 4 and start the discussion of the situation in individual global cities in the light of the regression and ANOVA findings.

5.3 The individuality of cities: a discussion

Table 5-7 presents the top 10 cities with increased socioeconomic segregation and the top 10 cities with decreased socioeconomic segregation. An observation drawn from table (5-7) may explain why the global/divided city model is widely accepted despite the lack of extensive empirical evidence supporting it. In the top 10 list of cities with increased segregation, the 1st, 2nd, and 5th places are filled by all three Mexican cities in the global ranking. Oddly, Mexican cities are sharing similar spatial transformations with Swedish cities that fill the 6th and 9th places on the same list. The oddness here is derived from the huge variations between the two countries in terms of quality of life (see Prescott-Allen, 2001), welfare system, and different ranking categories and their levels of integration into global economy are not the same. While, an in-depth discussion of the situation in

140 | Analysis and findings the two contexts will reveal that they might share similar changes in socioeconomic segregation, yet the changes are a result of entirely different reasons.

Top ten cities with maximum increase and decrease in socioeconomic segregation and their respective ranks and regions Top 10 cities with increased segregation City Rank Region 1 Guadalajara gamma L. America 2 Monterrey beta L. America 3 Amsterdam alpha Europe 4 Denver beta N. America 5 Mexico city alpha L. America 6 Gothenburg gamma Europe 7 Oslo beta Europe 8 Sydney alpha Australia 9 Stockholm beta Europe 10 Seattle beta N. America

Top 10 cities with decreased segregation City Rank Region 1 Brussels alpha Europe 2 Rotterdam gamma Europe 3 Berlin beta Europe 4 Rio de Janeiro beta L. America 5 Austin gamma N. America 6 San Francisco alpha N. America 7 Copenhagen beta Europe 8 Atlanta alpha N. America 9 Toronto alpha N. America 10 Johannesburg alpha Africa

Socioeconomic segregation in global cities | 141 demographic changes. And still, both Mexican and Swedish cities are heading toward more segregated configuration of cities. Nevertheless, similarities between the two different contexts cannot be pinned on economic globalization, because all five cities are falling into different global categories. Overall, the absence of a collective pattern of spatial change in global cities proves that socio-spatial transformations in these cities cannot be fully understood away from their local circumstances and contexts. Therefore, the subsequent sections are dedicated to discuss in more detail the motivations and factors involved in the production of diverse socio-spatial transformations in individual global cities. And according to the results of the ANOVA, the sections are organized based on the broad geographic regions of the world, which was already proven to have a significant influence on the outcomes of spatial restructuring in global cities. For easier reading of the results presented in figures (5-1), figure (5-3) presents global cities categorized according to their location in the broad geographic regions of the world instead of their global rank. Clearly, cities in a certain region are not expected to show similar changes in their socioeconomic segregation level, due to the huge variations exist among cities within the same region in terms of national policies, history, geography, subcultural differences… etc. Yet, the observations drawn from figure (5-3) suggest that: On one hand, although the changes in income inequality and socio- spatial polarization in European cities are recognized to be ‘moderate’ when compared to the change in American cities over the last 20 years (Van Kempen and Murie, 2009), still, the results in figure (5-3) shows a similarity between the two contexts, where only 50% of global cities in both regions scored an increase in their socioeconomic segregation level, while the higher tendency towards intensified socioeconomic segregation is observed among global cities of Latin America and Australia, where 71.4% of cities of each

142 | Analysis and findings

Brussels

Rotterdam e Berlin p Copenhagen Stockholm Oslo n index index n Gothenburg Euro W. Amsterdam Austin San Fransisco Atlanta Toronto Calgary Montreal Miami Chicago Columbus Phoenix San Diego Washington St Louis Los Angeles Minneapolis indianapolis Vancouver Charlotte Houston Dallas

Edmonton lo America

Baltimore g San Jose us Philadelphia An Cleveland Kansas city Tampa New York Pittsburgh Milwakee Boston Orlando Cincinnati Portland Richmond Detroit Ottawa Seattle Denver Rio de Janeiro Bogota Panama city Sao Paulo Mexico city Monterrey Guadalajara L. America Singapour

Hong Kong Asia Adelaide Wellington Brisbane Perth Auckland Melbourne Oceania Sydney Johannesburg Cape Town

Durban Africa

4.00 3.00 2.00 1.00 0.00 -1.00 -2.00 -3.00 -4.00 Figure 5-3 global cities categorized according to their geographical region and their yearly change in socioeconomic segregatio socioeconomic in change yearly their and region geographical to their according categorized cities 5-3 global Figure Figure 5-3 global cities categorized according to their geographical region and their yearly change in socioeconomic segregation index

Socioeconomic segregation in global cities | 143 of the two regions showed an increase in their SD(m) value (note that Africa and Southeast Asia are underrepresented in this analysis due to lack of data). On the other hand, the results also show that even among cities within the same national border, similar patterns of change may not exist. For example, Brazilian cities of Sao Paulo and Rio de Janeiro show a stark contradiction in both intensity and direction of change in socioeconomic segregation, same for Johannesburg and Cape Town, Rotterdam and Amsterdam, and Austin and Denver. These cases are discussed in more detail as follows:

5.3.1 Latin American cities

The academic literature on urban development of Latin American cities underlines distinct characteristics of the region’s urban areas. On one hand, the unprecedented rapid urbanization in the second half of the twentieth century, where roughly, the share of the population living in cities grew from 40% in 1950 to more than 75% in 2000 (Lattes et al., 2002). As a result, mass numbers of rural migrants – in their search for accommodation in cities of destination - have built informal settlements at the periphery of cities (Rolnik, 2011; Monkkonen, 2011). Which led the urban structure of Latin American cities to be characterized by a center/periphery dichotomy, where high-income families are concentrated in wedge-shape areas, pointed at the historical center and expanded in a single geographical direction towards the periphery. These central and affluent urban wedges are embedded in a peripheral zone of poor and working-class families, living mostly in informal settlement (Telles, 1995; Sabatini et al., 2001; Sabatini, 2003; Feitosa et al., 2007; Audirac et al., 2012). On the other hand, uneven industrialization and growing social inequality have always been of the basic characteristics of the region (Telles, 1995). Unlike Western Europe where deindustrialization has been accompanied by a shift to services and spatial shift of manufacturing to countries of the global South such as Brazil and Mexico. Latin American

144 | Analysis and findings global cities such as São Paulo and Guadalajara faced a complex process of economic restructuring. Where metropolitan cores are deindustrializing, while at the same time, urban peripheries and hinterland are expanding and receiving off-shore manufacturing industries from Western Europe and North America (Audirac et al., 2012). Later in 2000s, industrial peripheries lost manufacturing competitiveness to Asia. Arguably, the global processes of economic, social, and spatial restructuring have impacted the longstanding history of socio-spatial inequality in Latin American cities (Telles, 1995; Audirac et al., 2012). One example is the emergence of “gated communities” for medium and high- income families within the urban periphery including poor neighborhoods (Feitosa et al., 2007; Monkkonen, 2011). The ‘islands of wealth’ as Janoschka (2002) calls them are gradually replacing the previous large-scale segregation patterns of poor peripheral areas and wealthy central cities by fragmented and more complicated urban structure (Monkkonen, 2011). In fact, Gated communities are a form of gentrification as they result from increasing demand for urban space to accommodate high-income groups, and the low land costs in poor peripheries (ibid.). Yet the difference between gentrified areas in Latin American cities and those in Western Europe and North America is the ‘walling out’3, where the gentrified areas are totally isolated from its surrounding that are still dominated by housing for low- income groups, therefore, real estate developers favored the exclusive, closed, and controlled housing project out of fear of crime and violence in the surrounding neighborhoods (Caldeira, 2000; Coy and Pohler, 2002). In general, the distinct characteristics of Latin American cities explain the historically similar patterns of socioeconomic segregation shared by cities across the region. As shown in figure (5-3), Latin American cities may still be sharing patterns of change in segregation in the past decade, where 71.4% of cities of the region scored an increase in income segregation (except for Rio de Janeiro and Bogota). The recent and collective increase in segregation is usually explained in the specialized literature by the ongoing

Socioeconomic segregation in global cities | 145 urbanization along with relocation of manufacturing industries from cities’ peripheries to Asia (Audirac et al., 2012). This relocation is accompanied by plant closures, job losses, and eventually skyrocketing unemployment rates. For example, the main industrial center of Guadalajara (Mexico’s Silicon Valley in the 1990s) has lost more than 45,000 jobs in electronics industry to China between 2001 and 2003 (Dussel, 2005). To sum up, the influx of rural – mostly poor - immigrants, increasing unemployment, and the rise of gated communities contributed to the intensification of socioeconomic segregation in a number of global Latin American cities. However, this assumption may explain why all three Mexican cities are showing sharp increase in their segregation level, but it fails to explain the inconsistency in spatial change between Sao Paulo and Rio de Janeiro. Moreover, a closer look to the situation in Mexican cities reveals that intensified segregation is generated by variables other than the historical urban development of the region. In particular, Mexico’s urban development is marked by distinct government actions, which contributed to the sharp increase in segregation in Mexico City, Guadalajara, and Monterrey. According to Monkkonen (2011), Mexico’s new federal housing finance system - initiated in the 1990s - is a government intervention that has contributed to greater levels of segregation. As, it triggered the construction of large tracts of middle-class single-family houses in the poor peripheries of cities across the country, by providing mortgages exclusively to registered salaried employees in the formal sector. Roughly, 220,000 mortgages were issued in Mexico during the 1990s, this figure jumped to 400,000 per year between 2000 and 2005 (ibid.). And since loans have been given only to registered employees, thousands of poor informal workers were excluded from the new suburban development. While at the same time, local governments were unable to effectively manage the growth of informal settlements of poor families (Audirac et al., 2012). The drawbacks of housing reform in Mexico are the creation of homogeneous large areas for residents of a similar socioeconomic status, along with the expansion of informal settlements for

146 | Analysis and findings the poor in cities’ peripheries, leading to rapid increase in socioeconomic segregation levels. The results in figure (5-3) supports Monkkonen criticism of the federal housing reform in Mexico. However, the assumption that poor population tends to expand in peripheral informal settlements could not be verified. Figure (5-4) and (5-5) show the distribution of population earning up to the minimum wage or less in Mexico City municipalities in 2000 and 2010. Unfortunately, census data does not differentiate between informal sector workers and registered employees. Nevertheless, the 2010 map shows that low- paid workers are not expanding in peripheral municipalities. Instead, poor residents were displaced and concentrated in higher densities in less number of municipalities in the center. Also, unlike ethnic segregation, socioeconomic segregation is not entirely dependent on residential mobility; part of the changes in socioeconomic compositions of municipalities is due to the changes in employment status caused by lay-offs. Therefore, the recent significant presence of poor workers in municipalities of the southeast and northeast is not necessarily due to the displacement of the poor. Instead, it can be a result of plants closure and the consequence income decline in those neighborhoods. Similar tendencies showed in Guadalajara and Monterrey, regardless the different global status of each city. Unlike the unified trend in Mexican cities, Brazilian cities showed different outcomes. Sao Paulo and Rio de Janeiro varied in both intensity and direction of change in income segregation, as shown in figure (5-3), Sao Paulo scored 4.97% increase in segregation between 2000 and 2010, while Rio de Janeiro scored 24.76% decrease in segregation over the same period of time. The noticeable difference between the two cities is explained by the uneven industrialization in previous decades and different urban development approaches in recent years. Firstly, uneven industrialization is a common characteristic of urban development in Latin America in general, and Brazil is one of the best examples of this imbalance, where metropolitan areas range from highly

Socioeconomic segregation in global cities | 147

N

10 mi

10 km Figure 5-4 distribution of workers earning the minimum wage or less in Mexico City municipalities in 2000

N

10 mi

10 km

Figure 5-5 distribution of workers earning the minimum wage or less in Mexico City municipalities in 2010

Figure 5-4 distribution of workers earning the minimum wage or less in Mexico City municipalities in 2000 Figure 5-5 distribution of workers earning the minimum wage or less in Mexico City municipalities in 2010

148 | Analysis and findings industrialized ones like Sao Paulo to others “that have grown without the benefit of industrialization” (Tells, 1995: 1200) and are consequently have higher poverty and unemployment rates. Since “Rio had never been able to attract the consumer goods sectors which led to the dynamic industrialization and strong local markets of Sao Paulo, which overtook Rio de Janeiro as Brazil’s largest city in the 1950s. Rather, the economy of Rio had been dependent on the local tertiary sector, comprised of small modern segment (finance, computers, commerce) and a large personal services sector.” (Ribeiro and Telles, 2000: 80). Therefore, Rio de Janeiro has always had higher levels of segregation between the poor and the middle class than in Sao Paulo with a clear north and south division (Tells, 1995) (see figure 5-7). While at the same time, the impact of deindustrialization in the 1980s and 90s was more severe on Sao Paulo, it followed the Mexican scenario where manufacturing industries left first to the hinterlands then to Asia in later decades (Audirac et al., 2012). Accordingly, it can be argued that uneven industrialization in previous decades has impacted the level of vulnerability of cities toward the impacts of current economic transformations on both global and local levels of urban development. Secondly, urban governance in Brazil on federal, regional, and municipal levels contributed to the production of different spatial outcomes in cities. Decentralization of the administration of urban services started in the 1960s, materialized in the form of regional offices created to manage urban growth (Levy, 2001). Then, in 1988, the new constitution gave municipal government more political, administrative, financial, and legislative independence, and made local governments responsible for all services including land use regulations (ibid.). Also, the constitution reflects the rising public awareness of cities governance, as the constitution contains a chapter dedicated to urban policies, revolving around three concepts; the social function of cities, the importance of direct participation of citizens in urban policy decision-making processes, and recognition of the land rights

Socioeconomic segregation in global cities | 149 of millions of inhabitants of informal settlements on the periphery of cities in Brazil (Rolnik, 2011). The multi-level decentralized urban governance in Brazil resulted in several urban development initiatives, on both federal and local level. Such as, Favela-Bairro project by the local government of Rio de Janeiro in 1994 (UN-Habitat, 2003), public-private partnerships projects in Sao Paulo’s master plan in early 1990s (Siqueira, 2012), and the housing program ‘minha casa, minha vida’ initiated by the state in 2008 (Barbosa, 2010). The different approaches adopted by each initiative have led to different spatial outcomes in cities across the country. As shown in figures (5-6) and (5-7), the north-south division in Rio de Janeiro in 2000 appeared to fade in 2010 despite Telles’s (1995) prediction of the continuation and increase of socioeconomic segregation in globalizing Rio. Feitosa et al. (2007) and Monkkonen (2011) explain the decrease in segregation in Rio and other Brazilian cities by the changing scale of segregation. In more detail, when a gated community appears in a poor peripheral area, the segregation level on the municipality level decline because the municipality hosts different income groups, but on the micro level, wealth is still unevenly distributed in pockets of self-segregation. Arguably, the case of Rio de Janeiro cannot be explained by this scale issue, since it depends essentially on the emergence of affluent gated communities in the poor north. Figure (5-7) shows that poor population over 10 years of age with monthly income less and up to half the minimum wage started to concentrate in the affluent south, causing income segregation levels to fall significantly. Changes in spatial distribution of the poor can be partially explained by their recent access to middle-class housing unites facilitated by the local government. The case of Sao Paulo differs significantly from Rio de Janeiro. As shown in figures (5-8) and (5-9), the concentration of poor population with monthly income less and up to half the minimum wage in 2010 increased in

150 | Analysis and findings

N

10 mi

10 km Figure 5-6 distribution of population with monthly income less and up to half the minimum wage in Rio de Janeiro in 2000

N

10 mi

10 km Figure 5-7 distribution of population with monthly income less and up to half the minimum wage in Rio de Janeiro in 2010

Figure 5-6 distribution of population with monthly income less and up to half the minimum wage in Rio de Janeiro in 2000 Figure 5-7 distribution of population with monthly income less and up to half the minimum wage in Rio de Janeiro in 2010

Socioeconomic segregation in global cities | 151

N 10 mi 10 km 9 distribution of population with monthly - Figure 5 income less and up to half the minimum wage in Sao Paulo in 2010 N 10 mi 10 km f 8 distribution of population with monthly - Figure 5 income less and up to half the minimum wage in Sao Paulo in 2000

Figure 5-8 distribution of population with monthly income less and up to half the minimum wage in Sao Paulo in 2000 Figure 5-9 distribution of population with monthly income less and up to half the minimum wage in Sao Paulo in 2010

152 | Analysis and findings neighborhoods that were already mostly poor in 2000, causing higher level of income segregation. Unlike Rio de Janeiro, the poor population did not gain access to central affluent neighborhoods. Which matches Siqueira’s (2012) description of the duality of the urban structure of Sao Paulo, where large investments are dedicated for development of strategic locations in order to “construct[ing] a skyline that now resembles other global cities around the world.” (p: 391), while poor areas lack basic services. The variations among Sao Paulo, Rio de Janeiro, and Mexican cities reflect the role of the local governments in promoting higher levels of integration or segregation among different socioeconomic groups. Latin American cities in general share historical characteristics, where their shared spatial structure is derived from colonial and pre-colonial times (Telles, 1995), they also share similar impacts of industrialization and premature deindustrialization (Audirac et al., 2012). Yet clearly, the ability of central and local governments in managing urban development is an important contingency that affects the spatial outcomes of Latin American cities under globalization.

5.3.2 Australian cities

Figure (5-3) indicates that Australian cities show parallel changes in socioeconomic segregation to that of Latin American cities, where 71.4% of Australian cities witnessed an increase in socioeconomic segregation. Generally, patterns of spatial segregation in Australian cities are discussed expansively in the academic literature. Theoretically, factors contributing to the intensified spatial segregation in Australian cities are gentrification, increased international migration, and the consequent changes in housing market including home ownership rates, housing densities, and rent costs. The common transformations in Australian cities (Baum, 2008) are basically due to similar characteristics shared by the large urban areas across the country throughout their relatively short histories (Forster, 2006). They were all highly suburbanized with high levels of home ownership. Their

Socioeconomic segregation in global cities | 153 social structure was reflected spatially in ‘doughnut’ shaped cities, where the declined inner cities are circled by growing suburban development (ibid.). Since the 1970s, it is argued that the socioeconomic segregation has worsened generally in Australian cities (Hunter, 2003), as the post-Fordist economic and social restructuring resulted in two contrasting outcomes; the increasing suburbanization of poverty and significant gentrification in inner cities (Badcock, 1997; Forster, 2006; Baum, 2008). Accordingly, the demand for housing units has undergone noticeable changes since the early 1990s. On one hand, the concentration of advanced producer services – especially in Sydney and Melbourne – predominately in the CBD and few adjacent inner suburbs has generated unprecedented demand for high-quality housing close to the CBD to serve the affluent highly-skilled professionals (Stimson, 2001; Baum, 2008). On the other hand, shifts in immigration policies have supported large-scale in-migration of international students and skilled workers (Weller and Van Hulten, 2012). Which have raised the demand for cheap residence, mostly in the outer suburbs (Stimson, 2001). The supply side has been affected by the planning policies adopted in the late 1980s, which aimed to restrain suburban expansion by promoting the construction of new dwellings within the existing built-up areas of cities (Forster, 2006). As a result, residential densities have increased and medium density housing such as apartments, townhouses, and duplexes started to gradually replace the prevalent single-family suburban houses (ibid.). Another incentive for the changes in housing supply is the decline of public housing sector in Australian cities since 1991 (Badcock, 1999), where the federal government decreased funding to public housing authorities, causing shrinkage in new public housing built in recent years, along with the demolition of deteriorated public housing built in the 1950s and 1960s to be replaced by privately owned dwellings (Arthurson, 1998). Consequently, the restricted suburban expansion, the changing residential landscape, and the declining public housing caused the growth rate of housing stock to fall

154 | Analysis and findings gradually since the early 1990s to reach the lowest rate in the last 50 years in 2009 and 2010 (Kulish et al., 2011). In what seems to be a typical example for the global/divided city model, the limited housing supply and increased demand have led to increased competition over limited residential space, which is responsible for housing prices to rise significantly in absolute terms or relative to household incomes (Forster, 2006; Kulish et al., 2011) leading to differential access to the housing market based on households’ ability to pay. Moreover, with the traditionally high rates of residential mobility in Australia (Badcock, 1997), differential access has contributed to the increased ‘neighborhood sorting’ where people with similar preferences (and the ability to pay) tend to live with others with similar characteristics (see chapter two) (Hunter, 2003). While at the same time, differential access limits the opportunities for low-income households to find affordable housing, and confine them to low-quality dwellings in private rental sector (Randolph and Holloway, 2005; Forster, 2006). Ultimately leading to a further segregation between the affluent and the disadvantaged. Figures (5- 10) and (5-11) show the distribution of poor population (with weekly income less than 160$) over Melbourne urbanized area in 2001 and 2006. Clearly, the inner-city poor in 2001 have decreased significantly in 2006, which conform to the gentrification process described in the literature. In fact, Inner Melbourne has witnessed the most extensive gentrification process across Australia (Weller and Van Hulten, 2012), according to Raskall (1995); the extent of the inner-city gentrification in Melbourne exceeded even that observed in the inner-west of Sydney. So far, Australian cities appear to be textbook cases of how globalization can affect the local housing market and generate visible changes on neighborhood level. Where, retreating policy actions and deregulated housing market have compromised cities’ resistance to gentrification and poverty concentration, leading eventually to a higher level of intensified socioeconomic segregation.

Socioeconomic segregation in global cities | 155

N 10 mi 10 km 11 Distribution of population with income less 11 - Figure 5 than 160$ per week in urbanized Melbourne 2006 N 10 mi 10 km 10 Distribution of population with income less - Figure 5 than 160$ per week in urbanized Melbourne 2001

FigureFi 5-105 10 distributiondi ib i of f population l i with ihi income less l than h 16$ per week ki in urbanized b i dM Melbournelb iin 2001 Figure 5-11 distribution of population with income less than 16$ per week in urbanized Melbourne in 2006

156 | Analysis and findings

However, as shown in other cases, government intervention is an influential contingency that can either lead to intensified socioeconomic segregation (as the case of Mexico City) or promote higher levels of integration (as the case of Rio de Janeiro). The Australian city of Adelaide is also an example of such influence of local government’s decision on the outcomes of spatial restructuring on neighborhoods level. Originally, Adelaide had a reputation to be economically disadvantaged compared to other Australian cities, where some areas of Adelaide are among the nation’s poorest suburbs (Baum, 2008). Then, in late 1990s, local government focused on issues of inequality and responded by several urban regeneration projects such as ‘The Parks neighborhood renewal’ in Adelaide’s west (Forster, 2006) and ‘Salisbury North urban improvement project’ in Adelaide’s north (Arthurson, 2012). Those projects aimed to improve housing and infrastructure of deteriorated neighborhoods, as well as implementing community development programs that reduce welfare dependence and promote social inclusion of poor residents (Forster, 2006; Arthurson, 2012). As a result, despite the economic restructuring process shared by Australian cities, Adelaide’s relatively poor outer suburbs do not show the same stark contrast to the inner city witnessed in Melbourne or Sydney. Which once again highlights the influential role of local governments on the outcomes of economic restructuring in individual cities.

5.3.3 African cities

As noted, African cities are underrepresented in the final dataset due to data availability, and the only three African cities in the dataset are all situated in South Africa. Yet the discussion here is still central to the understanding of the role of local contingencies in the complex process of cities restructuring. The case of Johannesburg shows an alpha city that does not follow the global/divided city model (since the city scored a decrease in socioeconomic segregation by 0.7% by year), while the explanation of its

Socioeconomic segregation in global cities | 157 spatial transformation is strongly linked to historical and social factors that existed on the local level in previous decades. In more detail, South African social classes have always been defined according to color line; “[wealth] was largely in the hands of white South Africans” (Beall, 2002: 47). After the first non-racial, democratic elections in 1994, which coincided with the rapid growth of service sector and changes in employment patterns, race and class association began to erode, and new black middle class started to grow to support the service industries that were predominated by whites (ibid.). For that matter, the decrease in socioeconomic segregation is explained by changes in the labor market that are caused, not only by economic changes in the global context, but also caused by changes in racial segregation in post-apartheid Johannesburg. Moreover, Beall et al. (2000) acknowledge the role of urban governance in decreasing segregation, through reaching a balance between achieving global competitiveness and tackling spatial segregation by executing programs of black empowerment and poverty reduction. Despite the fact that Cape Town and Durban share the same national, political, and historical context of Johannesburg, both cities showed different direction of spatial transformation than their fellow city. The reason behind this inconsistency is individual situation of each city. Where unlike Johannesburg, ‘shack-dwellers' movement in both Cape Town and Durban have resisted the forced eviction of blacks from slum areas in the center of these cities to peripheral townships (Huchzermeyer, 2011), which caused both racial and socioeconomic segregation to be harder to tackle. Accordingly, South African cities highlight the importance of recognizing cities as individual cases. Where generalized characteristics of cities on global or regional levels are eventually reshaped by distinct situation in each city.

158 | Analysis and findings

5.3.4 North American cities

In total, 51.2% of cities of the Anglo American region have witnessed an increase in socioeconomic segregation levels since the late 1990s, yet regional differences appear between cities in Canada and USA, and even on sub-country level especially within the USA. In case of the Canadian cities, only 33% of Canadian cities showed an increase in socioeconomic segregation level, Fong and Shibuya (2000) claim that socioeconomic segregation in these cities is influenced by racial and ethnic segregation rather than the socioeconomic status of the segregated poor. Where racial and ethnic groups have differential access to housing choices. Therefore, governmental actions that promote ethnic integration eventually reduce the associated socioeconomic segregation. Which explains the tendency for Canadian cities to be less segregated by income. This argument is discussed in more detail in the subsequent chapters. In case of American cities, issues of socioeconomic segregation have a longer history than those in Canada. Moreover, in spite of the strong historical association between socioeconomic status and race in the US (Massey and Denton, 1993), race alone cannot explain the increased poverty concentration in inner cities (Wilson, 1987), especially when considering the macro economic changes impacting western societies such as deindustrialization. Income inequalities in the US have emerged as a result of the uneven economic development on the national level since the late 1940s (Brenner, 2002). Originally, the 1950s were the golden age of manufacturing industries in central cities of the Northeast and Midwest, then the increasingly footloose industries shifted slowly - during the 1950s to the 1990s – to the Southeast then the Southwest/West then finally to Asia in the 1990s to present (Ross, 2011). With the massive impacts of deindustrialization on cities of the Northeast and Midwest, they were labeled as Rustbelt cities, a term that summarizes the process of fiscal crisis and the

Socioeconomic segregation in global cities | 159 consequent joblessness, tax revenue loss, and concentrated poverty (ibid.). As a result, the decline of Rustbelt cities accelerated the shift in population and employment to the South into the fast growing Sunbelt cities (Chapple and Lester, 2010). The North and South division in the USA was intensified in the 1980s due to neoliberal federal policies in Reagan’s era, where “state and municipalities began to adopt entrepreneurial strategies in order to attract external capital investment to their territorial jurisdictions” (Brenner, 2002: 8) including infrastructure investments and locational decisions of industries based on the supply of cheaper labor (ibid.). (See chapter six – Sunbelt cities receive sizable in- migration surge from Central and South America who are mostly low-pay workers). Apparently, as shown in figure (5-12), the socio-spatial impacts of the historically uneven development between Rustbelt cities and Sunbelt cities - since 1940s to 1980s - are not yet reversed. The map shows the 33 American cities, of which 14 cities are Sunbelt cities, the rings indicates the value and direction of change in socioeconomic segregation from 2000 to 2009. In the North, 14 out of 19 (73.7%) cities have witnessed an increase in socioeconomic segregation, compared to only 4 out of 14 (28.6%) cities in the Sunbelt. Brenner (2002) explains the continuation of the north-south division in the 21st century to be a result of the ineffective reform of federal policies under Bush and Clinton administrations, both modified the pure neoliberal policies of the preceding decades. However, urban policies are still market- based which sustained the regional competitiveness over investment. In addition, the growth of Sunbelt cities in terms of territorial size, population and economic output have widened the gap between the north and the south. According to Chapple and Lester (2010), the growth of the south is associated with the booming information technology economy of the 1990s - for example, average earnings per worker in Austin increased by 42% due to the fast growing high-tech industries in Austin metro area. Therefore, the changes in socioeconomic segregation showed in figure (5-3) conform to

160 | Analysis and findings

Figure 5-12 increase/decrease in socioeconomic segregation in both Sunbelt and Rustbelt cities in the USA USA the in cities Rustbelt and Sunbelt both in segregation socioeconomic in increase/decrease 5-12 Figure Figure 5-12 increase/decrease in socioeconomic segregation in both Sunbelt and Rustbelt cities in the USA

Socioeconomic segregation in global cities | 161

Chapple and Lester interpretation of economic development of Sunbelt cities. The increasing average earnings per worker in Austin is reflected spatially, where Austin scored the highest degree of socioeconomic integration (18.4% decrease in segregation from 2000 to 2009 which equals -2% of yearly change). While, the formerly leading industrialized cities of Detroit and Denver are amongst the top cities with increased socioeconomic segregation. Arguably, neoliberal globalization has clear impacts on American cities, exhibited in deindustrialization, increased competitiveness, and uneven development of the north and the south. However, the generalized association between globalization and socio-spatial divide does not apply to American cities as well as North American cities in general. Where, first, only half of North American cities had a significant increase in socioeconomic segregation. Second, inconsistencies in cities’ direction and value of spatial changes suggest that spatial transformations need to be explained on sub-regional and local level of individual cities. For example, the north-south division of winner and loser cities cannot explain the spatial transformations in Chicago and Dallas; both are alpha cities where command and control centers of global economy are located (Sassen, 1991), but Chicago scored less segregation despite its location in the Midwest (the losers region), while Dallas scored more segregation despite its location in the South (the winners region). Accordingly, finer grain analysis is required to comprehend the contingent factors that influence spatial transformations in individual American cities.

5.3.5 Western European cities

As shown in chapter two, several qualities have distinguished the socio-spatial transformations of Western European cities from those of American cities or any other region in general. According to Van Kempen and Murie (2009), ‘the corporatist and social democratic welfare states in Europe’ has contributed to the mitigation of the inequalities associated with

162 | Analysis and findings industrial restructuring, through strong employment protection and benefit systems. Accordingly, unemployment rates and loss of income are relatively low compared to those produced under the liberal welfare state of the USA. However, in terms of quantity, figure (5-3) shows that patterns of spatial changes in Europe are not so different than those in USA. 50% of European cities of the dataset have witnessed a significant increase in income segregation over the past decade, compared to 51.2% in North America. Also, the intensity of spatial changes appear to be higher in European cities in either directions toward more segregation or integration. The region hosts the top three cities – of the entire dataset - with less socioeconomic segregation, Brussels, Berlin, and Copenhagen, along with Amsterdam which scored the highest increase in socioeconomic segregation after Guadalajara and Monterrey. On the national level, Swedish cities of Stockholm and Gothenburg showed a consistent and strong tendency for increased socioeconomic segregation, while patterns of segregation in Amsterdam and Rotterdam are inconsistent in terms of direction and intensity of the change. Noticeably, the results do not indicate that absolute levels of spatial segregation in Europe are higher than in USA. Instead, results only refer to the pace and intensity of spatial changes over a certain period of time. The academic literature on spatial segregation in Western European cities focuses on urban policies and public housing as key aspects in defining residential segregation patterns in European cities (Musterd and Andersson, 2005; Van Kempen and Murie, 2009). Since low-income households are (in most cases) overrepresented in public housing dwellings. Then, the spatial concentration of public housing in certain areas of the city can lead to the spatial concentration of low-income households (Van Kempen and Murie, 2009). Therefore, the difference in location, quantity, and quality of public housing from city to city can explain the variations in segregation levels among European cities. However, the dynamic process of change in socioeconomic segregation patterns requires the preexisting

Socioeconomic segregation in global cities | 163 density and location of public housing to be altered, through either introducing additional public housing projects, privatization of existing ones (e.g. Musterd and Fullaondo, 2008), or demolition of old projects as part of the renewal and diversification projects adopted by several European states (e.g. Kleinhans, 2004; Musterd, 2005). Evidently, urban policies vary significantly among Western European countries (Van kempen and Murie, 2009). The next section briefly discusses housing policies and its consequent change in segregation patterns in three states, the Netherlands, Sweden, and Belgium. In the Netherlands, The national Urban Renewal Policy - since 1997 – aimed to reduce spatial concentration of low-income households and ethnic minorities to finally promote social mix in residential areas (Kleinhans, 2004; Bolt et al., 2009). The renewal processes seemed to follow a typical route of gentrification, as neighborhood restructuring resulted in the demolition of old social rented dwellings and the construction of owner- occupied expensive dwellings (ibid.). Unfortunately, that supposedly successful strategy for improving social cohesion has resulted in a “zero-sum outcome” (Musterd and Andersson, 2005: 766). As, diversification of social structure by introducing higher income households to the target area is accompanied by the displacement of low-income households who used to reside in the demolished dwellings. According to Bolt et al. (2009: 515), low-income household “tend to move to neighborhoods with a similar population composition as the areas they are leaving behind. In other words; maybe concentrations of low-income households are broken in the targeted areas, but there is a strong suggestion that new concentrations emerge elsewhere.” Causing the average segregation on city level to increase or - at least – remain the same. This theorization of residential restructuring in Dutch cities may explain the increased socioeconomic segregation in Amsterdam along with other factors discussed in chapter six. However, Rotterdam has a different story, where diversification projects are accompanied by strict housing

164 | Analysis and findings allocation rules set by local government. Those rules were defined in 2003 by the ‘Rotterdam zet door’ action program (Kleinhans, 2004), which aimed to control the spatial distribution of low-income households and prevent their re-concentration in disadvantaged neighborhoods. Apparently, Rotterdam plans succeeded in achieving social mix – at least statistically, as table 5-1 shows that Rotterdam is in the second place among the top 10 cities with decreased socioeconomic segregation. Again, the cases of Amsterdam and Rotterdam are examples of how existing local situations and governmental initiatives can alter or even reverse spatial outcomes. In Sweden, the country has always had a reputation for having a well- developed welfare system, progressive housing policy, and good quality housing (Musterd and Andersson, 2005; Andersson 2007). In 1965, the government initiated ‘The Million Homes’ program to overcome housing shortage by constructing one million new dwellings in ten years (Bråmå, 2008). In later decades – as in most western societies – neoliberal housing policies were adopted by the state. As a result, the housing sector in Sweden became deregulated and market-oriented (Hedin et al., 2012), and existing public rental dwellings are increasingly being privatized. Moreover, public housing companies became more selective in their choice of tenants, and the very poor were excluded from affordable housing (Sahlin, 2008), as a result, between 1999 and 2005, homelessness rates has doubled (Hedin et al., 2012). The overview on public housing in Sweden gives a broad idea on residential shifts expected to take place in both Stockholm and Gothenburg. Also, the absence of a counter action from both central and local governments – as in the case of Rotterdam - suggests that privatization of public housing and potential displacement of low-income households will eventually lead to higher level of concentration of disadvantaged population, which conform to the increase in segregation index in both cities shown in the results. Furthermore, growing inequalities in Sweden is perceived to be drastic, according to Musterd (2005), Swedish society is known to be more

Socioeconomic segregation in global cities | 165 egalitarian than any other society in Europe. Therefore, the recent inequalities “have led to a relatively strong labeling and stigmatization reaction in the Swedish context.” (Musterd, 2005:344). Following the same logic, the intense increase in socioeconomic segregation shown in figure 5-3 is not simply a result of the neoliberal reform of welfare system, but also because the Swedish welfare system – at least until the 1990s - was considered to be ideal and comprehensive. Thus, socio-spatial consequences of neoliberal reform are magnified, especially with the absence of governmental actions that can resist, modify, or reverse the increasingly segregated spatial outcomes. In fact, the beta and gamma cities of Stockholm and Gothenburg appear to be ideal examples of the global/divided city model discussed in the literature. In Belgium, the alpha city of Brussels offers totally different model from those discussed in the Netherlands and Sweden. Where the role of the state in housing provision has never been very large (Kesteloot and Cortie 1998), and when compared to other European cities, Belgian cities in general have relatively small shares of social housing (Kesteloot et al. 1997; Musterd, 2005). Accordingly, the changes in residential patterns in Brussels are not occurring due to demolition or privatization of public housing dwellings as in Amsterdam or Stockholm. Instead, among other factors, the expansion of the European Union institutions in Brussels as a capital of Europe has led to increased demand for additional office space and other supporting services, which eventually led to gentrification (Kesteloot, 2000; Swyngedouw et al., 2002). However, despite the fact that gentrification and its associated displacement theoretically contribute to higher levels of spatial segregation (see chapter three), still, Brussels managed to come in the first place among the top 10 cities with highest level of spatial change towards more integration. This spatial outcome is mainly due to two distinct characteristics that mark the process of gentrification in Brussels. Firstly, Kesteloot (2000) describes gentrification in Brussels to be “modest” (p: 203), compared to level of ongoing restructuring of the city.

166 | Analysis and findings

Also, the gentrification is contained in certain areas such as Leopold Quarter (Swyngedouw et al., 2002) and in the eastern edge of the 19th-century belt in general (Kesteloot, 2000), where EU institutions are located. In fact, residential blocks in the area are systematically bought by property developers and eventually demolished and replaced with offices. Yet, there is no evidence that the displaced households are re-concentrating in other neighborhoods. Secondly, according to Van Criekingen and Decroly (2003), the comparative analytical study they performed over Brussels and Montreal shows that scale (size) of gentrification projects in Brussels are small, as they put it “smaller than census tract” (p: 2466). Accordingly, the decreased segregation on city level can be explained by the scale of gentrification, where municipalities that host a small scale gentrification project will have higher social mix, which cause the average level of segregation on the municipalities level to drop significantly. Nevertheless, higher levels of income segregation may appear on smaller scale such as the neighborhood4 level, where due to ‘small scale gentrification’, adjacent blocks in the same municipality can show clear contrast in their income structure and housing quality, still, this assumption requires another detailed analytical study to be confirmed.

Socioeconomic segregation in global cities | 167

Summary

In the light of the global city literature, global cities are expected to have an increasingly fragmented socio-spatial structure due to economic restructuring and its associated growing inequality. However, the analytical study presented in this chapter concludes that generalized patterns of spatial transformations on the global level are not detected; the 66 cities of the dataset have shown diverse directions of change in segregation level, where only 54.5% of global cities have witnessed an increase in socioeconomic segregation, which challenges the validity of the generalized assumptions in the literature. This conclusion is also supported by a regression analysis, which confirmed that the global network connectivity of cities explains only 0.2% of the variability in spatial segregation index. Moreover, the one-way analysis of variance ANOVA revealed that cities’ position in the global network represented in alpha, beta, and gamma categorization are not associated with certain changes in socioeconomic segregation. While the location of global cities in certain geographic region is proved influential on the resultant levels of segregation Cases discussed in this chapter show that local situations in individual cities such as housing regulations, the existence/condition of public housing, and gentrification levels define the changes in the residential canvas of each city. And since local situations on cities level vary greatly within the same region and sometime within the same nation-state (as in cases of Johannesburg - Cape Town and Amsterdam - Rotterdam), then, generalized patterns of spatial changes in global cities are to be questioned. In fact, the impacts of globalization-related phenomena (such as gentrification and privatization of public housing) on the spatial transformation within cities are not uniform. Gentrification in Brussels and Melbourne has led to two distinct spatial outcomes due to the scale and intensity of the gentrification process. Similarly, the state provision of public housing in Mexico has contributed to higher levels of socioeconomic

168 | Analysis and findings segregation, while at he same time, the privatization of public housing in Sweden has also led to more segregated outcome. Another example for local situations that define spatial outcome is the cases of Adelaide, Rio de Janeiro, and Rotterdam. The three cities are situated in different global, regional, and sub-regional contexts. Yet, local government of each city could reach higher levels of integration through deploying different strategies such as urban development projects in Adelaide and Rio, and housing allocation regulations in Rotterdam. As a result, the three cities showed a significant decrease in socioeconomic segregation. The individuality in cities also explain the significant variations in direction and intensity of spatial change in American cities, where the factor of race is included in defining patterns of socioeconomic segregation along with other issues of affordability and access to the housing market. Race and ethnicity also show relevance to socioeconomic segregation in other contexts such as South Africa and Canada. The next chapters shed more light on patterns of ethnic segregation and the intersection of ethnicity and socioeconomic segregation in global cities.

Socioeconomic segregation in global cities | 169

Notes

1 2010 classification is compared only to 2008 classification due to the different methodology used by the GaWC in previous classifications, because the network connectivity became the main measure of importance of cities only from 2008 onwards (Taylor et al., 2011), 2 The more advanced producer service firms the city host the more stark the changes in the city’s occupational structure, income structure, pace and intensity of gentrification, and eventually intensified socioeconomic segregation. 3 By definition, ‘walling out’ is the extreme physical form of voluntary and deliberate separation of a socially and economically dominant group. Walling out may be involved in the formation of ‘exclusionary enclave’ and is also involved in the formation of a ‘citadel’ (Marcuse, 2005). 4 The 19 municipalities of Brussels are divided into 724 neighborhoods (quartiers), 606 of them have more than 200 inhabitants (Brussels Instituut voor STATISTIEK en Analyse, 2005). Statistical data of income groups on neighborhood level are not available before 2008 (only median income is available for every year since 1993). For that matter, the analysis is performed on the larger divisions of municipalities due to the availability of comparable data.

170 | Analysis and findings

Chapter 6

6 Ethnic segregation in global cities: Analysis and findings

As noted in chapter four, type of data collected by each census vary greatly from country to country. Yet, since ethnic composition within cities is related to issues of migration flows and labor mobility, which are accelerated in the age of globalization. Therefore, data collected for different cities meant to reflect the impacts of those factors, such as1, population by citizenship status (foreigners): in countries of Spain, France, and Switzerland, population by nationality by region: in countries of Belgium and the Netherlands, and foreign born population by region: in USA, Canada, and Australia. Noticeably, developing countries that are sending migration to other parts of the world have no data related to in-migration flows to global cities in these regions2, and hence are excluded from the final dataset. Accordingly, the final list contains 81 cities, to be analyzed within the period from 1995 to 2012, divided into 29 alpha, 26 beta, and 26 gamma cities. In the light of the discussion in chapter two and three, levels of ethnic segregation are expected to increase in global cities due to one of two reasons. On one hand, the increased demand for both skilled professionals and low-skilled migrants - with a relatively easier labor mobility - lead to a higher complexity of the ethnic composition of global cities, this complexity eventually contributes to higher level of spatial segregation, especially when

172 | Analysis and findings considering the preferences and behaviors of individual new-comers to the city, as most migrants follow ‘beaten paths’ and go where their fellow nationals have already established a community, making it easier to find work and accommodations (Castles, 2002). On the other hand, regardless of the in-migration flows to the city, levels of segregation among historically coexisting ethnic groups can still rise due to the socioeconomic status of ethnic groups and its related issues of access to education, labor market, and eventually, housing opportunities. This argument is discussed in more detail in chapter seven. Despite the expected increase in ethnic segregation levels in global cities, as shown in figure (6-1), global cities of the dataset showed the same lack of shared patterns of change in ethnic segregation as in the case of socioeconomic segregation. Where 45 cities (55.6% of the dataset) show a tendency for decreased levels of ethnic segregation compared to only 36 cities (44.4% of the dataset) with an increase in ethnic segregation levels. As a general observation, The maximum increase in ethnic segregation is scored by the alpha city-state of Singapore, with a 4.7% of average yearly change in SD(m) value (47% increase in ethnic segregation from 2000 to 2010), followed by the alpha city of Sydney with 2.5% by year. On the other end of the spectrum, the maximum decrease in ethnic segregation is also scored by an alpha city, that is Milan with a -5.8% of average yearly change in SD(m) value (-52.27% decrease in segregation between 2001 and 2010), followed by the beta city of Manchester with -4.5%. See subsequent sections for the in-depth interpretation of the observation.

6.1 Examining the correlation between the intensity of spatial change and cities’ global network connectivity.

This section presents the results of the linear regression analysis introduced in chapter four. The calculations are generated using IBM - SPSS

Ethnic segregation in global cities | 173

London New York Hong Kong Paris Singapore Tokyo Shanghai Chicago Sydney Milan Beijing Toronto Madrid Los Angeles ha

Frankfurt p

Amsterdam al Seoul Brussels San Fransisco Washington Miami Melbourne Zurich Munich Boston Dallas Atlanta Barcelona Philadelphia Dusseldorf Stockholm Montreal Rome Hamburg Houston Berlin Copenhagen Vancouver Seattle Auckland Oslo Manchester

Minneapolis beta Brisbane Geneva Detroit Denver Bratislava Stuttgart Cologne St Louis San Diego Perth Cleveland Calgary Cincinnati Charlotte Antwerp Baltimore Adelaide Portland San Jose us Valencia Kansas city Phoenix Lyon Rotterdam Tampa Columbus amma

indianapolis g Pittsburgh Edmonton Tallin Orlando Gothenburg Marseille Ottawa Richmond Austin Milwakee Wellington 6 5 4 3 2 1 0
-1
-2 -3 -4 -5 -6 -7 Figure 6-1 alpha – beta – gamma cities and their yearly percentage of change in ethnic segregation index index segregation ethnic in of change percentage yearly their and cities – gamma – beta 6-1 alpha Figure Figure 6-1 alpha – beta – gamma cities and their yearly percentage of change in ethnic segregation index

174 | Analysis and findings statistics software, and the aim of the analysis is to test the strength of the association between the cities rank in the global network of cities and their corresponding change in ethnic segregation. The expected result - if the null hypothesis is correct – is a strong correlation between the two variables indicated by the value of R that is greater than 0.5. Also, the summary scatterplot in figure (6-2) represents the correlation between the dependent variable, which is the standardized change in the spatial dissimilarity index SDI on the y-axis, and the independent variable, which is the cities’ global network connectivity GNC on the x-axis.







  

 


 






Standardized change in SD(m) index value









Global Network Connectivity

Figure 6-2 Summary scatterplot for the correlation between the standardized change in segregation index and the cities’ global network connectivity

Ethnic segregation in global cities | 175

The SPSS also generates several tables of for the linear regression. In this section, we show the three main tables required to understand the results of the of the linear regression procedure: the model summary table, the ANOVA table, and the coefficients table.

Table 6-1 Regression output – model summary, ANOVA, and Coefficient tables Model Summary Adjusted Std. Error of the Model R R Square R Square Estimate 1 .048a .002 -.010 1.00516005675 a. Predictors: (Constant), AbsoluteGNC

ANOVAa Sum of Mean Model Squares df Square F Sig. 1 Regression .183 1 .183 .181 .672b Residual 79.817 79 1.010 Total 80.000 80 a. Dependent Variable: zscoreETHNICITY b. Predictors: (Constant), AbsoluteGNC

Coefficientsa Unstandardized Standardized Coefficients Coefficients

Model B Std. Error Beta t Sig. 1 (Constant) .112 .287 .392 .696

AbsoluteGNC .000 .000 -.048 -.425 .672 a. Dependent Variable: zscoreETHNICITY

176 | Analysis and findings

Firstly, the regression output table (6-2) provides the information needed to determine how well the regression model fits the data. In the model summary section, the column ‘R’ is the absolute value of the Pearson correlation coefficient between the dependent and the independent variables. It simply indicates the strength of the association between the two variables. In our case, R = 0.048 < 0.5, which indicates a weak correlation. The R2 value in the R2 column represents the proportion of variance in the dependent variable that can be explained by the independent variable. In our case, R2 = 0.002, which means that the independent variable, global network connectivity, explains only 0.2% of the variability of the dependent variable, change in segregation index. Please note that the R2 is calculated for the sample (81 global cities). SPSS generates another value called adjusted R2 that represent the proportion of variance if the analysis is performed over the entire population (i.e. the full list of global cities). Table (6-2) shows that the adjusted R2 = -0.010, which indicates a very low dependence between variables. The next table is the ANOVA table, which reports how well the regression equation fits the data (i.e., how accurately the model can predicts the dependent variable SDI based on the changes in the independent variable GNC). The number in the "Regression" row and the "Sig." column indicates the statistical significance of the regression model that was run. Here, the regression model is statistically insignificant, F (1, 79) = 0.181, p = 0.672, which is greater than 0.05, and indicates that, overall, the regression model cannot statistically predict the outcome variable, which conform with the low dependence between variables observed from the model summary table. The last table is the coefficients table, which provides the coefficients of the regression equation that is used to predict the dependent variable from the independent variable. y = b0 + (b1 x) where y is the change in

Ethnic segregation in global cities | 177 segregation, x is global network connectivity, b0 is the intercept and b1 is the coefficient, both can be found under the B column in the table, b0 = 0.112, and b1 = 0.000. In our case, it is already established that the model poorly fits the data due to the weak dependence between the variables. Recapitulating, the linear regression established that global network connectivity for a city could not statistically significantly predict the change in ethnic segregation level, as the two variables have a weak correlation. The global network connectivity of cities accounted for only 0.2% of the explained variability in spatial segregation index.

6.2 Examining the influence of location, history, and other factors

Parallel to the findings in the case of socioeconomic segregation, the analysis confirmed that cities’ position in the global network of cities is not associated with the direction or the intensity of change in ethnic segregation within these cities, which also suggests that there are other independent variables that can explain the variation in the dependent variable. This section aims to assess the influence of several factors that are assumed to be relevant to the outcomes of spatial restructuring in global cities. As explained in chapter four, the influence of these factors is evaluated by the one-way analysis of variance ANOVA. Firstly, to examine the influence of a city’s global history on the resultant change in spatial segregation, the sample is divided into two groups: cities newly added to 2010 classification, and cities in the classification since 2008. Table (6-3) reports the output of the one-way ANOVA generated by IBM- SPSS. The table provides the F ratio, as explained in chapter four, the calculated value of F can be compared with x that is the critical value below which there is no statistically significantly difference between groups’ means. In other words, when F

178 | Analysis and findings under test has no significant influence on the observed variation in segregation level and vice versa. In our case, F=2.017 < x=3.97 which

Table 6-2 ANOVA output when cities are grouped according to their global history

ANOVA Zscore: ETHNICITY Sum of Mean Squares df Square F Sig. Between Groups 1.106 1 1.106 2.017 .160 Within Groups 41.114 75 .548 Total 42.220 76 ONEWAY Zscores SD(m) BY global history indicates insignificant difference between cities that are already global cities for a longer period of time than cities that are recently added to the classification. This conclusion is also confirmed with the p-value under the column ‘Sig.’ in table, where p=0.16, which is greater than a significance level of 0.05. The p-value here is another indication on the absence of statistically significantly difference in groups’ means.

Table 6-3 ANOVA output when cities are grouped according to their global rank ANOVA Zscore: ETHNICITY Sum of Mean Squares df Square F Sig. Between Groups 1.074 2 .537 .966 .386 Within Groups 41.146 74 .556 Total 42.220 76 ONEWAY Zscores SD(m) BY global rank

Similarly, the sample is once again divided into three groups; table (6- 5) reports the output of the one-way ANOVA when cities are categorized in alpha, beta, and gamma groups. The calculated F ratio from the table equals

Ethnic segregation in global cities | 179

0.966 which is once again less than the critical value of x = 3.12. Accordingly, it can be concluded that changes in ethnic segregation witnessed by alpha cities is not statistically significantly different from those witnessed by beta or gamma cities. The p-value from the table also confirms this conclusion, where p= 0.386, which is greater than a significance level of 0.05. Table 6-4 ANOVA output when cities are grouped according to their geographic region ANOVA Zscore: ETHNICITY Sum of Mean Squares df Square F Sig. Between Groups 6.034 4 1.508 3.001 .024 Within Groups 36.186 72 .503 Total 42.220 76 ONEWAY Zscores SD(m) BY region

ANOVA Zscore: ETHNICITY Sum of Mean Squares df Square F Sig. Between Groups 5.682 3 1.894 3.792 .014 Within Groups 35.464 71 .499 Total 41.146 74 ONEWAY Zscores SD(m) BY region except Eastern Europe

Another factor to be tested in this section is the location of a global city in a certain geographic region. The cities of the sample are divided according to their broad geographic region into 5 groups: Western Europe, Eastern Europe, North America, Australia, and Asia. Table (6-5 – above) presents the ANOVA output, which reveals that – unlike the previous factors of global history and rank – the F ratio equals 3.001, which is greater than

180 | Analysis and findings x= 2.5. In this case, the analysis results suggest that cities located in certain region do share trends of change in spatial segregation, and these trends are significantly different from cities in other regions. This conclusion is confirmed by the p-value = 0.024 which is less than 0.05. As in the case of socioeconomic segregation, the table (6-5 - below) shows a re-run of the ANOVA test when cities of Eastern Europe are excluded from the sample, as the region contained only 2 cities and their exclusion aims to enhance the accuracy of the analysis and to avoid any misread conclusions. Accordingly the analysis is performed again among 75 cities grouped into four regions: Western Europe, North America, Asia, and Australia. Noticeably, the F ratio = 3.792 is still greater than x = 2.73, and the p-value has improved to reach 0.014, which indicates a high significance level of the influence of the geographic region on the changes of segregation in global cities.

6.3 The individuality of cities: a discussion

Table 6-7 presents the top 10 cities with increased socioeconomic segregation and the top 10 cities with decreased socioeconomic segregation. As shown in table 6-7, the list of top 10 cities with decreased ethnic segregation is dominated by alpha cities, where 5 out of the 10 cities are ranked as alpha global cities. Yet another observation suggests that regional and national contexts have a considerable effect on the final results; where the top 8 (out of 10) cities with a decrease in ethnic segregation level are all located in Western Europe, while cities of Australia occupy the 2nd, 3rd, and 4th places of cities with the highest increase in segregation. Accordingly, the absence of clear pattern of spatial change among global cities - as well as potential similarities among cities of the same region - indicates that the understanding of the complex process of spatial restructuring of global cities requires a closer look on the local situations and particularities of individual cities. For that matter, the subsequent sections

Ethnic segregation in global cities | 181 are dedicated to discuss in more detail the motivations and factors involved in the production of diverse patterns of ethnic segregation in individual

Top ten cities with maximum increase and decrease in socioeconomic segregation and their respective ranks Top 10 cities with increased segregation City Rank Region 1 Singapore alpha Asia 2 Sydney alpha Australia 3 Perth beta Australia 4 Brisbane beta Australia 5 Valencia gamma Europe 6 Cincinnati gamma N. America 7 Tallinn gamma Europe 8 Richmond gamma N. America 9 Denver beta N. America 10 Seattle beta N. America

Top 10 cities with decreased segregation City Rank Region 1 Milan alpha Europe 2 Manchester beta Europe 3 Zurich alpha Europe 4 Barcelona alpha Europe 5 Geneva beta Europe 6 Stuttgart beta Europe 7 Rome beta Europe 8 Munich alpha Europe 9 Indianapolis gamma N. America 10 Beijing alpha Asia

182 | Analysis and findings

Beijing Hong Kong Shanghai Tokyo Seoul Singapore S.E. Asia Milan Manchester Zurich Barcelona Geneva Stuttgart Rome Munich Oslo Antwerp Copenhagen e

Madrid p Brussels Rotterdam Dusseldorf Gothenburg Berlin Euro W. London Lyon Stockholm Cologne Paris Hamburg Frankfurt Amsterdam

Marseille e Valencia p Bratislava Tallin indianapolis Orlando E. Euro Charlotte Miami Minneapolis Ottawa Montreal Atlanta Houston Washington San Fransisco Dallas Chicago Phoenix San Jose us San Diego Boston New York Los Angeles St Louis Baltimore lo America

Columbus g Cleveland Tampa An Kansas city Calgary Portland Milwakee Vancouver Philadelphia Austin Toronto Pittsburgh Detroit Edmonton Seattle Denver Richmond Cincinnati Adelaide Melbourne Wellington Auckland Brisbane

Perth Oceania Sydney 6 5 4 3 2 1 0
-1
-2 -3 -4 -5 -6 -7

Figure 6-3 global cities categorized according to their geographical region and their yearly change in ethnic segregation index index segregation ethnic in change yearly their and region geographical to their according categorized cities 6-3 global Figure

Ethnic segregation in global cities | 183 global cities. And according to the results of the ANOVA, the sections are organized based on the broad geographic regions of the world, which was already proven to have a significant influence on the outcomes of spatial restructuring in global cities. Also, for easier reading of the results presented in figures (6-1), figure (6-3) presents global cities categorized according to their location in the broad geographic regions of the world instead of their global rank. Clearly, figure (6-3) shows that the dataset is geographically imbalanced, where 66 out of the 81 cities of the dataset are located in migration-receiving countries of Western Europe and North America, while migration-sending countries of Latin America and Africa have no representative cities in the dataset. As a general observation, cities of Australia and New Zealand showed higher tendency for increased ethnic segregation, where 71.4% of cities within the region have witnessed an increase in the value of SD(m), compared to 46.1% of cities in North America, and only 33.3% of cities in Western Europe. The following sections discuss in more detail the factors contributing to the changes in the spatial arrangement of different ethnic groups in individual global cities.

6.3.1 Australian cities

Spatial segregation of ethnic groups in Australian cities was initially caused, then reinforced, by several factors including location of job opportunities, location of housing opportunities, size of ethnic groups, language, and marginalization (Hugo, 1995). During the 1950s and 1960s, available cheap rental housing was located in the inner suburban areas of major cities. Accordingly, newly arrived immigrants were attracted to those areas. However, as shown in chapter five, with the flourishing service industries in the past two decades, a new middle class arose (Badcock, 2000) and inner cities have consequently been affected by gentrification (Hugo, 1995; Badcock, 2000). As a result, the inflow of higher-income residents - who do not necessarily belong to the same ethnic group residing in inner city

184 | Analysis and findings areas – contributed to changes in patterns of geographical concentration of certain ethnic groups. For example, the close proximity of the gentry group to the existing ethnic enclave can lead to the overall decrease in ethnic segregation level in the city. However, in case that the process of gentrification is accompanied with large-scale displacement and reconcentration of disadvantaged population who are – in most cases – also immigrants, then gentrification may lead to the exacerbation of ethnic segregation by dismantling the existing enclave in inner city and the creation of another enclave elsewhere in the city. This gentrification and displacement scenario may explain the growing ethnic segregation in Sydney, Perth, and Brisbane. In figures (6-4) and (6-5), areas on the waterfront of Brisbane River such as Highgate Hill, Indooroopilly, and West End Brisbane have witnessed noticeable change in their ethnic composition. Where Middle Eastern immigrants were pushed south away from the valuable waterfront area. Yet still, gentrification and displacement scenario fails to explain the decrease in ethnic segregation in Melbourne and Adelaide. Note that Melbourne has witnessed the most extensive gentrification process across Australia (Weller and Van Hulten, 2012), but the large-scale displacement of inner-city immigrants did not contribute to the intensification of ethnic segregation. The explanation here is that the displacement of immigrant is not followed by the creation of another enclave by the displaced group, instead, the displacement caused their dispersal in other poor neighborhoods that are already inhabited by other ethnic groups, leading overall level of ethnic segregation to drop. The case of Adelaide is also a good example of such dispersal of ethnic groups due to gentrification. As shown in figures (6-6) and (6-7), in 2001, Middle eastern immigrants for example were concentrated in inner city, with insignificant presence in inner suburbs. However, their displacement in 2006 from inner city areas such as, West Torrens, Marion, and Onkaparinga Woodcroft, did not only lead to their concentration somewhere else away from gentrified areas, but more specifically, they are displaced to

Ethnic segregation in global cities | 185

N

2 mi

2 km

Figure 6-4 Brisbane in 2001– distribution of North African and Middle Eastern migrants

N

2 mi

2 km

Figure 6-5 Brisbane in 2006– distribution of North African and Middle Eastern migrants

Figure 6-4 Brisbane in 2001– distribution of North African and Middle Eastern migrants Figure 6-5 Brisbane in 2006– distribution of North African and Middle Eastern migrants

186 | Analysis and findings

N 10 mi 10 km 7 urbanized area of Adelaide in 2006, distribution 7 urbanized area of - of Middle Eastern migrants Figure 6 N 10 mi 10 km 6 urbanized area of Adelaide in 2001, distribution 6 urbanized area of - Figure 6 of Middle Eastern migrants

Figure 6-6 urbanized area of Adelaide in 2001, distribution of Middle Eastern migrants Figure 6-7 urbanized area of Adelaide in 2006, distribution of Middle Eastern migrants

Ethnic segregation in global cities | 187 neighborhoods that are originally concentrated by immigrants from South and East Asia. As a result, residential displacement in the case of Adelaide promoted higher level of integration between different ethnic groups. Finally, another explanation for the decreased ethnic segregation in Adelaide is by taking into consideration that it is the only city in Australia that managed to achieve a decrease in socioeconomic segregation through efficient local urban policies – see chapter five. In this sense, based on the assumption that certain ethnic groups are also socioeconomically disadvantaged compared to nationals; then the tackling of socioeconomic segregation can lead in the process to the creation of housing opportunities for poor immigrants outside of the enclaves, and eventually contributes to decreased ethnic segregation. Further details about the association of ethnic and socioeconomic segregation in Australian cities are explained later in chapter seven.

6.3.2 Southeast and East Asian cities

As the case of Western European cities, countries of Southeast and East Asia vary greatly in their local situations which in turn affect levels of ethnic segregation within cities of the region. The dataset contains 6 Asian cities; Tokyo, Singapore, Beijing, Shanghai, Hong Kong, and Seoul. In Beijing, Shanghai and Chinese cities in general, the phenomenon of ethnic segregation is negligible, because according to Li and Wu (2008), there are 55 main ethnic minorities in China; however, their quantity is not large enough to constitute social areas. Therefore, socioeconomic status is the major dimension of residential segregation for Chinese cities. Similar situation is detected in Tokyo, where despite the doubling of foreign residents in Tokyo since the mid-1980s, still, foreign residents constitute only 3 percent of total residents of Tokyo (Matsumoto, 2009). Unlike American and European cities, Asian cities in general tend to receive migrants from within the region; the largest three foreign population groups in Tokyo are Chinese, Koreans, and Filipinos, where Chinese are the

188 | Analysis and findings largest group representing only 1% of the residents in Tokyo (ibid.). Moreover, the spatial allocation of foreigners is highly dependent on their socioeconomic status, where Americans concentrate heavily in the central area because they are most likely to be highly paid workers. While, Chinese, Koreans, and Filipinos tend to concentrate in eastern Tokyo because they are most likely to be poor students, small shop owners, and low- paid service and industrial workers (ibid.). As shown in figure (6-3), foreigner’s concentrations patterns have not changed dramatically, as Tokyo scored insignificant increase in ethnic segregation over the past decade (0.7% in the period from 2001 to 2009). The case of Tokyo as alpha city shows that global status of any city and its increasing immigrant’s inflows do not necessarily lead to spatial segregation, rather it is the local situations such as demography or population density that can affect the spatial outcomes within global cities. Another Asian alpha city is Singapore; it shares the same global status with Tokyo. However, unlike Tokyo, Singapore witnessed the highest increase in ethnic segregation of all cities of the dataset. As a leading global city, Singapore seems to fit the model of typical global/divided city characterized in the literature, where spatial segregation is the axiomatic outcome of increased in-flows of guest workers. Therefore, the next section discusses in detail the case of Singapore, in order to determine whether the increase in segregation is mainly due to the city’s global status, contextual particularities of the region, or due to local circumstances intersecting on city level. The individuality of Singapore is derived from two main conditions. First, Singapore is a historically multiracial society since its establishment in 1819, comprised of three main ethnic groups: Chinese, Malay, and Indian (Sin, 2003). Where, Chinese are the largest group with 79% of total population followed by Malay with 14%, Indian with 6%, and the remaining 1% - labeled as ‘others’ - are mainly European and other nationalities (Singapore Department of Statistics, 1996). Second, the state intervention in

Ethnic segregation in global cities | 189 housing sector is exceptionally high. In 1998, 86% of the total population resided in public housing constructed by the Housing and Development Board (HDB) (Sin, 2003). State level of control over housing market in a multiethnic society, along with government interest in achieving socio-spatial integration of ethnic minorities have led to several policy actions and restrictions imposed by the government – through HDB - over the public sector. During the 1960s, large numbers of public housing projects have been developed, offering a wide range of dwelling units for different income groups (Van Grunsven, 2000). In addition, the allocation of households over housing projects aimed to disperse ethnic groups in space (Van Grunsven, 2000; Sin, 2003). As a result, evidence from the 1970s and early 80s showed that the provision and allocation of public housing have contributed to the desegregation of ethnic groups (Van Grunsven, 2000). However, in mid 1980s, voluntary residential mobility caused re-grouping of ethnic minorities (Van Grunsven, 2000; Sin, 2003). To combat minorities’ re-grouping, a system of ethnic quotas was imposed in March 1989, where each neighborhood had to replicate the ethnic mix of Singapore. In other words, the HDB set a “maximum limit on the percentages of Chinese, Malay and Indian/Others living in each neighborhood” (Sin, 2003: 530), and when a neighborhood reached or exceeded quota limits, owners can resell their flats only to buyers of the same ethnic group. For example, in a neighborhood that is mostly inhabited by Malays, Malays can sell to Malays, but a Chinese, Indian, or others living in that neighborhood cannot sell to a potential Malay buyer (ibid.). However, the ethnic quota system did not succeed in re-achieving ethnic integration for two reasons. First, the quota was not comprehensively imposed over all neighborhoods, additionally, “household were not forced to move from or to existing neighborhoods and/or blocks which did not conform to the ethnic limit” (Van Grunsven, 2000: 119). Which implies that households of the same ethnicity could move and cluster in those

190 | Analysis and findings neighborhoods where ethnic quota is not considered. Second, defining, implementing, and following up ethnic quotas of every neighborhood require the government to impose full control over housing market transactions, where full control is ‘extreme intervention’ that the government abstained to adopt. As figures (6-8) and (6-9) show, despite the housing regulations, patterns of clustering of the Indian population have changed visibly from 2000 to 2010. As a result, the increased ethnic segregation showed in figure (6-9) reflects the spatial outcome of residential mobility of different ethnic groups based on their preferences to re-cluster with co- ethnics, under incapacitated housing regulations. The case of Singapore as presented above does not support the assumption that the in-flow of guest workers into the city is the main cause for intensified ethnic segregation. In fact, foreign workers’ regulations compel employers to provide suitable accommodation for their foreign workers by their arrival in Singapore (Cho, 2011). Also, foreign workers – with exception of highly skilled professionals – are legally banned from renting any residential properties on the market. Therefore, it is nearly impossible for low-skilled foreign workers to form their own ethnic enclave. For highly skilled professional, some ethnic groups show a tendency for clustering such as Japanese and French, however, their relatively low proportion to total population made their segregation insignificant (ibid.) especially on the macro level when compared to segregation of Malays or Indians. Comparing ethnic segregation levels of Singapore to those of Hong Kong sheds more light on the individuality of cities as the main factor in explaining spatial transformations within global cities. Both Singapore and Hong Kong are leading alpha cities located in the economically booming Asian region. However, Singapore witnessed increase in ethnic segregation due to its historical ethnic composition and local housing policies. While Hong Kong, witnessed a decrease in ethnic segregation, where its ethnic

Ethnic segregation in global cities | 191

N

5 mi

5 km

Figure 6-8 Singapore in 2000, the distribution of Indian population

N

5 mi

5 km

Figure 6-9 Singapore in 2010, the distribution of Indian population

Figure 6-8 Singapore in 2000, the distribution of Indian population Figure 6-9 Singapore in 2010, the distribution of Indian population

192 | Analysis and findings composition is mainly impacted by handover of Hong Kong to the Chinese authority in 1997, and the consequent changes in entry visa and work permits regulations especially for Mainland Chinese (Ullah, 2012). Where, 95% of total population of Hong Kong is Chinese, most of them are either descendants of migrants from China or were themselves born in China (ibid.). The handover of Hong Kong was followed by the outbreak of the H5N1 avian influenza, and then in 2003, Hong Kong and China suffered the epidemic of severe acute respiratory syndrome (SARS), where “[T]he epidemic not only severely affected the health of the people but also produced many related social, economic, and humanitarian problems including tourism, international travel and trade.” (Lee, 2006:11). Consequently, the in-flows of foreign (non-Chinese) workers to Hong Kong slowed down compared to other alpha cities in the same period of time. As a result, decreased level of ethnic segregation in Hong Kong is caused by the in-flows of Mainland Chinese in a basically Chinese community, along with relative retreat of foreign immigration. In general, the discussion of Asian cities showed that history and local circumstances are the main factors impacting socio-spatial transformations within cities. Alpha cities of Tokyo, Singapore, and Hong Kong proved that patterns of segregation are not related to global status, and the shifts on global levels are eventually defined by local contexts of each city.

6.3.3 North American cities

Despite the fact that it is a major destination of international immigrants, issue of residential segregation in the U.S. has always been orbiting around segregation of African Americans3 (Massey and Denton, 1993), and later the concerns included the segregation of Hispanic population (Cross, 1992). Only in recent decades, more focus was given to the impacts of the enormous international migration received by American cities. Between 1980 and 1989, the U.S. received third of all the world’s

Ethnic segregation in global cities | 193 international migrants (Zlotnik, 1993), where the state of California alone accommodates more permanent immigrants than any other country in the world (Massey et al., 2009). The same migration surge is observed in Canada, where in the 1990s, 16% of Canada’s total population was foreign- born residents (ibid.). According to figure (6-3), despite the increase in immigration flows received by the region, yet only 46.1% of North American cities have witnessed an increase in ethnic segregation levels. This observation matches Cutler et al. (2008) conclusion that changing pattern of immigrants’ segregation is not related to the share of immigrants of total population. Where in the U.S. – for example – the in-flows of immigrants are not distributed evenly among cities; instead, Sunbelt4 cities receive significantly higher percentage of immigrants, especially from Mexico (Thompson, 2011), compared to Rustbelt cities (Frey, 2002). Yet still, 11 out of 14 Sunbelt cities have witnessed a decrease in ethnic segregation levels over the past decade. This inverse correlation between the in-flows of immigrants and levels of ethnic segregation is explained by, on one hand, when the flows of immigrants are large enough to exceed the capacity of the existing ethnic community to accommodate the new comers, immigrants start to search for affordable housing in other neighborhoods, hence, they become dispersed in different areas of the city, which lead to higher ethnic mix and eventually the overall ethnic segregation level will drop accordingly. Note that the dispersal of immigrants in this scenario is similar to the situation in Australian cities; the only difference is that in Australia the dispersal is caused by gentrification and displacement of disadvantaged minorities, while in the U.S. the dispersal is a result of the growing in-flows of immigrants. On the other hand, instead of the dispersal of new comers to the city outside the ethnic enclave, the case of Houston (see figures 6-10 and 6- 11) shows that previous generations - who are in a better socioeconomic status then new comers - tend to move to the suburbs, while t he existing

194 | Analysis and findings

N 10 mi 10 km 10 Houston urbanized area in 2000, Distribution of Mexican, Caribbean, and Central American 10 Houston urbanized area in 2000, Distribution of Mexican, Caribbean, and Central - Figure 6

Figure 6-10 Houston urbanized area in 2000, Distribution of Mexican, Caribbean, and Central American

Ethnic segregation in global cities | 195

N 10 mi 10 km 11 Houston urbanized area in 2009, Distribution of Mexican, Caribbean, and Central American Houston urbanized area in 2009, Distribution of Mexican, Caribbean, and Central 11 - Figure 6

Figure 6-11 Houston urbanized area in 2009, Distribution of Mexican, Caribbean, and Central American

196 | Analysis and findings enclaves are sustained by new comers. Either way, ethnic segregation level decreases due to the presence of large portion of an ethnic group outside their enclave, yet in the first case, dispersal of poor immigrants can lead to the decrease in socioeconomic segregation as well, while in the second case, well-off settlers moving to the suburb can lead to the exacerbation of socioeconomic segregation, as they leave behind a concentration of disadvantaged immigrants isolated in poor neighborhoods (McGeary and Lynn, 1988). The next chapter discusses in more detail how ethnic and socioeconomic segregation are important contingencies in the understanding of one another. These explanations for the decreased ethnic segregation in Sunbelt cities clarify how cities like Miami, Phoenix, and San Francisco, where the foreign born population have increased by 15.2%, 13.2%, and 10.4% respectively 1990 to 2000 (Frey, 2002). Yet at the same time, all three cities scored significant decrease in ethnic segregation by 12.0%, 5.9%, and 8.7% respectively. While on the contrary, Cincinnati – which is a Rustbelt city - scored the highest degree of increase in ethnic segregation, although the growth of foreign-born population in Cincinnati was only 1.1% from 1990 to 2000, which strongly suggests that increased ethnic segregation is not necessarily a result of the growing in-flows of immigrants into global cities. Instead, changes in ethnic segregation patterns within cities depend highly on other factors such as characteristics of each ethnic group (Logan, 2000), their background, culture, education, and socioeconomic status. Another observation drawn from figure (6-3) suggests that there are inconsistent pattern of change in ethnic segregation in a number of individual cities. One example for such inconsistency is the cases of Austin, Houston, and Dallas; all three cities are Sunbelt cities and sharing similar characteristics of the region, firstly, they are all located in Texas. Secondly, they witnessed parallel growth in the share of immigrant workers especially from Mexico (Thompson, 2011), where the foreign born population have increased in Austin, Houston, and Dallas by 11.4%, 11.5% and 11.7%

Ethnic segregation in global cities | 197 respectively from 1990 to 2000 (Frey, 2002). And finally, the region is known for its thriving service industries, as Texas is one of the top 5 states with growth of service and high-tech industries in 2005 (The State of Texas, 2011). However, all those shared characteristics did not have equal effect on the outcomes of spatial transformation of the three cities. According to figure (6-3), Austin scored an increase in ethnic segregation by 4.8% from 2000 to 2009. While on the contrary, both Houston and Dallas scored a decrease in ethnic segregation by 9.0% and 8.5% respectively over the same period of time. The reason behind this unlikely increase in ethnic segregation in Austin is a very particular characteristic of the city, which is the interstate highway (I-35). The highway is built in the 1950s (Skop, 2009) and it is passing through the city while geographically isolating the historically persistent ‘East Austin enclave’ that is basically a Mexican community. In this case the interstate highway is acting like a physical barrier that prevents the enclave from dissolving into the city.

6.3.4 Western European cities

During the post World War II period, Western Europe witnessed sizable flows of guest workers pouring in cities of the region. Initially, guest workers were recruited from Southern European countries such as Spain and Italy, then later, workers were recruited from other countries such as Morocco, Turkey, Algeria, India, …etc. (Van Kempen, 2005). Then, with a large number of temporary guest workers turning into settlers in cities of destination – along with opening the door for family reunion – the result was the formation of ethnic communities (Castles, 2002), which became a common attribute of Western European cities The rise, then the persistence, of ethnic enclaves in Europe – as shown in chapter 3 - are perceived as an outcome of two parallel processes. First, self-segregation of minorities who choose willingly to live with co-ethnics for reasons of safety, enhancing sense of belonging, and maintaining customs and traditional values of their particular culture (Fischer, 1976).

198 | Analysis and findings

Second, in case that immigrants’ preferences are not leading to self- segregation, their low socioeconomic status comparing to nationals (Van Kempen, 2005) eventually limits their housing choices outside the enclave. Yet apparently, despite these mechanisms of voluntary and involuntary segregation of ethnic minorities, according to figure (6-3), levels of ethnic segregation have fallen in 66% of Western European cities of the dataset. Furthermore, a closer look on the situation in individual Western European cities reveals that changes in level of ethnic segregation occur for different reasons that vary greatly from country to country or even within the same country. Undoubtedly, European cities share distinctive characteristics, which differentiate them from other cities of the developed world such as North American cities (Kazepov, 2005). Yet, within Europe, dissimilarities exist on different territorial levels from states to cities (ibid.). In the process of explaining ethnic segregation, variations in the housing supply, policies, and practices on the local level can have different impacts on changes of ethnic segregation in individual cities. The role of housing policies in particular gained a special focus in segregation studies. On one hand, Van Kempen (2005) explained extensively how ethnic minorities’ choices in European cities could be affected greatly from city to city according to the type, quality, and affordability of housing supply on each market. He stated, “a lack of social rented housing in Belgium largely explains the concentrations of Turks in the city of Brussels, and the location and accessibility of social housing in the Netherlands, could be the principle explanatory factor in Dutch cities. Discriminatory regulations have a very important role in Austria, while patterns of ‘choice’ might be more important for different groups in the UK.” (Van Kempen, 2005:205). Therefore, cross-nation variations in housing policies on centralized national level or on local governments level will highly influence spatial transformation within cities. Results from figure (6-3) partially support this conclusion, on the national level, countries of France, Switzerland, and Italy

Ethnic segregation in global cities | 199 show consistent nation-based changes on their cities level, French cities of Marseille, Paris, and Lyon show parallel increase in level of ethnic segregation, while Italian cities of Rome and Milan as well as Swiss cities of Zurich and Geneva show significant decreased in ethnic segregation. On the other hand, German, Dutch, and Spanish cities show noticeable divergence in results between cities within each nation, as the cases of Frankfurt vs. Stuttgart, Amsterdam vs. Rotterdam, and Valencia vs. Barcelona. Those cases either reflect the influential role of local governance on sub-national level. Or as Musterd and Fullaondo concluded, that “more social housing and stronger welfare state neither relate to lower, nor to higher levels of segregation compared to contexts where the welfare state and the housing market are clearly different.” (Musterd and Fullaondo, 2008: 112). Their conclusion is based on comparative analytical study of ethnic segregation and the housing market in Amsterdam and Barcelona. Where the large social housing stock in Amsterdam and the relatively strong welfare state of Dutch cities are encountered by large number of migrants, causing segregation levels to be similar to that of Barcelona. Despite the fact that housing stock in Barcelona is much more private, and the welfare state is weaker compared to the Dutch, but the lower share of migrants in Barcelona – compared to Amsterdam - neutralizes the impacts of absent public housing - and the role of welfare state in general - on the process of ethnic segregation (ibid.). Resulting similar levels of segregation in two totally different situations, which imply that ethnic segregation is a complex process and it is not solely related to housing policies and issues of affordability or choice. In fact, the seemingly opposing perspectives of Van Kempen (2005) and Musterd and Fullaondo (2008) regarding the relevance of housing policies to ethnic segregation are supported by individual cases shown in figure (6-3). On one hand, the increased segregation in Amsterdam is coinciding with the retreating role of the state and the consequent privatization of socially rented housing in recent years. According to

200 | Analysis and findings

Ministry of VROM5 (2003), socially rented housing units in 2001 comprised 57% of total housing units in Amsterdam; this percentage has dropped to reach 47.4% in 2011 (Ministry of BZK6, 2012). Which backs up Van Kempen assessment of the important role of the state policies. On the other hand, Berlin has witnessed a similar and even more dramatic reduction of public housing stock than Amsterdam (from 30% of total housing units in 1990 to only 15% in 2008), along with increasing rent prices in privatized housing stock (Aalbers and Holm, 2008). Given those two conditions, the expected spatial outcome for Berlin would be increased segregation as Amsterdam. However, Berlin showed insignificant change in ethnic segregation over the years, which implies – as Musterd and Fullaondo (2008) suggested - that factors involved in producing and sustaining ethnic segregation cannot be reduced to housing policies or welfare weaknesses. That discussion of ethnic segregation in Western European cities leads to the following conclusions. First, tracing patterns on regional level are proved to be as difficult as on the global level, despite the similar characteristics shared by nations and cities of the region. Second, spatial outcomes of each city is distinctive, where individuality of cities is derived from their local situations that vary greatly in terms of housing supply on one side, and size, background, and socioeconomic status of minorities on the other side.

Ethnic segregation in global cities | 201

Summary

Spatial segregation of ethnic minorities is a complex phenomenon, the cases discussed in this chapter show that changes in ethnic segregation are highly dependent on wide range of variables, including history, housing policies, and socio-economic status of the segregated minority. Generalized patterns of spatial transformations on the global level are not detected; the 81 cities of the dataset have showed diverse directions of change in segregation level, where only 44.4% of global cities have witnessed an increase in ethnic segregation, which challenges the validity of the generalized assumptions in the literature. This conclusion is also supported by a regression analysis, which confirmed that cities’ global network connectivity explains only 0.2% of the variability in spatial segregation index. Moreover, the one-way analysis of variance ANOVA revealed that the cities’ position in the global network represented in alpha, beta, and gamma categorization are not associated with certain changes in socioeconomic segregation, while the location of global cities in certain geographic region is proved influential on the resultant levels of segregation. The regional comparisons of cities of Australia, North America, Western Europe, and Southeast Asia have showed that local situations on the level of cities vary greatly within the same region and sometime within the same nation-state (as the case of Texan cities or Amsterdam and Rotterdam). The impacts of global economic and political shifts on the spatial transformations within cities are not uniform. For example, the retreating role of the welfare state does not have the same effect on segregation levels in cities such as Singapore when compared to Amsterdam. Also, the in- flows of immigrant workers do not necessarily cause increase in ethnic segregation as the case of Californian cities. Recapitulating, the discussion over the situations in individual cities revealed a number of factors contributing to changes in ethnic segregation levels in global cities; firstly, gentrification and its associated displacement

202 | Analysis and findings and reconcentration (or dispersal) of disadvantaged minorities (as the case in Australian cities). Secondly, the growing in-flows of immigrants to the city, which can lead to the decreased ethnic segregation due to the dispersal of ethnic groups outside the enclave (as the case in American cities). Thirdly, the efficiency of urban policies and the characteristics of public housing provided by the state (in terms of distribution, access, …etc.) (As in the case of Singapore and Western European cities), can have varying effects on the resultant ethnic segregation patterns. Fourthly, a specific characteristic of the city – such as a physical barrier in Austin or the handover of Hong Kong - that can contribute to either higher or lower levels of ethnic segregation. Finally, since changes in ethnic segregation is tied to residential mobility, then segregation levels are related to the financial ability of households to move out/in the ethnic enclaves. The next chapter discusses the spatial outcome resulting from the intersection of patterns of ethnic segregation with patterns of socioeconomic segregation within the same city.

Ethnic segregation in global cities | 203

Notes

1 The variations are based on the official census methodology of each country 2 Note that alpha global cities of Mumbai (India), Johannesburg (South Africa), Jakarta (Indonesia), and beta global cities of Cairo (Egypt), Bogota (Colombia), Bucharest (Romania), and Karachi (Pakistan) are all receiving rural immigrants from within their respective regions (immigrants with the same ethnicity and therefore are irrelevant to the analysis), while at the same time, the regions are sending immigration to the usual destinations of North America, Western Europe, and Australia, and any amount of international in-migration to those global cities is outweighed by the enormous outer-migration 3 Early European migration to the U.S was also a concern for research. However, European enclaves of Irish, Germans, then later for Poles, Italians, and Czechs were perceived – for Chicago school - as a transitional phenomenon after which assimilation is inevitable (Park, 1926; Wirth, 1928; Burgess and Newcomb, 1933). 4 Sunbelt cities are located in the Southern states, while Rustbelt cities are the deindustrialized cities of the Northeastern and the East North Central States. 5 The Ministry of Housing, Spatial Planning and the Environment (Dutch: Ministerie van Volkshuisvesting, Ruimtelijke Ordening en Milieu or VROM. It was merged with the Ministry of Transport, Public Works and Water Management into the new Ministry of Infrastructure and the Environment on 14 October 2010. 6 The Ministry of the Interior and Kingdom Relations (Dutch: Ministerie van Binnenlandse Zaken en Koninkrijksrelaties; BZK) is the Dutch Ministry responsible for Home Affairs, Civil service, Intelligence and the relations with the other countries in the Kingdom of the Netherlands.

204 | Analysis and findings

Chapter 7

7 The color of poverty: the correlation between socioeconomic and ethnic segregation

The aim of this chapter is to further discuss the influential role of local contingencies on the changes in spatial segregation levels in individual global cities. The chapter follows up on the conclusions drawn from the findings of the analysis presented in chapters five and six, with a particular focus on the potential correlation between the socioeconomic status of a certain population group and their ethnic background. In other words, a number of cases presented in the previous chapters revealed that ethnic segregation patterns are changing due to the change in the socioeconomic status of the segregated ethnic groups (as in the case of Houston, where the well-off Mexican group could afford to move to the suburbs and leave the ethnic enclave). This observation matches what Timberlake et al. (2012: 74) found in their study of income polarization in 57 US cities; they concluded that cities with higher immigration in-flows are more likely to show an increase in income polarization due to the agglomeration of global economic activities in these cities compared to cities with less in-flows of immigrants. In this case, socioeconomic segregation is itself a contingency that affects patterns of ethnic segregation, and vice versa, where socioeconomic segregation patterns might be determined by patterns of clustering or regrouping of a certain ethnic group (as the case of Singapore, where the

206 | Analysis and findings voluntary regrouping of Indian population – who are mostly less affluent than the other groups - contributed to the concentration of poverty). In this case, pattern of ethnic segregation in individual city can be considered an important contingency that affects the changes in socioeconomic segregation. Investigating this correlation between socioeconomic and ethnic segregation is particularly relevant to the purpose of this research. In more detail, the three school of thoughts discussed in chapter two acknowledge the influence of macro economic changes on the micro socio-spatial settings of cities, but in case that the resultant socio-spatial change is caused (or at least modified) by other local contingent factors. Then, it can be argued that the acknowledged role of economic globalization as a main generator of socio-spatial division is overstretched. Accordingly, the subsequent sections investigate the mutual influence of socioeconomic and ethnic segregation on one another; the analysis findings are not conclusive. However, it raises several questions on the generalization tendencies in the global city literature.

7.1 Intersection of socioeconomic and ethnic segregation: a discussion

“Ethnic and social spatial inequalities cannot be understood in a one-dimensional way. Ethnic segregation has socio-economic components” (Musterd, 2005: 339)

According to Burgers and Musterd’s (2002) study of Amsterdam and Rotterdam, the ‘subcultural’ variable emerged as an important factor that affects the socioeconomic status of Turks and Moroccans, as well as their spatial distribution in both cities. Similarly, Wilson’s mismatch theory (1987) paid a particular attention to the racial/socioeconomic segregation of African Americans, as the residential segregation of African Americans is

The color of poverty | 207 not only historically persistent, but it is also institutionalized, and connected to other racial discriminatory processes; such as, limited opportunities for African Americans in the labor market, their chances of getting decent education, exclusion from the housing market, and their exposure to higher poverty rates. Accordingly, a wide range of segregation studies – especially on U.S urban areas - is dedicated to explore the connection between race/ethnicity on one hand, and poverty/socioeconomic status on the other hand. Massey and Denton (1993) explained in detail that the association between race and poverty is a key issue in explaining the dynamics of residential segregation in urban areas of the U.S. In fact, both Wilson (1987) and Massey (1990) dealt with race/poverty connection from different angles; they used the terms “minority underclass” and “black poverty” respectively, both terms are combined of two sections, the first refers to race/ethnicity status (minority – black), and the second refers to the socioeconomic status (underclass – poverty) of the segregated population, linking a certain race to a certain social and economic conditions. However, Wilson and Massey disagreed on the degree of importance of race in this equation. In Wilson’s view (1987), race alone cannot explain the sharp increase in inner city poverty in the 1970s, especially when considering the macro economic changes impacting western societies (such as deindustrialization). While, Massey (1990) and Massey and Denton (1993) insist that in a racially segregated environment, any increase of poverty will automatically lead to geographic concentration of poverty, because the higher poverty rates are absorbed by few number of racially segregated neighborhoods. In order to explain how racial segregation contributes to the concentration of poverty, Massey (1990) constructed a hypothetical city of 128,000 people, distributed evenly among 16 equal-sized neighborhoods, 8000 persons each. Massey tried to distribute the population race groups based on the real situation in many American cities in the 1970s. Therefore, the distribution of the population in the hypothetical city was determined

208 | Analysis and findings based on a previous study by Massey and Eggers (1990), in which they examined the poverty rates for different races in 60 metropolitan areas in the U.S. In average, African Americans were 25% of the population, with poverty rate of 20%, compared to 75% whites, with poverty rate of 10%. Massey started by distributing blacks and whites by those ratios evenly over the 16 neighborhood. Accordingly, average poverty rates were equal for all neighborhoods and there is no racial segregation. Then to create a racially segregated environment, he redistributed black population to be concentrated in 12 then 8 then 4 neighborhoods of the 16. Each time he raise the level of black segregation, he finds that poverty rates among black neighborhoods steadily rise while poverty rates among white neighborhoods steadily fall. Because in average, blacks tend to have higher poverty rates then whites, and their concentration in few neighborhoods lead to the geographic concentration of poverty (Massey, 1990). Since racial segregation and poverty concentration are theoretically contributing to the persistence of one another, then it is logical to assume that any external factor (macro changes) affecting either race or class composition of the city will automatically affect the other. For instance, if deindustrialization and privatization are causing higher levels of unemployment, welfare dependence, and eventually increasing poverty (Marcuse and Van Kempen, 2000; Mingione, 2005), and under the condition that the city is already highly segregated by race, then – hypothetically – increased poverty will be concentrated and the city will show higher levels of socioeconomic segregation over time – note that poverty concentration in this case is partially caused by the existing racial segregation and not only due to the rising levels of unemployment or welfare dependence . Similarly, increased poverty will contribute to the persistence of the existing racial segregation. Massey and Denton (1993) support this hypothesis by describing the situation of Blacks during the economic hardship of the 1930s and 1970s; “During the 1930s, […], the Great Depression brought a wave of factory closings, bankruptcies, bank failures, and very high rates of

The color of poverty | 209 unemployment in the black community. During the 1970s, successive recessions, bursts of inflation, and increased foreign competition eliminated many high-paying jobs in manufacturing, lowered wages, and decreased the real value of welfare payments. These dislocations took a heavy toll on the distribution of black income, especially among families in the industrial cities of the northeast and midwest. As a result of the downward shift in black incomes, poverty rates increased substantially in both decades.”(Massey and Denton, 1993:125). Moreover, as noted in chapter one, the high poverty rates among the segregated ethnic minority create a vicious cycle of social exclusion and spatial segregation, which causes spatial segregation patterns to persist. As, poverty creates very difficult conditions for Blacks or other segregated minorities to escape the segregated neighborhoods (Massey et al., 1994). Because poverty is associated with other social problems such as violence, crime, unwed child bearing, divorces, single parenting, low educational achievement, poor health, drug use ... etc. (Wilson 1987; Soja, 1989; Massey and Denton, 1993; Andersen, 2002; Wassmer, 2005; Varady, 2005). These problems affect the ability of the segregated population to find well-paying jobs, decent education, and better housing, which consequently limit their opportunities for social upward mobility (Wilson, 1987; Kazepov, 2005; Boal, 2005; Massey and Denton, 1993; Madanipour et al., 1998). Eventually, the socioeconomic status of the segregated minority is the key factor contributing to the persistence of the ethnic enclave and vice versa.

7.2 Socioeconomic and ethnic segregation: parallel or divergent changes?

In total, out of the 81 cities with available data on the spatial distribution of different ethnic groups, only 56 cities have comparable data about the distribution of socioeconomic groups. The 56 cities are categorized into 17

210 | Analysis and findings

New York Hong Kong

Singapour Chicago Sydney Toronto Los Angeles Amsterdam Brussels

San Fransisco alpha Washington Miami Melbourne Boston Dallas Atlanta Philadelphia Stockholm Montreal Houston Berlin Copenhagen Vancouver Seattle Auckland Oslo   
  

Minneapolis beta Brisbane Detroit Denver St Louis San Diego Perth Cleveland Calgary Cincinnati Charlotte Baltimore Adelaide Portland San Jose us

Kansas city   
  
Phoenix Rotterdam Tampa Columbus

indianapolis gamma Pittsburgh Edmonton Orlando Gothenburg Ottawa Richmond Austin Milwakee Wellington  





 



Figure 7-1 alpha – beta – gamma cities and their yearly percentage of change in both socioeconomic and ethnic segregation index segregation ethnic and socioeconomic both in change of percentage yearly their and cities gamma – beta – alpha 7-1 Figure

Figure 7-1 alpha – beta – gamma cities and their yearly percentage of change in both socioeconomic and ethnic segregation index

The color of poverty | 211 alpha, 18 beta, and 21 gamma cities. Figure (7-1) presents the changes in both types of segregation in cities of each global level. For a better understanding of results in figure (7-1), figure (7-2) shows, cities categorized into four types based on the possible directions of change in both socioeconomic and ethnic segregation:

1. Cities with increase in both types of segregation 2. Cities with increased socioeconomic segregation and decreased ethnic segregation 3. Cities with decreased socioeconomic segregation and increased ethnic segregation 4. Cities with decrease in both types of segregation

The aim from this categorization is to detect if there is a tendency among cities to show parallel changes in both socioeconomic and ethnic segregation levels. According to figure (7-2) 19 cities follow the first case (increase in both types), 10 cities follow the second case (increased socioeconomic with decreased ethnic segregation), 7 cities follow the third case (decreased socioeconomic with increased ethnic segregation), and 20 cities follow the fourth case (decrease in both types). In total, the results show that 39 out of 56 cities (almost 70 percent) showed a parallel increase (or decrease) in both socioeconomic and ethnic segregation, compared to only 17 cities showed contrasting directions of change between both types of segregation. As noted, the parallel changes in both types of segregation are explained theoretically based on the credible assumption that immigrants are more likely to be in a lower socioeconomic status than nationals (Van Kempen, 2000; Massey and Denton, 1993). Consequently, any factor that contributes to more concentration (or dispersion) of ethnic minorities will also lead to socioeconomic segregation (or integration). While, the less prevalent outcome of divergent changes in both types of segregation is explained by the particular situations in individual cities. For example, the

212 | Analysis and findings

Edmonton Brisbane Philadelphia Perth

Cleveland segregation Kansas city Tampa Pittsburgh Milwakee Cincinnati Portland (+,+) Richmond Detroit Auckland Seattle Stockholm Sydney Denver Amsterdam Dallas Baltimore San Jose us New York Boston

Orlando (+,-) Ottawa   
  
Melbourne Oslo Gothenburg Austin Toronto Calgary Singapour Columbus (-,+) Wellington Vancouver Brussels Rotterdam Berlin San Fransisco Copenhagen

Atlanta   
  
Adelaide Montreal Miami Chicago

Phoenix (-,-) San Diego Washington St Louis Los Angeles Minneapolis indianapolis Hong Kong Charlotte Houston  





 



Figure 7-2 cities of the dataset categorized into four groups of possible directions of change in both socioeconomic and ethnic and socioeconomic both in of change directions of possible groups four into categorized dataset the of 7-2 cities Figure

Figure 7-2 cities of the dataset categorized into four groups of possible directions of change in both socioeconomic and ethnic segregation

The color of poverty | 213

A matrix shows cities of the dataset classified according to their global status and their direction of change in both socioeconomic and ethnic segregation: alpha beta gamma (+) Increased Sydney Stockholm Cincinnati socioeconomic Amsterdam Seattle Portland segregation Philadelphia Auckland Kansas City Brisbane Tampa (+) Increased Detroit Pittsburg ethnic segregation Denver Edmonton

Perth Richmond Cleveland Milwaukee (-) Decreased Hong Kong Montreal Charlotte socioeconomic Chicago Houston Adelaide of segregation segregation Los Angeles Berlin Phoenix Brussels Copenhagen Rotterdam Parallel changes in both types (-) Decreased San Francisco Minneapolis Indianapolis ethnic segregation Washington St. Louis Miami San Diego Atlanta (+) Increased New York Oslo Baltimore socioeconomic Melbourne San Jose segregation Boston Orlando Dallas Gothenburg

(-) Decreased Ottawa ethnic segregation (-) Decreased Singapore Vancouver Austin

segregation socioeconomic Toronto Calgary Wellington segregation Columbus

(+) Increased Divergent changes in both types of ethnic segregation displacement of minorities from central areas - as a result of gentrification – to adjacent neighborhood dominated by another minority group, causing increase in socioeconomic segregation while at the same time higher level of ethnic integration (see the case of Adelaide). Another aspect affects spatial changes of socioeconomic and ethnic segregation is self-segregation, where

214 | Analysis and findings affluent minorities prefer to live with co-ethnics regardless their socioeconomic status, causing socioeconomic segregation to decrease while maintaining the ethnic divsion (see Varady, 2005). The discussion in chapters two and three suggests that both socioeconomic and ethnic segregation are expected to intensify in global cities as a result of the macro economic changes impacting these cities. However, according to table (7-1) - which presents a matrix of cities of the dataset classified according to their global status and their direction of change in both socioeconomic and ethnic segregation - only 33.9% of global cities of the dataset followed the global/divided city model and showed parallel increase in both socioeconomic and ethnic segregation. While, 35.7% of cities showed a parallel decrease in both types of segregation. Another remark drawn from table 7-1 is that – statistically speaking - alpha cities are not necessarily more prone to increase in spatial segregation, where only 3 out of 17 alpha cities (17.6%) showed parallel increase in both types of segregation. Instead, alpha cities are more likely to have a parallel decrease in both types compared to beta and gamma cities.

7.2.1 Examining the strength of association between socioeconomic and ethnic segregation

In terms of direction of change, the observations in figure (7-2) and table (7-1) suggest that the change in one type of segregation tends to follow the change in the other type, and a chi-squared test already confirmed this observation. While, in terms of the intensity of change, a linear regression analysis is performed to assess the strength of the association between the changes in socioeconomic and ethnic segregation in global cities. Figure (7- 3) presents a summary scatterplot for the correlation between the two variables. Also, The SPSS generates several tables for the linear regression analysis. In this section, we show the three main tables required to

The color of poverty | 215 understand the results of the of the linear regression procedure: the model summary table, the ANOVA table, and the coefficients table.

6

5

4

3

2

1

0 -4 -3 -2 -1 0 1 2 3 -1 Standardized change in ethnic segregation y = 0.2332x - 0.0633 -2

-3 Standardized change in socioeconomic segregation

Figure 7-3 Summary scatterplot for the correlation between the standardized change in socioeconomic segregation and ethnic segregation in global cities

Firstly, the regression output table (7-2) provides the information needed to determine how well the regression model fits the data. In the model summary section, the column ‘R’ is the absolute value of the Pearson correlation coefficient between the dependent and the independent variables. It simply indicates the strength of the association between the two variables. In our case, R = 0.226, and since the R value ranges from 0 to 1, then the observations indicate low to moderate correlation between the changes in socioeconomic and ethnic segregation in global cities. The R2 value in the R2 column represents the proportion of variance in the dependent variable

216 | Analysis and findings that can be explained by the independent variable. In our case, R2 = 0.051, which means that the independent variable, change in socioeconomic segregation, explains 5.1% of the variability of the dependent variable, change in ethnic segregation. Please note that the R2 is calculated for the sample (56 global cities), SPSS generates another value called adjusted R2 that represents the proportion of variance if the analysis is performed over the entire population (i.e. the full list of global cities). Table (7-2) shows that the adjusted R2 = 0.033.

Table 7-2 Regression output – model summary, ANOVA, and Coefficient tables Model Summary Adjusted Std. Error of the Model R R Square R Square Estimate 1 .226a .051 .033 .75379511236 a. Predictors: (Constant), zscoreINCOME

ANOVAa Sum of Mean Model Squares df Square F Sig. 1 Regression 1.650 1 1.650 2.904 .094b Residual 30.683 54 .568 Total 32.334 55 a. Dependent Variable: zscoreETHNICITY b. Predictors: (Constant), zscoreINCOME

Coefficientsa Unstandardized Standardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) .239 .101 2.362 .022 zscoreINCOME .188 .110 .226 1.704 .094 a. Dependent Variable: zscoreETHNICITY

The color of poverty | 217

The next table is the ANOVA table, which reports how well the regression equation fits the data (i.e., how accurately the model can predicts the dependent variable – ethnic segregation - based on the changes in the independent variable socioeconomic segregation). The number in the "Regression" row and the "Sig." column indicates the statistical significance of the regression model that was run. Here, the regression model is statistically significant, F (1, 54) = 2.904, p = 0.094, which is greater than 0.05, and indicates that, overall, the regression model cannot statistically significantly predict the dependent variable. The last table is the coefficients table, which provides the coefficients of the regression equation that is used to predict the dependent variable from the independent variable. y = b0 + (b1 x) where y is the change in ethnic segregation, x is change in socioeconomic segregation, b0 is the intercept and b1 is the coefficient, both can be found under the B column in the table, b0 = 0.239, and b1 = 0.188. In our case, it is already established that the model poorly fits the data due to the weak dependence between the variables. Recapitulating, the linear regression established that the change in both types of segregation is moderately correlated, as the change in socioeconomic segregation explains 5.1% of the variability in ethnic segregation. Also, the regression model cannot statistically significantly predict the change in one type of segregation based on the change in the other. A comparison of these observations with the findings of the previous chapters reveals that change in any type of segregation is weakly correlated to the position of the city in the global network, while moderately correlated to the change in the other type of segregation within the same city, as only 0.2% of socioeconomic segregation and 0.2% of ethnic segregation are explained by cities’ global network connectivity, while this percentage jumps to 5.1% when segregation types are explained by one another. In other words, a considerable part of the increased (or decreased) spatial

218 | Analysis and findings segregation in global cities is a result of the long-standing socioeconomic and ethnic structure, rather than the cities’ level of integration in the global economy. Accordingly, a closer look on the local situation of individual cities is required, especially for cities showing divergent changes in both types of segregation, and cities excluded from the analysis for being outliers.

7.3 Regional and local patterns of association between socioeconomic and ethnic segregation

As clarified in chapter four, the sample is regionally unbalanced due to data availability limitations. Therefore, this section focuses on patterns of spatial changes in three regions: Australia and New Zealand by 7 cities, North America by 39 cities, and Western Europe by 8 cities, along with only 2 cities representing Southeast Asia that are discussed individually in the end of this chapter. Figure (7-4) shows that the increase in both types of segregation is most likely to occur in Australia and New Zealand followed by North America then Western Europe. Where, 4 out of 7 cities (57.1%) of all Australia and New Zealand cities in the dataset show parallel increase in socioeconomic and ethnic segregation, compared to 33.3% of North American cities, and 25% of Western European cities.

7.3.1 Australian cities

Healy and Birrell (2003) and Gwyther (2005) claim that ethnic and socioeconomic segregation in Australian cities are reinforcing one another, where Australian-born and English-speaking residents move to gentrified areas, while residents from non-English-speaking origins are concentrated in low-income neighborhoods. The strong and clear association between low socioeconomic status and immigrants resulted in simultaneous increase in both socioeconomic and ethnic segregation in almost 60% of Australian global cities. Also, the same logic explains the parallel decrease in both types of segregation in the case of Adelaide, where despite the inner-city

The color of poverty | 219

6

5

4

3

2

1

.2

.3

.4

.5
!!'
 "
)'( #
((!
"%
$'($#
(&$ (
(!#(
!&-
#*&

((+
 $
(
$) '
&!#$
$&$#($
$# ,
$)'($#
$&(!#
$#(&!
&!$((
 !+ 
#
 $
!*!#
$!)")'
+
$&
!( "$&
 # ##(
 "$#
"$#($#
 ((')&
#$)*&
#
$'
)'
#''
 (-
$'
#!'
 ##%$! '
' #($#
# #%$! '
 !!% 
#
&#' '$
/70/70
/.0/.0
/70/.0
/.0/70
a - North America 6
6
6
5
5
5

4
4
4

3
3
3

2
2
2

1
1
1

.2
.2
.2

.3
.3
.3

.4
.4
.4

.5
.5
.5
'!$
&(
&! #
-#-
! 
&)''!'
& '#
) !#
 #%$)&
($ $!"
!$)&#
$#
$#
$((&"

!! #($#
"'(&"
$(#)&
$%##
/.0/70
/.0/.0
/70/70
/.0/.0
/70/.0
/70/70
/.0/.0
/70/.0
/.0/70 b – Western Europe c - Oceania d - Asia

Figure 7-4 Global cities of the dataset categorized according to broad regions and the four modes of income/ethnicity association

220 | Analysis and findings gentrification, the urban regeneration projects promoted by the local government contributed to the mitigation of income inequalities (Forster, 2006; Arthurson, 2012) socially as well as spatially, and consequently reduced the associated ethnic segregation. However, as argued earlier, the predictable spatial outcome of that simple and abstract correlation between both types of segregation can be altered in the presence of other contingent factors involved in the complex process of spatial changes. As in the case of Melbourne, it is a typical Australian city, but there are two distinct characteristics caused ethnic segregation to drop despite of the increased socioeconomic segregation. Firstly, Melbourne is marked by intense gentrification that surpassed any other Australian city including Sydney (Weller and Van Hulten, 2012). Consequently, poor ethnic groups are displaced from inner-city neighborhoods (see chapter five) to other poor neighborhoods that are ethnically mixed or dominated by another ethnic group. Secondly, the rapid growth of transnational students in Melbourne has shifted neighborhood housing and services structure (Fincher and Shaw, 2009), since students are attracted to affordable and relatively small apartments in accessible locations (Forster, 2006) regardless their ethnic background or preferences. Which contributes to the decreased ethnic segregation.

7.3.2 North American cities

In the United States, as suggested by Massey (1990) and Massey and Denton (1993), racial segregation and poverty concentration are historically interrelated, which is clear in the popular term of ‘black poverty’. However, segregation of foreign-born immigrants and their socioeconomic status is less acknowledged in the literature, to the extent that several studies classify foreign-born population according to color lines instead of national or regional origins (i.e. Iceland and Scopilliti, 2008). Figure (7-4) shows that only 12 out of 33 (36.4%) American cities showed increase in both types of segregation, all of them are located in the north except for Tampa.

The color of poverty | 221

Despite the fact that Rustbelt cities receive significantly lower percentage of immigrants compared to Sunbelt cities (Frey, 2002), increased ethnic segregation is more common in the north where socioeconomic segregation prevails (see chapter five). Iceland and Scopilliti (2008) explain the segregation patterns of ethnic groups by their median income, where the higher the ratio of median income of minorities to that of whites, the lower the level of income inequality, and accordingly the lower the level of ethnic segregation and vice versa. For that matter, income inequalities in Rustbelt cities contribute to a parallel increase in ethnic segregation in those cities. Noticeably, other Rustbelt cities such as Boston and Baltimore did not follow Iceland and Scopilliti (2008) explanation of parallel increase in both types of segregation. Figure (7-4) shows that only 8 American cities witnessed divergent directions of change in both types of segregation. Clearly, since divergent changes are outcome of individual situations of housing and social structure on the local level, then no unified rationalization can explain spatial changes in those 8 cities at once, especially that cities do not only show variation in directions of spatial change toward more segregated or integrated urban structure, but they also show considerable variations in the intensity of spatial change. For example, the Texan cities of Austin and Dallas, both share Sunbelt cities’ qualities. However, the decreased socioeconomic segregation in Austin is accompanied by sizable increase in ethnic segregation. While, Dallas showed insignificant change in socioeconomic segregation accompanied by decrease in ethnic segregation. In fact, Austin is the only 1 American city where socioeconomic integration failed to promote parallel ethnic integration, which is uncommon in Sunbelt cities. As noted earlier, Austin’s distinct spatial outcome is due to the historically persistent segregation of the ‘East Austin enclave’, which is physically isolated by the interstate highway (I-35) built in the 1950s (Skop, 2009), the highway is passing through the city and limits the chance of the enclave to dissolve into the city. Furthermore, despite the fact that the

222 | Analysis and findings enclave receives large numbers of poor immigrants every year (especially Mexican), socioeconomic segregation did not increase accordingly, because the fast growing Mexican community has developed its own resources and networks, which support business of immigrants entrepreneurs to flourish (ibid.) and consequently lower levels of socioeconomic segregation. Similarly, spatial changes in other US cities can only be rationalized individually. In Canada, observations from table (7-4) indicate that 4 out of 6 Canadian cities show divergent directions of change in both types of segregation. This result opposes Fong and Shibuya’s (2000) claims discussed in chapter five, as they explain the reduced socioeconomic segregation in Canadian cities as a result of governmental actions that promote ethnic integration, and eventually reduce the associated socioeconomic segregation. In fact, Montreal is the only city that matches Fong and Shibuya’s assumption where the two types of segregation decreased simultaneously. While divergent segregation patterns in Toronto, Vancouver, Ottawa, and Calgary cannot be explained accordingly. The two neighboring and largest cities in Ontario, Toronto and Ottawa show the same paradox of Austin and Dallas. However, increased ethnic segregation in Toronto – compared to Ottawa – has always been perceived as a direct outcome of the exceptional growth rate of immigrants in Toronto. According to Statistics Canada (2006), Toronto greater area alone hosts 40% of all immigrants entered Canada from 2001 to 2006, compared to only 3% of immigrants who settled in Ottawa greater area. In addition to the elevated growth rate of immigrants in Toronto, other factors - such as median income of immigrants and their residential preferences - have shaped the complex segregation patterns in the city. Haan (2005) has traced homeownership preferences of 12 ethnic groups in Toronto; his analytical study concluded that Chinese and Italians have above average tendency to self-segregate through buying their homes near or inside their ethnic enclave. Which explains the decrease in socioeconomic

The color of poverty | 223 segregation associated with increased ethnic segregation, especially that Italians in particular have higher median income than that of Toronto in general.

7.3.3 Western European cities

In Europe, the association of socioeconomic and ethnic segregation is highly acknowledged in the literature. The simultaneous changes in both types of segregation in each city are based on one (or more) of three assumptions. Firstly, ethnic groups are more likely to belong to low-income groups (Bolt et al., 2009; Van Kempen, 2000; Massey and Denton, 1993). This assumption can explain the parallel increase/decrease in both types of segregation in several European cities, because any factor contributes to the change in one type of segregation will eventually lead to similar changes in the other. Such as the case of Rotterdam, where housing allocation rules are controlling levels of concentration of low-income households, who mostly belong to ethnic minorities. The second assumption suggests that poor ethnic groups tend to concentrate in public housing – as the case of Amsterdam (Musterd and Fullaondo, 2008), and Stockholm (Bråmå, 2008). This assumption can lead to different spatial outcomes of segregation overtime. Initially, if public housing is unevenly distributed in certain neighborhoods, ethnic minorities will also be concentrated according to public housing allocation. But over time, the recent privatization of public housing may (or may not) lead to the displacement and re-concentration of poor minorities, based on the extent and intensity of the privatization process, which explains the different spatial outcomes in Swedish cities. In Gothenburg, ethnic segregation decreased despite the increased socioeconomic segregation. As most minority enclave areas are dominated by public rental housing (Bråmå, 2008). Therefore, privatization of public housing may reduce affordable housing opportunities for the poor minorities, but at the same time it introduces high-income households to the enclave, and change the ethnic composition of the targeted

224 | Analysis and findings area. Also, in case of large-scale displacement of poor minorities, they relocate to other neighborhoods with affordable housing regardless their cultural preferences to live with co-ethnics, which can push them to look for housing in neighborhoods populated by another ethnic group. Eventually, limited opportunities of affordable housing may contribute to ethnic integration accompanied by higher level of poverty concentration. Thirdly, affluent ethnic groups tend to self-segregate. This assumption explains patterns of change in cities with decreased socioeconomic segregation associated with increased ethnic segregation. This trend of spatial change is common in British cities. According to Peach (2007), Indian population in British cities has maintained high level of segregation in spite of their high socioeconomic status. Unfortunately, the analysis cannot confirm this assumption due to lack of comparable data for British cities until the analysis process was performed.

7.3.4 Southeast Asian cities

Singapore and Hong Kong are the only two global cities with available data (on both types of segregation) in the East Asian region. As shown in chapter six, both cities have their special histories and qualities. On one hand, in Singapore, the enormous increase in ethnic segregation and the associated decrease in socioeconomic segregation can be simply explained by self-segregation of different ethnic groups (see chapter six for more details). However, the decreased socioeconomic segregation is relatively low (in terms of the intensity of the change) compared to the change in ethnic segregation, which indicates that other factors are involved in the complex process of spatial change beyond the issue of self- segregation. In fact, regrouping patterns of different ethnic groups in Singapore are supposed to produce higher level of socioeconomic segregation, because Malays – for example – have higher socioeconomic status and they tend to occupy higher-quality public housing units (Sin, 2002). While, Indians are more isolated in lower-quality public housing

The color of poverty | 225 units (ibid.). Accordingly, self-segregation of Malays and Indians can lead to a wider gap between income groups. However, allocation system of public housing units overcame the socioeconomic segregation resulting from ethnic regrouping. In fact, the HDB reformed its public housing allocation system in the late 1980s, by the aim of making it “more efficient and fair” (Tu, 1999: 104). According to Sin, “[T]he entire public-housing landscape in Singapore has been engineered in such a way as to yield low segregation scores.” (2002:434). Where each planning area has fair shares of all four types2 of apartments vary in sizes and prices (Singapore Department of Statistics, 2009). And since public housing hosts 86% of the population, therefore, the housing allocation system can have an enormous effect on segregation levels of the city as a whole. On the other hand, the case of Hong Kong shows no clear association between levels of ethnic segregation and the change in socioeconomic segregation. The reason behind this special case of disassociation is the absence of clear ethnicity lines between Hong Kong Chinese, and Mainland Chinese, who are the largest and only significant immigration group in Hong Kong. According to Cheung and Leung (2012: 2), “[B]ecause the ethnicity is the same, mainland migrants to Hong Kong seem to be indistinguishable by appearance, and they can permeate everywhere in Hong Kong without encountering segregation and exclusion.” Cheung and Leung also highlighted that the only visible differences between Hong Kong Chinese and mainland migrants are some cultural practices and slightly different Cantonese dialect. Notably, younger migrants who attend school in Hong Kong are more likely to overcome the language and culture differences and blend in totally in the host society. Another evidence for the absence of differentiation between Hong Kong natives and Mainland migrants is the census data collected by ‘Census and Statistics Department of Hong Kong”, where their classification of resident population is confined to only three categories: Chinese (including natives and Mainland migrant – 95% of total population), other Asian migrants, and other non-Asian migrants.

226 | Analysis and findings

Accordingly, any changes in socioeconomic segregation cannot be explained by the non-evident segregation of Mainland immigrants. Instead, socioeconomic segregation patterns in Hong Kong are highly dependent on other characteristics of the distinctive urban landscape of the city. Hong Kong is an island with a mountainous geography, it has very high density of population, high income inequalities, and its skyline is marked by high-rise buildings that reach up to 50 stories even in areas away from the central districts (Monkkonen and Zhang, 2011). Accordingly, the high density causes the issue of scale to be particularly relevant in defining levels of segregation, where pockets of high-income/high-rise gentrification emerging in scattered locations around the city will lead to decrease in socioeconomic segregation at least on the neighborhood scale, while the fragmented urban fabric is maintained on smaller scales. This assumption conforms to the results shown in figure (7-4), as the change in socioeconomic segregation index is measured on relatively large areas (district council/constituency areas level). Furthermore, other analytical studies of Hong Kong performed by Forrest et al. (2004) and Monkkonen and Zhang (2011) have yield similar results to the findings of this research. Forrest et al. concluded that high levels of income inequality is not reflected spatially in form of residential segregation, while Monkkonen and Zhang found that levels of socioeconomic segregation have actually fallen over the period from 1991 to 2006. Among other reasons, Monkkonen and Zhang (2011) explained lower income segregation by the fragmented urban redevelopment of older areas of the city by the private sector, which have brought the new high-rise buildings in close proximity to older residential stock inhabited by low-income households. Evidently, both cases of Singapore and Hong Kong show different models of income and ethnicity association than those found in Western Europe, North America, and Oceania. Where in Singapore, the association of Malay and Indian groups to distinct socioeconomic status had no impacts

The color of poverty | 227 on segregation levels due to the spatial distribution of public housing. While in Hong Kong, there is no evidence that this association does even exist.

228 | Analysis and findings

Summary

Cases discussed in this chapter suggest that part of the change in either socioeconomic or ethnic segregation in global cities is actually caused by the change in the other. Results reveal that 70% of global cities tend to have parallel increase or decrease in both types of segregation. The regression analysis showed a moderate correlation between the change in both socioeconomic and ethnic segregation, where 9.7% of change in ethnic segregation is explained by the change in socioeconomic segregation. Moreover, the generalized assumption that alpha cities are more prone to higher levels of segregation is not confirmed. Instead, beta and gamma cities showed higher tendency to have parallel increase in both types of segregation, while a higher ratio of alpha cities showed a parallel decrease in both types. Noticeably, divergent changes in socioeconomic and ethnic segregation could not be linked to certain global status. On the regional level, results show that Australian cities are more likely to have parallel increase in both types of segregation, followed by North American cities, then Western European cities. The level of spatial change produced by the association of an ethnic group to certain socioeconomic status varies significantly from region to region, and from city to city according to the local situations in each city. For example, intense gentrification in Australia, privatization of public housing in Europe, and lower median income of immigrants in Rustbelt American cities are factors contributing to the increase of at least one type of segregation, and consequently the increase/decrease in the other. In fact, this conclusion undermines the role of macro economic change as a main generator of divided socio-spatial outcomes within global cities. Where in several cases, the local contexts proved powerful enough to modify, intensify, or reverse spatial outcomes produced on higher levels of the urban hierarchy.

The color of poverty | 229

Finally, detailed studies of individual cities reveal distinct models of urban change that are inconsistent with the global city’s characteristics discussed in the literature. Austin, Brussels, and Hong Kong are just few examples for this controversy.

230 | Analysis and findings

Notes

1 The other American city showing similar spatial change is Columbus, however, the change in ethnic segregation is insignificant compared to Austin. 2 Price of public housing units is often defined in terms of the number of rooms. Types of housing units are: 1-/2- room flats, 3-room flats, 4-room flats, 5- room/Executive flats (Sin, 2002).

Chapter 8

8 The hybrid outcomes of urban change: conclusions and discussion

Throughout the chapters, the discussion of the socio-spatial transformations in global cities revealed certain generalization tendencies in the global city literature. The generalization includes; firstly, the changes in cities’ economic function in the global economy are decisive for other social and spatial changes occurring within these cities (Friedmann, 1986), where the agglomeration of advanced producer service firms in these cities is considered the key motivation behind urban change (Sassen, 1991; Castells; 2000; Taylor, 2000; Haussermann and Haila, 2005). Secondly, as a result of economic restructuring, occupational and income structures of global cities are altered towards more polarization (Friedmann and Wolff, 1982; Sassen, 1991), dualization (Mollenkopf and Castells, 1991), and inequality (Fainstein et al., 1992; Hamnett, 2003). Consequently, the intensified social division between different population groups within the city is spatially manifested in the form of high-income gentrification in inner city, spatial concentration of poverty, and spatial segregation based on race or ethnicity (Sassen, 2001; Massey, 1996; Goldsmith, 2000; Pacione, 2009). Thirdly, despite acknowledging the relevance of local situations in individual global cities to the understanding of the complex process of urban change (Burgers and Musterd, 2002; Van Kempen, 2007; Marcuse and Van

232 | Conclusions and discussion

Kempen, 2000), yet Friedmann and Wolff (1982), Sassen (2006), along with a wide range of studies performed over individual cities (i.e. Badcock, 1997), strongly suggest that economic restructuring and its consequent socio-spatial changes are evident in different cities around the world as a result of their integration into the global economy.

8.1 The assessment of the ‘global/divided’ city model

The key findings of the analysis process presented in chapters five, six and seven answer several questions about the validity of these three widely accepted assumptions in the global city literature. Briefly, the study was designed to capture the intensity and magnitude of change in spatial segregation levels in a large number of global cities, then, by correlating these changes to the city’s level of integration into the global economy. The propagated association between the processes of economic globalization on one side, and the increasingly fragmented socio-spatial patterns within global cities on the other side, is evaluated. Also, during the analysis process, the specific local situations that affect the spatial change in individual cities were discussed. The general observations have led to the belief that the general applicability of the ‘global city/divided city’ model on global cities around the world is overstretched. The analysis was performed over three stages; stage one confirmed that global cities do not share statistically significant tendency to be more segregated based on the socioeconomic status and ethnic background of their population. The regression analysis also confirmed that; the global network connectivity of cities could explain only 0.2% of the change in socioeconomic segregation and 0.2% of the change in ethnic segregation. While this percentage jumps to 5.1% when segregation types are explained by one another. In other words, part of the increased (or decreased) spatial segregation in global cities is more likely a result of the long-standing

Conclusions and discussion | 233 socioeconomic and ethnic structure, rather than the cities’ level of integration in the global economy. The second stage of the analysis revealed that certain characteristics of individual global cities could have an influence on the resultant change in spatial segregation in these cities. A one-way analysis of variance ANOVA was performed over different groups within the sample; the analysis results confirmed that the changes in segregation differ significantly among cities located in different geographic regions of the world. On the contrary, cities divided into alpha, beta, and gamma ranks didn’t show a significant difference in their changing segregation levels. Finally, in stage three, the discussion of the situation in individual global cities aimed to find an explanation for the lack of shared pattern of spatial change among global cities. The key findings are as follows: Cities of Amsterdam, Mexico City, and Sydney are seemingly supporting the validity of the general global/divided city model. Where the three cities are sharing the same global status as alpha cities, they are situated in totally different contexts, and all of them witnessed an increase in socioeconomic segregation (see chapter five) over the past decade. However, when compared to other alpha cities of Singapore, Miami, and Brussels, - which have actually witnessed more integration between socioeconomic groups over the same period of time – it can be argued that socio-spatial segregation is not necessarily the spatial outcome associated with economic globalization as promoted in the literature. Furthermore, it is to be noted that this conclusions is not based only on these 6 cities. Instead, this conclusion is supported by the lack of common spatial change shared by the large number of cities of the dataset (total of 91 cities to avoid misleading patterns suggested by smaller datasets). As mentioned, the results of the regression analysis deny any possibility that economic restructuring is associated with intensifying socio-spatial divisions within global cities. Even among alpha cities - which have higher ‘service value’ than beta and gamma cities, the results show no tendency for intensified

234 | Conclusions and discussion socioeconomic and ethnic segregation within these cities, while on the contrary, alpha cities – statistically speaking - were less likely to witness increase in segregation levels compared to gamma cities. Beside the generalized spatial outcomes, the generalization of a mechanism of the production of urban change was also challenged. The examples of Amsterdam, Mexico City, and Sydney reveal that even if these cities share similar spatial changes. Yet, a closer look on each case clarifies that these changes occur for different reasons and through different mechanisms. In more detail, the global/divided city model suggests that the declining role of the state will lead to privatization of services and subsidies cutbacks in general, including the housing sector (see chapter two for details). This might be true for Amsterdam, where the increased socioeconomic segregation is coinciding with the privatization of socially rented housing, which dropped from 57% of total housing units in 2001 to 47.4% in 2011 (Ministry of BZK, 2012). However, the same tendencies for privatization and declined provision of public housing are not evident in Mexican cities. On the contrary, the federal government initiated a housing finance system in the 1990s, and provided around 400,000 mortgages every year between 2000 and 2005 (Monkkonen, 2011). Yet, the allocation and size of the newly introduced housing projects, along with the inequitable selection criteria for those whom eligible for mortgage contributed to higher levels of socioeconomic segregation after all. Similarly, the generalized model of the global city also suggests that labor mobility contributed to the massive increase in flows of immigrants, and consequently contributed to the intensifying of ethnic segregation - due to the concentration of new comers in existing ethnic enclaves or affordable housing in poor neighborhoods in general. Yet, alpha cities of Singapore, San Francisco, and Melbourne suggest otherwise. Despite being magnets for immigrants, changes in ethnic segregation in those cities cannot be linked to labor mobility. In Singapore, foreign workers are legally banned from renting any residential properties on the market (Cho, 2011). Therefore, any

Conclusions and discussion | 235 increase in immigrants flow will not lead to the formation of any ethnic enclaves. Instead, ethnic segregation in Singapore is historically inherited, then intensified due to inefficient housing policies (see chapter six for details). In San Francisco, as in most Sunbelt cities in the USA, the huge in- flows of immigrants to the region (especially from Mexico) (Frey, 2002; Thompson, 2011) are not followed by intensified ethnic segregation. On the contrary, most Sunbelt cities witnessed significant decrease in ethnic segregation, especially when compared to Rustbelt cities that have higher tendency for increased ethnic segregation and much less flows of immigrants, this can be explained by the ‘super-diversity’1 (Vertovec, 2007) generated in Sunbelt cities by the enormous flows of immigrants. As, the existing enclaves are no longer able to accommodate all new comers to the city, which encourage them to look for housing opportunities elsewhere in the city. While in Melbourne, the in-flows of immigrants were also accompanied by decrease in ethnic segregation, due to the demographic characteristics of immigrants. As, Melbourne attracts particularly large numbers of transnational students (Fincher and Shaw, 2009), who their preferences are based on affordability and accessibility rather than the ethnic composition of the desired neighborhood. Besides, intense gentrification in inner Melbourne contributed to the decreased ethnic segregation, due to the displacement of poor immigrants, and their concentration in other neighborhoods that are already dominated by other ethnic groups. Accordingly, labor mobility is not necessarily responsible for the complicated patterns of ethnic segregation in global cities. Increased segregation in Singapore is not even remotely related to labor mobility, while despite the evident flows of immigrants to San Francisco and Melbourne, levels of ethnic segregation dropped significantly in both cities.

236 | Conclusions and discussion

8.2 The production of spatial change in global cities

In view of the above, the lack of a general trend of change in spatial segregation levels among global cities of the dataset does not only challenge the acknowledged association between economic restructuring and the intensified socio-spatial divisions within cities, but also, the findings support a point of view that criticizes the global city literature for the oversimplification of the global/local interplay through which the complex process of urban change is taking place. In more detail, the only explanation for how cities are expected to share common trends of urban change despite their local differences is because local situations in individual cities are rendered irrelevant to the outcomes of urban change. For example, the polarization thesis acknowledges that global impacts on local settings are channeled through different scales of the urban hierarchy. It is believed that transnational institutions (global level) are downscaling central governments2 (nation-state level), which will impact urban policies (on both national and metropolitan level), to finally contribute to increased spatial segregation (on neighborhood level). For that matter, the problem with this logic is not the overlooking of the intermediate scales between the global and the local. Instead, the problem is the overlooking of the imprint each level can leave on the process as a whole. In other words, the thesis simplifies the restructuring process through the standardization of the influence of global forces and neutralizing any possible influence of national, metropolitan, or local particularities. Therefore, the lack of a shared pattern of spatial change among global cities challenges the assumptions that local situations are irrelevant, as the divergent changes in individual cities are only explained through acknowledging the important and relevant role of local situations on the outcomes of urban changes, and consequently, the findings challenges the over-simplistic view of the global/local interplay, which renders powerful global processes that modify local settings within cities in an unproblematic

Conclusions and discussion | 237 manner. Accordingly, the complex interaction between macro processes of globalization and micro situations in individual cities needs to be approached from another perspective that includes the different factors involved in the process of urban change in general, and involved in the production of distinct and non-generalizable spatial outcomes for individual global cities in particular. I don’t claim that this research offers such well-formulated alternative perspective for understanding the complex dynamics of urban change in global cities. Instead, this chapter highlights the theoretical and empirical issues that need to be addressed in future research to finally reach a framework for the systemic study of urban change in a global context.

8.2.1 Urban change and the concept of ‘hybridity’

One of the possible explanations for the observed divergent outcomes of spatial restructuring in global cities is that the potential impacts of the new functions of global cities are exaggerated, as Pretecille (1994) argues, the services industry may represent only a small part of urban employment, and accordingly, the occupational and income structures of global cities are not expected to change dramatically as a result of the cities’ integration into the global economy. In this case, a considerable part of the recent social and spatial changes in global cities is triggered by several factors that are not necessarily economic or global in reach. As in the case of Mexican cities, Guadalajara is a gamma city, Monterrey is a beta city, and Mexico City is an alpha city, yet in all three cases, the drastic intensification of socioeconomic segregation has nothing to do with these cities’ position as global cities. Instead, the change in segregation is explained by the characteristics of Mexico’s federal housing finance system (see chapter five for details). However, this perspective - which considers the influence of global economic changes on cities to be negligible - fails to fully explain the more complex situations in cities like Amsterdam, Sydney and other cases where the impacts of global economic changes on other social and spatial facets of

238 | Conclusions and discussion restructuring is more pronounced. Therefore, another systemic explanation is required in order to capture the complexity of the process of restructuring of global cities. It is already established that the generalization of a certain outcome of spatial restructuring to be applicable in global cities around the world is not valid, neither is the total elimination of the possibly influential role of global economic functions of cities on the outcomes of restructuring. Accordingly, the aim of this section is to reach ‘the middle ground’, and to introduce a potential systemic explanation that avoids any generalized assumptions by capturing the complexity of the restructuring process based on the findings of the analytical study. Briefly, the cases presented in chapters five, six and seven reveal that the spatial outcomes in global cities are a product of the ‘interaction’ between global economic functions with the existing contextual and historical particularities that varies from city to city. The nature and main characteristics of this interaction are what our proposed theoretical construct is attempting to capture. As noted in chapter two, one way to describe this ‘interaction’ is Burgers and Musterd’s (2002) three-layered model that refers to the contextual variables as the layer through which the global changes are ‘mediated’ to the local level. However, the term ‘mediate’ here implies that the direction of the change is from a top layer (the global economic change) to the bottom layer (the level of the locality). Accordingly, despite the fact that Burgers and Musterd’s model effectively explains different mechanisms of growing inequalities in post-industrial labor markets, yet the same idea of the top-down layers of change may not be suitable to describe the spatial change in global cities in general, as this model refers to the local level as a receptor for the change impinged on it by macro process that take place on higher scales of the urban hierarchy instead of acknowledging the mutual interaction between the global and the local. Other concepts in the academic literature – such as glocalization and hybridity – are one step closer to describe the nature of this global/local interaction and its outcome.

Conclusions and discussion | 239

In another publication (see Ismail, 2013), I argued that the concept of hybridity – which is widely used in cultural studies – could be extended to describe the process (and outcomes) of spatial changes taking place in urban areas as a result of globalization. The notion of hybridity originated in biology with the development of Mendelian genetics in the 1870s, which refers to the process of ‘mixing’ of genetic traits, as the resulting hybrids typically have intermediate traits of both geneses involved in the hybridization process (McCarthy, 2006). Then, the term was extended to describe the mixing of ‘cultural’ forms as a result of increasing global interconnectedness (Eriksen, 2007). Hybridity is also used to explain other types of ‘mixing’ of institutional (de Ruijter, 1996), organizational (Oliver and Montgomery, 2000), and structural forms of social organization (Nederveen Pieterse, 2001). Since global processes and local contexts are not exactly ‘idealtypical’ (Brandsen et al., 2005) entities that can be tangibly mixed to produce a hybrid outcome. Therefore, approaching such global/local interplay - from hybridity perspective - has to capture the in-between structural practices that influence one another on different hierarchal scales, then by verifying that the spatial outcome of this interplay is actually marked by traits inherited from different practices (including those on the local level), only then can the concept of hybridity be broaden to describe geographical features produced from the interaction of non-hegemonic global forces and non- transparent local contexts. The analytical study has already eliminated the possibility that spatial outcomes are defined solely by cities’ integration in the global economy, while at the same time, the detailed discussion of individual cases reveals a number of regional, national, metropolitan, and local particularities that are clearly influencing the intensity and direction of spatial changes in global cities. To finally assert that spatial changes are outcome of complex processes of structural and chronological hybridity

240 | Conclusions and discussion

a. Structural hybridity

Nederveen Pieterse explains that globalization in structural terms means “the increase in the available modes of organization: transnational, international, macro-regional, national, micro-regional, municipal, local” (1995: 50). In this sense, the global process of structural hybridization refers to the formation of a hybrid outcome as a result of the interaction of social, institutional, and economic process across the different scales of the urban hierarchy. He adds “hybrid formations constituted by the interpenetration of diverse logics manifest themselves in hybrid sites and spaces.” (p. 51). Global cities themselves are an example of such hybrid spaces, as they surpass the national scale in their importance to the global economy (Sassen, 2006), while at the same time, they are still a place of the local, where the global spaces of firms and gentrified neighborhoods are juxtaposed by the deprived neighborhoods and ethnic enclaves. In order to describe the recent spatial changes in global cities as a form of structural hybridization, the global city literature acknowledges the fact that the global impacts on local settings are channeled through different scales of the urban hierarchy. As mentioned earlier, it is believed that transnational institutions (global level) are downscaling central governments3 (nation-state level), which will impact urban policies (on both national and metropolitan level), to finally contribute to increased spatial segregation (on neighborhood level). In the light of the findings of the analytical study, all these processes on the different scales are affecting the resultant patterns of spatial segregation in terms of the direction and the magnitude of change. For that matter it is no longer possible to overlook the influence of the intermediate processes on the multitude of scales between the global and the local, through standardizing the impact of global forces and neutralizing any possible influence of national, metropolitan, or local particularities. Cases discussed in previous chapters show that, for example, public housing is

Conclusions and discussion | 241 being privatized in some cities, while new public housing provision programs are being implemented in other cities despite their high integration into the global economy. In this case, the geographical distribution of public housing units and the accessibility of the program (by disadvantaged population groups) will outline the maps of spatial segregation in unpredictable manner. Similarly, urban regeneration projects are not necessarily leading to gentrification (and its consequent displacement/ reconcentration of the poor), few examples show that urban regeneration projects are still able to achieve improvements in residents’ well-being despite the profit-based atmosphere clouding the decision-making process. Furthermore, on neighborhood level, homeownership rates, housing market flexibility, and residents’ preferences are strong modifiers of residential mobility and hence spatial segregation. One example is the preferences of affluent entrepreneurs who belong to an ethnic minority to reside in the (usually poor) ethnic enclave near to their ethnic business; their choice will probably contribute to higher level of ethnic segregation, yet socioeconomic segregation can decrease accordingly. Accordingly, it can be argued that the spatial change in global cities is a structural hybrid, formed by the mélange of the global, national, municipal, and local processes and particularities in individual cities.

b. Chronological hybridity

The metaphor of the ‘layered city’ (Massey, 1984; Kesteloot, 2000; 2005) portrays how the socio-spatial structure of any city is a product of historical processes occurring in successive rounds of capital accumulation, each round has its own arrangement of economic sites and residential areas deposited in layers one upon another across geographical space. While, each layer represents distinct spatial arrangement of economic and residential functions that belong to a certain round of accumulation, still, spatial features from previous layers may still be present in recent layers of spatial development.

242 | Conclusions and discussion

Similarly, the findings of the analytical study reveal that the patterns of spatial segregation in recent years are defined by both the historical patterns of spatial segregation, as well as the current economic, political, and social particularities in individual cities. In this sense, it can be argued that spatial outcomes in global cities are a ‘chronological hybrid’. As in the case of structural hybridity, the chronological hybridity refers to the formation of hybrid from the interaction of different processes, but instead of the interaction of these processes across different scales, the processes are taking place across the historical rounds of capital accumulation. In other words, the recent spatial changes in global cities are a product of the interaction between the historically inherited spatial forms and present-day dynamics (Soja, 1985; Kesteloot, 2000). For example, patterns of ethnic segregation in global cities are not defined solely by recent flows of international migration and the increased labor mobility. Instead, concentrations of new comers to the city are highly dependent on already existing ethnic enclaves that represented nuclei for the new comers to settle around. Therefore, labor mobility alone cannot explain present concentrations of Turkish population in German cities nor the Indian population in British cities, because the first is formed by historical migration waves since WWII, while the later is an outcome of historical ties formed in the colonial era (Peach, 2002; Massey et al., 2008). If it has any influence, labor mobility merely contributes to the altering of existing conditions that are accumulated over the years to produce a hybrid outcome, instead of imposing a non-rooted spatial outcome. To sum up, since the process of economic restructuring alone is proved inadequate to explain the complex process of urban change in global cities, then it can be argued that by acknowledging the influential role of local contexts, and through perceiving the global/local interplay as processes of structural and chronological hybridization, we can reach a better understanding of the mechanisms through which the divergent patterns of spatial change in individual cities are produced. The historically uneven

Conclusions and discussion | 243 development in Brazil, social inclusion programs in Johannesburg and Adelaide, housing allocation restrictions in Rotterdam, geography of Hong Kong, and ethnic composition of Singapore are just few examples of how the spatial transformations in global cities are a product of both global processes mélanges with historical, political, cultural, and geographical particularities across the national, metropolitan, and local contexts of each city. To finally produce a ‘hybrid’ spatial outcome; hybrid in terms that the outcome carries distinct qualities that can be traced back to all the modifiers exist – structurally - on different levels of the urban hierarchy from global to local, and – chronologically - on different layers of previous developments. One example of this ‘hybridization process’ is the gentrification of poor neighborhoods in both contexts of Latin American cities, and the city of Hong Kong. In both cases, as well as in many other European, Australian, and American cities, the demand for attractive housing conditions has increased. However, instead of the large-scale gentrification process that took place in Melbourne or Montreal for example, the gentrification process in Latin American cities was affected by the already existing informal settlements that surround urban areas due to the massive urbanization of the region since the 1950s. The mélange of gentrification (as a global process (Atkinson and Bridge, 2005)) and historical patterns of urbanization, produced what is known as the ‘gated communities’ (Feitosa et al., 2007; Monkkonen, 2011), which is technically gentrified areas4, but they are isolated by walls and secured by monitored entrances due to fear of crime and violence in the surrounding neighborhoods (Caldeira, 2000; Coy and Pohler, 2002). Similarly, the demand for attractive housing conditions in Hong Kong was met by the gentrification of deteriorated neighborhoods. Yet, due to the mountainous topography of the island of Hong Kong, its very high density of population, and its limited urban space (Monkkonen and Zhang, 2011). The gentrification process took the shape of high- income/high-rise buildings - in scattered locations in the city, and in a close proximity to older residential stock inhabited by low-income households. In

244 | Conclusions and discussion this sense, despite the fact that gentrification is a global process that occurs in cities around the world. Yet, it can be argued that, local situations – such as historical layers of urban development in Latin America, or the topography of Hong Kong – are able to bend the course of the gentrification process and produce a unique and hybrid type of gentrified areas that reflects the interaction of the global process of gentrification with the specific characteristics of every context.

8.3 So what? A research agenda for the systemic study of spatial segregation in global cities

This research aimed to answer several research questions, the main question as described in chapter one; “is the global city a divided city?” According to the theoretical discussion and the findings of the analytical study, it is argued that global cities are not necessarily divided cities; instead, it is evident that global cities vary greatly in the direction and the magnitude of change in their spatial segregation levels. Accordingly, the findings shed the light on another research problem that needs to be addressed in the future. Concisely, it can be argued that the available research lacks the ability to theoretically and empirically explain the complex dynamics of the spatial restructuring of global cities. The problem is manifested in the generalization of a number of assumptions to be applicable on global cities around the world, which indicates an oversimplification of the global/local interplay in the specialized literature. Also, the problem emerges from the evident overlooking of the possibly influential particularities of different local contexts on the process and outcomes of spatial restructuring. Briefly, in the light of this research problem, the objectives of the future research should include; the development of an all-encompassing research approach, or an alternative hypothesis that overcomes the shortcomings of the available research, as well as, the development of a comprehensive

Conclusions and discussion | 245 research methodology for the systemic study of spatial segregation in global/local context. An example of such framework for the cross-city comparison of spatial segregation patterns in global cities is proposed later in this section. Unfortunately, due to several constrains, this study could not offer such alternative hypothesis. Instead, the study examined the validity of the assumptions of the global city literature and reached the conclusion that these assumptions require to be revisited. For that matter, the future research should overcome the shortcomings of this study through: (1) enhance and update the dataset for the empirical study, (2) an in-depth investigation of the most influential contextual particularities, as not all local factors are proved to be relevant to the situation in individual cities, and (3) answer these pressing research questions; how to approach the restructuring process of global cities in a way that capture its complexity? What are the local particularities that need to be included in the study of cities’ restructuring? And how to develop the concept of hybridity to assist in the understanding of the evolving patterns of spatial segregation in global context?

8.3.1 Towards a heuristic model for the comparative studies of spatial segregation

This sub-section presents briefly an attempt to develop a framework for the comparative studies of spatial segregation in global cities. The aim here is to highlight a number of local factors that affect the course of change in spatial segregation, while at the same time, spot the possible variations in these factors from context to context. Briefly, the discussion in chapter two highlighted a number of ‘aspects’ (Burgers and Musterd, 2002) or ‘local contingencies’ (Marcuse and Van Kempen, 2000), by which the impacts of global economic change on different localities are explained. These aspects include historical and subcultural differences, as well as other variations in national policies, geography, and existing levels of inequality. Yet, another conclusion drawn

246 | Conclusions and discussion from the analytical study suggests that these aspects are too general to describe and to compare patterns of spatial segregation in individual global cities. In other words, the comparison of socioeconomic segregation patterns that are a result of – for example – the change in public housing provision in two different cities, cannot be based solely on the number or ratio of public housing units provided by the state in each city. Instead, the comparison should include other factors, such as the allocation of public housing units (are they concentrated in low-income neighborhood, or distributed across the city in a close proximity to middle and high-income households?), who has access to the service? (low-income groups, poor immigrants, the unemployed), and patterns, as well as, the pace of privatization of public housing units (if any). All these variations of a single contingency, that is public housing provision, can lead to remarkably divergent changes in segregation levels between two cities, even if both cities are witnessing a parallel decline in shares of public housing stock. Still, issues of allocation and access are able to define the resultant segregation patterns in the two cities. Similarly, based on the analysis findings, the possible variations of other relevant local contingencies can be fruitfully developed into an explanatory model for analyzing the changing patterns of spatial segregation, especially from a comparative perspective. The model compares cities along four main axes, while each axis includes a set of variables, which represent the range of contextual differences evident from city to city that are most likely related to change in the segregation level. The four axes are: (a) Gentrification, (b) Urban policies, including governmental actions, regulations, and public housing provision, (c) The mutual correlation between socioeconomic and ethnic segregation, and (d) Other distinctive characteristics of the cases under study.

Conclusions and discussion | 247

a. Gentrification

Gentrification as a global phenomenon (Atkinson and Bridge, 2005) is considered a direct outcome of the rise of service industries, while at the same time, it contributes to higher level of socioeconomic segregation by displacing the poor and creating citadels for the rich. This standardized and flat perception of gentrification has different meanings when situated in the context of specific cases. Gentrification patterns in the alpha cities of Hong Kong, Brussels, Amsterdam, and Sydney suggest that the ‘scale’ of gentrification is an important variable that is linked to certain changes in segregation levels. In more detail, the small-scale gentrification projects in both Brussels (Van Criekingen and Decroly, 2003), and Hong Kong (Monkkonen and Zhang, 2011) have contributed to the overall decrease in socioeconomic segregation levels in both cities, and vice versa, where the large-scale gentrification in Sydney and Amsterdam are contributing to the overall increase in segregation levels. Another variable related to the ‘scale’ of gentrification, is the pace and intensity of displacement of the poor and immigrants from inner city, where the larger the scale of gentrification, the larger the number of households who have to relocate. Also, the greater the difference in socioeconomic status between the poor residents and the gentry group, the starkest the upgrade in the built environment (Clark, 2005), and eventually, the higher the pace of displacement of the poor from gentrified areas. Similarly, the process of reconcentration of poor and immigrants households - that follows their displacement from gentrified areas – varies from city to city according to the distribution of affordable housing on the city level, as well as the presence and the efficiency of housing regulations that prevent the concentration of poverty and counteract the drawbacks of gentrification. As in the case of Rotterdam, the strict housing allocation rules set by local government in 2003 by the ‘Rotterdam zet door’ action program (Kleinhans, 2004), have successfully controlled the spatial distribution of low-income households and prevent their reconcentration in disadvantaged

248 | Conclusions and discussion neighborhoods. While in Amsterdam, the lack of such effective regulations have led urban development projects to contribute to the reconcentration of displaced households in other poor areas elsewhere in the city (Bolt et al., 2009). In view of the above, the gentrification process in general does not necessarily lead to higher levels of socioeconomic and ethnic segregation. Instead, the different scales of gentrification projects, the pace of the associated displacement of the poor and immigrant households, and the regulations that control the reconcentration patterns of the displaced households are the variables that need to be taken into account, especially when discussing the implications of gentrification process on spatial segregation level in cities in different contexts.

b. Urban policies, regulations, and public housing provision

It is acknowledged in the specialized literature that state regulations and housing policies can promote either higher social mix or higher degree of segregation on neighborhood levels (Kazepov, 2005; Roberts and Wilson, 2009). As, state regulations have the ability to prevent/permit both poverty and affluence concentration. Yet, it is also acknowledged that economic globalization undermined the ability of local government to impose regulative decisions on local markets (Hancher and Moran, 1989), through different trends such as deregulation, privatization, diminished public spending ...etc. (Beall, 2002; Held, 2004). As a result, the local housing market became increasingly deregulated, social housing provision declined, and existing social housing are being privatized or demolished to clear room for profitable housing projects. Accordingly, changes in local housing markets in global cities are expected to contribute to higher level of segregation. However, it is already established that generalization of certain consequences associated with economic globalization is sort of an exaggeration, because local situations in individual cities is highly influential on the outcomes of cities’ restructuring. In this sense, the

Conclusions and discussion | 249 presence of an efficient government intervention in individual cities can significantly resist segregation tendencies, or even alter existing patterns of segregation. In order to evaluate the influence of local urban policies as an important contingency affecting the course of urban change, the analytical study aimed to correlate levels of segregation to three main aspects of urban policy: housing regulations, urban development initiatives, and public housing provision. The results suggest that presence of effective housing regulations, as well as urban development initiatives has a noticeable mitigating impact on the spatial segregation level. While, the change in public housing stock – as discussed above – affect the change in segregation levels differently from city to city, based on several variables such as the allocation of public housing units, and the access of the disadvantaged groups to the service. Cases of Singapore and Rotterdam clearly reflect the influence of housing regulations on segregation levels. In Singapore, the households’ allocation rules - imposed in the 1960s - have effectively contributed to the decreased ethnic segregation in the 1970s and early 80s (Van Grunsven, 2000). Then, in 1989, the government imposed the ethnic quota system (see chapter four) to control the voluntary residential mobility of minorities (Sin, 2003). However, the government ineffective pursuance of the quota system has led eventually to the re-grouping of ethnic groups. In this sense, effective and properly implemented housing regulations are expected to contribute to lower levels of segregation. This conclusion is also supported by the case of Rotterdam as discussed above. Similarly, Johannesburg, Adelaide, and Rio de Janeiro showed special cases of socioeconomic integration in highly segregated regions. The decreased socioeconomic segregation in these cities is explained by the presence of governmental intervention aiming for improving housing and infrastructure of deteriorated neighborhoods, as well as promoting social inclusion of the poor and minorities. Apparently, black empowerment and

250 | Conclusions and discussion poverty reduction programs in Johannesburg, urban development initiatives in Rio de Janeiro and urban regeneration projects in Adelaide (see chapter five) succeeded in reducing socioeconomic segregation significantly despite the process of economic restructuring witnessed in all three cities. In this sense, it can be argued that the absence of the same governmental actions in other cities has contributed to their rising levels of residential segregation due to the unchallenged social exclusion. Finally, the debate over the relevance of public housing (in terms of size and distribution) to levels of residential segregation can be summarized as follows: On one hand, it is assumed that quality, size, and distribution of public housing are highly relevant to levels of segregation (Van Kempen, 2005). As, public housing provides affordable housing opportunities for the deprived population. So, the less the share of public housing units in the market, the less housing opportunities for the poor. Also, uneven distribution of public housing units in certain neighborhoods can lead to concentration of poverty. The analytical study revealed that Swedish, Australian, and Mexican cities support these assumptions. Where the declined provision of public housing in Sweden (Hedin et al., 2012) and Australia (Arthurson, 1998; Badcock, 1999), (as well as the demolition of deteriorated public housing dwellings) are coinciding with the increase in both ethnic and socioeconomic segregation in Swedish cities and most of Australian cities. While in Mexico, public housing provision remained steady. Yet, the distribution of public housing units contributed to segregation (Monkkonen, 2011), due to the concentration of people with similar socioeconomic attributes in certain neighborhoods. While on the other hand, it is also assumed that housing policies are not necessarily related to lower or to higher levels of segregation (Musterd and Fullaondo, 2008). This assumption is supported by cases of Brussels, Barcelona, Berlin, and Singapore. Where, ethnic integration in Barcelona and socioeconomic integration in Brussels occurred anyway, despite the

Conclusions and discussion | 251 documented small shares of public housing in both cities. In Berlin, the reduction of public housing stock - from 30% of total housing units in 1990 to 15% in 2008 (Aalbers and Holm, 2008) - is not associated with increased socioeconomic segregation as the case of Swedish cities. Instead, Berlin showed insignificant change in levels of socioeconomic segregation. While in Singapore, 86% of the population resides in public housing dwellings (Sin, 2003). Yet, the increase in ethnic segregation is exceptionally high despite the fair distribution and high quality of public housing in Singapore. To sum up, contingency of local urban policies is proved to be – to some extent - relevant and influential on the course of spatial changes in global cities. As, housing regulations and inclusion policies can have a great impact on the accessibility of low-income households to decent housing opportunities, the prevention of poverty concentration, and the improvement of social cohesion. It is also concluded that size and distribution of public housing stock may or may not be related to a certain change in spatial segregation levels in individual cities.

c. The mutual correlation between socioeconomic and ethnic segregation

As shown in chapter seven, the intersection between maps of socioeconomic and ethnic segregation in the same city suggests that both type of segregation are mutually affecting one another. Where, 70% of cities of the dataset support this statement by showing a parallel increase (or decrease) in both socioeconomic and ethnic segregation. For example, in the case of Toronto, the decreased socioeconomic segregation is explained and only understood by the tendency among affluent Italian immigrants to cluster near or inside their ethnic enclave. As a result, the ethnic background and preferences of a certain population group resulted in the change in socioeconomic segregation. Similarly, increased ethnic segregation in Rustbelt cities in the USA is explained by the low median income of minorities compared to that of whites. As a result, the limited opportunities

252 | Conclusions and discussion of affordable housing for the poor minorities have contributed to their concentration in poor and deteriorated neighborhoods, while leading levels of ethnic segregation to rise accordingly. In this sense, comparing ethnic segregation levels in two cities needs to incorporate the socioeconomic status of the segregated ethnic minority as an important contingency and vice versa.

d. Other distinctive characteristics of the cases under study

Other factors that need to be considered in comparative studies of spatial segregation may include special and distinctive characteristic of a certain city. On the regional level, historically uneven economic development between cities of the same geographic region may affect the recent changes in spatial segregation levels. Where, historically uneven economic development between Rustbelt and Sunbelt cities in the USA, or between Sao Paulo and Rio de Janeiro in Brazil (see chapter five for details) is reflected in the tendency for Rustbelt cities, as well as Sao Paulo, to be more segregated. Apparently, impacts of uneven industrialization - since the 1940s - are still visible in the two contexts. On level of individual cities, distinctive characteristics of Hong Kong and Austin have noticeably impacted levels of spatial segregation in both cities. Despite being an alpha city, Hong Kong witnessed a relative retreat in foreign immigration in 2000s with the outbreak of H5N1 avian influenza, followed by SARS (severe acute respiratory syndrome). Also, the handover of Hong Kong to the Chinese authority in 1997 sparked the flows of Mainland Chinese into Hong Kong. As a result, local circumstances of Hong Kong contributed to less ethnic segregation in the city despite its global status. Likewise, gamma city of Austin receives fewer immigrants than other Sunbelt alpha cities such as Miami and San Francisco (Frey, 2002). Yet, Austin suffered an increase in ethnic segregation due to the historically persistent ‘East Austin enclave’, which is physically isolated by the interstate highway (I-35) built in the 1950s (Skop, 2009). Accordingly, it is

Conclusions and discussion | 253 clear that patterns of segregation in Austin and Hong Kong are strongly linked to the distinct characteristics of each city rather than its global status. For that matter, paying enough attention to the distinctive characteristics of different cities can assist in explaining the seemingly inconsistent changes in segregation levels showed by individual global cities. Finally, this sub-section underlines the remaining local contingencies that were not included in the comparative model, because they are either proved to be irrelevant to the process of spatial segregation, or its relevance is not confirmed based on the findings of the analytical study. On one hand, Marcuse and Van Kempen (2000) highlight the contingency of globalization to be influential on the spatial transformations in contemporary cities. In their view, the contingency of globalization is defined as; the city’s “position in the processes of globalization” (p: 268), where the processes of globalization are not limited to the concentration of advanced producer service firms in the city, but the processes also include other aspects such as the importance of international trade and extent of technological development. Accordingly, Marcuse and Van Kempen perception of globalization as a multi-dimensional process has led to their conclusion that spatial transformation in global cities are dependent on cities’ position in the processes of globalization. Which is not necessarily the case when a city’s global status is primarily determined based on the size of producer service firms hosted by the city. In fact, the results show no association between levels of segregation (both socioeconomic and ethnicity) and cities’ global status. On the contrary, each of the alpha, beta, and gamma categories had its share of cities with various directions and magnitudes of change in residential segregation. Furthermore, the discussion of the spatial transformations in individual cities did not pinpoint cities’ global status as a contingency that affects levels of segregation in anyway. Instead, there are numerous examples of alpha cities that show contrasting transformations in terms of the direction and magnitude of change in segregation level despite the fact that these cities share the same global

254 | Conclusions and discussion status as alpha cities such as Amsterdam, Brussels, Johannesburg, and Los Angeles to name a few. On the other hand, existing patterns of racial and ethnic segregation is another local contingency stated by Marcuse and Van Kempen. The discussion in chapter seven - and in the previous sub-sections - reveals that the existing patterns of ethnic segregation are highly relevant to the production of distinct patterns of socioeconomic segregation in individual cities. While in case of ethnic segregation, the historically persistent patterns of ethnic segregation are not associated with a recent intensification of ethnic segregation. For example, almost 60% of Western European cities have witnessed a significant decrease in ethnic segregation despite the persistent post-war ethnic enclaves in those cities. Accordingly, the relevance of existing patterns of ethnic segregation as an influential contingency is not confirmed. In more detail, the only case where the size and location of the existing ethnic enclave is proved relevant to recent patterns of ethnic segregation is when preferences of self-segregation are particularly high, as in the case of Indian population in Manchester, or the voluntary ethnic regrouping in Singapore. Other than that, the persistence or the dismantling of the existing enclave is determined by other factors such as the process of gentrification, where the displacement of poor immigrants from inner city is followed by either their concentration elsewhere, hence, ethnic segregation level is expected to rise, or their dispersal in other low- income and ethnically-mixed neighborhoods, which leads ethnic segregation level to drop. Similarly, the size of in-flows of immigrants to the city determine whether existing patterns of ethnic segregation is sustained or not, because when the flows of immigrants are large enough to surpass the capacity of the existing enclave to accommodate the new comers, immigrants start to search for affordable housing in other neighborhoods, and the resultant patterns of ethnic segregation in this case is also determined by either the concentration or the dispersal of immigrants outside of the existing enclaves, not by the presence of the enclave itself. Note that

Conclusions and discussion | 255 patterns of concentration of immigrants are also dependent on other aspects such as housing regulations and allocation schemes that vary from city to city. To sum up

The main contributions of this research are the in-depth investigation of the complex process of cities’ multifaceted restructuring, its motivations, outcomes, and dynamics. Also, the research managed to overcome a number of shortcomings of the available research by providing an empirical study - performed over a large dataset - for investigating the presumed socially and spatially divided nature of global cities. The research concluded that the widely acknowledged global/divided city model is insufficient to capture the potential influence of local contexts of individual cities on the outcomes of spatial changes, as the model oversimplify the global/local interplay by marginalizing the influence of the distinctive characteristics of individual cities on the local level. Therefore, the research offered a research agenda for future research; its main objective is developing an alternative theoretical construct for interpreting the socio-spatial transformations in global cities. The proposed construct should acknowledge the complexity of the global/local interplay through perceiving the process of urban change as a form of structural and chronological hybridity (see Ismail, 2013). Where, the spatial transformations in global cities are a product of both global processes mélanges with historical, political, cultural, and geographical particularities across the national, metropolitan, and local contexts of each city. To finally produce a ‘hybrid’ and unique spatial outcome; hybrid in terms that the outcome carries distinct qualities that can be traced back to all the modifiers exist – structurally - on different levels of the urban hierarchy from global to local, and – chronologically - on different layers of previous developments. A clear example for such hybridity is the case of Hong Kong, where segregation patterns are an outcome of complex interaction of (1) the city’s economic functions in the global economy, (2) the political milestone of the

256 | Conclusions and discussion handover of Hong Kong to the Chinese authority in 19975, (3) the natural outbreak of H5N1 avian influenza, then the epidemic of severe acute respiratory syndrome (SARS) in 20036, (4) the city’s demography and its high population density, and finally (5) the mountainous topography of the island. All these factors have contributed to the distinctive urban characteristics of Hong Kong including its skyline, social structure, ethnic composition, and its spatially fragmented urban fabric. Similarly, each global city has its own sets of ‘modifiers’ that cannot be overlooked, because these modifiers are proved to be an essential part of the complex global/local dynamics. The absence of standardized spatial outcome for global cities affirms that it is no longer possible to perceive the global/local interplay as a unidirectional cause-effect relation, in which the global processes are expected to reshape local settings of cities. Instead, context matters, history matters, and both are reflected on the outcomes of cities’ restructuring, even if this restructuring is taking place under the seemingly prevalent conditions of globalization. Moreover, the study also developed a scheme for studying the changes in socioeconomic and ethnic segregation within global cities, especially from a comparative perspective. The proposed scheme is based on a number of relevant and highly influential local contingent factors that are drawn from the discussion of a large number of individual cases throughout the chapters. The scheme is an attempt to facilitate the cross-city comparisons of the motivation behind spatial change in different local contexts. Finally, although the study could not provide a detailed discussion of the situation in all individual cities of the dataset (due to either lack of data or other time constrains), still, unexplained patterns of spatial change in cities and regions are to be the focus of an upcoming research that aims to develop the argument presented in this study. Overall, the study acknowledges the complexity of the urban changes taking place in contemporary cities. Generalized or not, the transformations in any city are an outcome of a dynamic process that can never come to halt. Yet, the nature

Conclusions and discussion | 257 of these transformations, as well as their motivations, signifies that each city is unique.

258 | Conclusions and discussion

Notes

1 Vertovec uses the term super-diversity to describe the unprecedented level of diversity witnessed in Britain’s demographic and social patterns. He explains, “Britain can now be characterized by ‘super-diversity,’ a notion intended to underline a level and kind of complexity surpassing anything the country has previously experienced. Such a condition is distinguished by a dynamic interplay of variables among an increased number of new, small and scattered, multiple-origin, transnationally connected, socio-economically differentiated and legally stratified immi- grants who have arrived over the last decade.” (p. 1024) 2 One example is the pressure of transnational institutions on central government to alter migration restriction to promote easier labor mobility, the result is the increased cross-border migratory flows on nation-state level (Castles, 2002), and with concentration of immigrants in metropolitan areas, ethnic segregation is expected to increase on the local level due to the deregulation of housing markets in global cities. 3 One example is the pressure of transnational institutions on central government to alter migration restriction to promote easier labor mobility, the result is the increased cross-border migratory flows on nation-state level (Castles, 2002), and with concentration of immigrants in metropolitan areas, ethnic segregation is expected to increase on the local level due to the deregulation of housing markets in global cities. 4 Gated communities are a from of gentrification, as – by definition – gentrification involves the gradual up-scaling of land-users where the new users are of a higher socioeconomic status than the previous users, together with an associated change in the built environment through a reinvestment in fixed capital (Clark, 2005) 5 Which altered the entry visa and work permits regulations especially for mainland Chinese (Ullah, 2012) causing unprecedented influx of workers to the city 6 Consequently, in-flows of foreign (non-Chinese) workers to Hong Kong slowed down compared to other alpha cities in the same period of time. Which altered the ethnic composition of the city due to the relative retreat of foreign immigration and the increased inflows of Mainland Chinese in a basically Chinese community

Conclusions and discussion | 259

Appendices

Appendix I

i. Tables of results

ii. iii. iv. This Appendix presents the values of the ‘spatial multi-group dissimilarity index’ SD(m) calculated for all cities of the dataset. As well as the total percentage and average yearly percentage of change in SD(m) value over time. v. vi. vii. viii. ix. x. xi. xii. xiii. xiv. xv. xvi.

Tables of results | 261

Table i-1: Socio-economic segregation index value for global cities, census year, and total/yearly change in index value

Table i-1 Socio-economic segregation index value for global cities, census year, and total/yearly change in index value in index change and total/yearly year, census cities, global for value index segregation Socio-economic i-1 Table

262 | Appendix I

xvii. e in index index in e g chan y earl y and total/ and , ear y census census , lobal cities g ation index value for value index ation g re g Cont. Table i-1: Socio-economic se Socio-economic i-1: Table Cont.

Tables of results | 263 xviii.

index in change total/yearly and year, census cities, for global value index segregation Socio-economic i-1: Table Cont.

264 | Appendix I

Table i-2 Ethnic segregation index value for global cities, census year, and total/yearly change in index value index in change total/yearly and year, census cities, global for value index segregation Ethnic i-2 Table

Tables of results | 265

Cont. Table i-2: Ethnic segregation index value for global cities, census year, and total/yearly change in index value in index change total/yearly and year, census cities, for global value index segregation i-2: Ethnic Table Cont.

Table i-2 Ethnic segregation index value for global cities, census year, and total/yearly change in index value

266 | Appendix I

xix.

Cont. Table i-2: Ethnic segregation index value for global cities, census year, and total/yearly change in index value index in change and total/yearly year, census cities, for global value index segregation i-2: Ethnic Table Cont.

Tables of results | 267

xx. xxi.

xxii. xxiii. xxiv. xxv. xxvi. xxvii. xxviii. xxix. xxx. xxxi. xxxii. xxxiii. xxxiv. xxxv. xxxvi. xxxvii. xxxviii. xxxix. xl. xli. xlii. xliii. xliv. xlv.

Cont. Table i-2: Ethnic segregation index value for global cities, census year, and total/yearly change in index value in index change total/yearly and year, census cities, for global value index segregation i-2: Ethnic Table Cont.

Appendix II

ii. Statistical hypothesis testing

Basically, statistical hypothesis tests aim to determine whether the observations drawn from the dataset are near the expected results defined by the hypothesis under examination (H0), or the observations are significantly various to the extent that there is high probability the ‘null’ hypothesis (H0) is rejected in favor of an alternative hypothesis (H1) (Gentle, 2002; Gibbons and Chakraborti, 2003). Simply put, the null hypothesis (H0) in our case is the association between economic restructuring and intensified spatial segregation within global cities. And according to (H0), the expected result is a tendency among global cities to have parallel changes towards increased segregation. However, in case that a number of global cities in the dataset shown a change in the opposite direction (towards more integration instead of more segregation); then, how to determine whether this number of cities is significant enough to rule out the null hypothesis or not? In other words, do cities with divergent behavior provide enough evidence against the global/divided city model? In order to decide whether to reject the null hypothesis or not based on the study observations, a statistical measure called the p-value (also known as probability value or the significance probability (Gibbons and Chakraborti, 2003)) is computed. By definition, the p-value is the probability of obtaining observations from the study that are within the normal range defined by the null hypothesis (Goodman, 1999). If this probability is high (higher than a pre-determined significance level of – for instance - 1% or 5%) (Stigler, 2008; Gentle, 2002), then the null hypothesis cannot be rejected as there is high probability that the observations are

Data catalogue | 269 within normal range and the null hypothesis might be true. While in contrast, if the p-value is below 5%, then the null hypothesis is rejected, because the observations are significantly outside the normal range defined by the null hypothesis. For example, if the global/divided city model requires – as an assumption - 85% of cities of the dataset to show increase in their spatial segregation, yet the observations confirmed that only 70% of cities show such increase, then the calculation of the p-value will determine whether that 70% is significant enough to oppose the global/divided city model, or it is within the normal range of the expected observations. One of the ways1 to determine the p-value is through performing the Pearson's Chi-squared test (χ2); which basically compares the expected counts of particular cases to their observed counts (Kirkman, 1996) where:

Where

Oi = the observed count of cases

Ei = the expected count of cases asserted by the null hypothesis n = the number of possible categories within the cases - for example, a coin flipping test has two possible categories, heads and tails. In our case cities are also categorized into two: cities with increase in segregation, and cities with decrease in segregation. After calculating the χ2 value, statistics textbooks - such as Gibbons and Chakraborti (2003) - already provide Chi-squared distribution tables that determine the equivalent p-value for the calculated χ2 for different values of n (see the subsequent chapters for more details about how the test is performed over the final dataset). In case that the p-value is less than the defined significance level (usually 5%), then the generalized global/divided city model is opposed, and the alternative hypothesis (H1) is supported.

270 | Appendix II

Where, (H1) suggests that local contextual differences are producing significantly divergent outcomes of restructuring for global cities in every context. The results in chapter 5 showed inconclusive patterns of change in socioeconomic segregation in global cities, more than half of cities of the dataset (54.5%) showed a tendency towards increased segregation, yet the only way to assert that the remaining (45.5%) of cities is significant enough to rule out any association between cities’ integration in the global network and their change in spatial segregation is through determining the ‘significance probability’ p-value of this null hypothesis. The process of determining the p-value is explained in the following steps:

1. Confirm that the ‘null hypothesis’ H0 is stated clearly, according to

the discussion in the preceding chapters, H0 states that ‘there is a tendency among global cities to show intensified patterns of spatial segregation due to the growing income inequality and social polarization within these cities as a direct outcome of their economic restructuring’. 2. Identify the expected results of the analysis in case the null hypothesis is true. In view of the above, the tendency among global cities to show increase in segregation level suggests that a ‘high percentage’ of cities of the dataset must show such increase. However, this ‘high percentage’ is not clearly defined, does it mean 95% or 80% or even 70% of cities of the dataset with increased segregation constitute a tendency? Therefore, since it is not an easy task to translate this ‘tendency’ into a definitive numerical value, then the χ2 is calculated several times based on the range of several possible expected observations suggested by the null hypothesis. See table 5-1 for more details 3. Calculate the χ2 value. 4. Determine the DF (degree of freedom), where DF = n -1, (n= the number of possible categories within the cases under study – as noted

Data catalogue | 271

in chapter 4, cities are categorized into two: cities with increase in segregation, and cities with decrease in segregation. Accordingly, DF = 2 - 1 = 1 5. Choose a significance level for p-value. By convention, scientists usually set the significance value for their experiments at 0.05, or 5 percent (Vaughan, 2001). If the P value is less than or equal to 0.05,

the decision is to reject H0; otherwise, the decision is not to reject H0 (Gibbons and Chakraborti, 2003). 6. Finally, in a Chi-squared distribution table (see table ii-1), each row in the table represents the corresponding χ2 values for different degrees of freedom, while the columns represents intervals of the p-value for each χ2 value. Given that the DF = 1, then by reading the row where DF equals 1 across from the left to the right until the first cell with a value bigger than the calculated χ2 value appears, the corresponding p-value appears in the top of this column.

Chi-squared distribution table As noted in step 2, the expected number of cities with increased segregation varies based on the definition of the word ‘tendency’ in the null hypothesis. Accordingly, χ2 is calculated when 95% of the dataset is

272 | Appendix II expected to show an increase in their socioeconomic segregation, then the χ2 is calculated again if this ration is expected to be 90%, 85%, 80%, 75%, or 70%. With each value for χ2 the p-value is determined and a decision about rejecting the null hypothesis or not is made. Table ii-2 shows the different χ2 value and their corresponding’s p-value. Note that the dataset contains total of 66 cities, the observations confirmed that only 36 cities showed an increase in their socioeconomic segregation index compared to 30 cities showed a decrease in their segregation index. The (36 to 30) ratio is compared to other expected ratios as follows:

Chi-squared value and their corresponding p-value χ2 value when expected Expected ratio ratio is Cities (of the of cities with compared dataset) with increased to to the tendency for cities with observation Null increased decreased (36 to 30) hypothesis segregation segregation cities p-value (H0) 95% 62 to 4 180 0.000 < 0.05 Rejected 90% 59 to 7 84.5 0.000 < 0.05 Rejected 85% 56 to 10 47.1 0.000 < 0.05 Rejected 80% 52 to 14 23.2 0.000 < 0.05 Rejected 75% 49 to 17 13.4 0.000 < 0.05 Rejected 70% 46 to 20 7.17 0.005 < 0.05 Rejected

Noticeably, based on a significance value 0.05, the null hypothesis is rejected as the p-value is founded less than the significance value in every case shown in table ii-2 (whether the null hypothesis require 95% of cities of the dataset to show an increase in segregation or even only 70%). In other

Data catalogue | 273 words, the 45.5 % of cities with decreased segregation are significant enough to rule out any suggested association between cities’ integration in the global network of cities and the intensified socioeconomic segregation within them. The case of socioeconomic segregation required performing the Chi- squared test, as the observations were not conclusive on their own. However, in the case of ethnic segregation, the test is not necessary as it confirms the obvious, because only less than half of cities of the dataset showed an increased in ethnic segregation index, which suggests that there is no tendency among global cities to show an increase in their ethnic segregation as propagated in the world/global city literature.

Notes

1 The Pearson’s chi-squared test is recommended by Dr. Manal Nassar - Professor of statistics in Ain Shams University to be suitable for the requirement of this research, compared to other statistical hypothesis tests including: t-test, student’s test, and The Kolmogorov-Smirnov test (KS-test).

Appendix III

iii. Data catalogue

This appendix presents all cities of the dataset with respect to their available data, data type, data source, as well as a guide basemap showing the boundaries and the internal sub-city divisions of each case. Cities are listed in alphabetical order.

Data catalogue | 275

City: Adelaide Country: Australia Global status: gamma (+) Rank: 127 Types of divisions: statistical local areas Number of divisions: 54 Income based segregation: Type of data: Gross weekly individual income of population over 15 years of age Population categories: 0$, 1$ to 159$, 160$ to 399$, 400$ to 599$, 600$ to 799$, 800$ to 999$, 1000$ to 1549$, 1550$ or more Ethnic based segregation: Type of data: Population by birthplace Population categories: People born in UK, surrounding territories, Anglo America, Western Europe, Eastern Europe and Former SU, South/East Asia, Middle East, Africa, and elsewhere. Data source: 2001 Census of Population and Housing 2006 Census of Population and Housing Australian Bureau of Statistics

City: Amsterdam Country: The Netherlands Global status: alpha Rank: 21 Types of divisions: neighborhoods Number of divisions: 94 Income based segregation: Type of data: Households in top middle and bottom of the income structure Population categories: Bottom 40% Middle 40% Top 20% Ethnic based segregation: Type of data: Population ethnic origin Population categories: Surinamese, from Antil and Aruba, Turkish, Moroccan, other not European, other European, and natives Data source: Gemeente Amsterdam Dienst Onderzoek en Statistiek: Stadsdelen in cijfers 2000, 2002, 2003, 2011

276 | Appendix II

City: Antwerp Country: Belgium Global status: gamma(+) Rank: 118 Types of divisions: Wijken (quarters) Number of divisions: 62

Income based segregation: Type of data: No detailed data available, 2001 and 2008 income data are available on city level

Ethnic based segregation: Type of data: Population by nationality Population categories: Belgian, EU 15 countries, EU 12 countries, other European, Turkish, Moroccan, Africa, Asia, America/Oceania, and Refugees Data source: Stad Antwerpen Stad Antwerpen in Cijfers, 2000, 2006, 2012

City: Atlanta Country: USA Global status: Alpha (-) Rank: 40 Types of divisions: census tracts Number of divisions: 601 Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/Australia, Caribbean/Central America, South America, North America, and not foreign born population. Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

Data catalogue | 277

City: Auckland Country: New Zealand Global status: beta Rank: 71 Types of divisions: area units Number of divisions: 292 Income based segregation: Type of data: Total Personal Income, for the Census Usually Resident Population Count Aged 15 Years and Over Population categories: 5000$ or less, 5001$ to 10000, 10001$ to 20000$, 20001$ to 30000$, 30001$ to 50000$, 50001$ or more Ethnic based segregation: Type of data: Population by ethnic group Population categories: European, Maori, Pacific, Asian, MELLA, and other Data source: 1996 Census, 2000 Census, 2006 Census, Statistics New Zealand

City: Austin Country: USA Global status: gamma(-) Rank: 168 Types of divisions: census tracts Number of divisions: 204 Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/Australia, Caribbean/Central America, South America, North America, and not foreign born population. Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

278 | Appendix II

City: Baltimore Country: USA Global status: gamma(+) Rank: 121 Types of divisions: census tracts Number of divisions: 543 Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/Australia, Caribbean/Central America, South America, North America, and not foreign born population. Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Barcelona Country: Spain Global status: alpha (-) Rank: 41 Types of divisions: zones de recca Number of divisions: 248 Income based segregation: Type of data: No comparable data on income is available

Ethnic based segregation: Type of data: Population by nationality Population categories: People from Spain, other European countries, Africa, North America, Central America, South America, Asia, and Oceania Data source:

Ajuntament de Barcelona Xifres oficials de població a 1 de gener de 2003 y 2009

Data catalogue | 279

City: Beijing Country: China Global status: alpha (+) Rank: 12 Types of divisions: Districts Number of divisions: 12 Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by citizenship status

Population categories: Foreigners Not foreigners Data source:

2010 Beijing sub-counties major statistical data directory.

City: Berlin Country: Germany Global status: beta(+) Rank: 56 Types of divisions: districts Number of divisions: 12 Income based segregation: Type of data: Personal monthly net income Population categories: Under 500€, 500€ to 899€, 900€ to 1299€, 1300€ to 1999€, 2000€ to 2599€, 2600€ and more Ethnic based segregation: Type of data: Population by citizenship status Population categories: Foreigners not Foreigners Data source:

Statistik Berlin: Results of the micro-census In Berlin in 2006, and 2010

280 | Appendix II

City: Bogota Country: Colombia Global status: beta(+) Rank: 62 Types of divisions: districts Number of divisions: 19 Income based segregation: Type of data: Population by income covers basic needs Population categories: People with income not enough to cover minimum needs People with income just enough to cover minimum needs people with income cover more than minimum needs Ethnic based segregation: Type of data: No data available

Data source:

DANE - DAPD, Quality of Life Survey 2003 Bogotá DANE - SDP, Quality of Life Survey 2007 Bogotá

City: Boston Country: USA Global status: alpha (-) Rank: 36 Types of divisions: census tracts Number of divisions: 895 Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/Australia, Caribbean/Central America, South America, North America, and not foreign born population. Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

Data catalogue | 281

City: Bratislava Country: Slovakia Global status: beta(-) Rank: 93 Types of divisions: districts Number of divisions: 5

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by nationality by continent. Population categories: Slovak, Russian, Hungarian, Gipsy, Czech, Ruthenian, Ukrainian, German, Polish, and others.

Data source: Statistical Office of the Slovak Republic – Regional Office of Bratislava

City: Brisbane Country: Australia Global status: beta(-) Rank: 87 Types of divisions: Statistical Local Areas Number of divisions: 66 Income based segregation: Type of data: Gross weekly individual income of population over 15 years of age Population categories: 0$, 1$ to 159$, 160$ to 399$, 400$ to 599$, 600$ to 799$, 800$ to 999$, 1000$ to 1549$, 1550$ or more Ethnic based segregation: Type of data: Population by birthplace Population categories: People born in UK, surrounding territories, Anglo America, Western Europe, Eastern Europe and Former SU, South/East Asia, Middle East, Africa, and elsewhere. Data source: 2001 Census of Population and Housing 2006 Census of Population and Housing Australian Bureau of Statistics

282 | Appendix II

City: Brussels Country: Belgium Global status: alpha Rank: 25 Types of divisions: municipalities Number of divisions: 19

Income based segregation: Type of data: Population by yearly taxable income

Population categories: Less than 10000€, 10001€ to 20000€ 20001€ to 30000€, 30001€ to 40000€ 40001€ to 50000€, 50001€ and more

Ethnic based segregation: Type of data: Population by nationality (broad regions)

Population categories: European, Turkish, North Africa, Sub-Sahara and West Africa, and others.

Data source: Nationaal Instituut voor de Statistiek. Algemene Directie Statistiek en Economische Informatie.

City: Calgary Country: Canada Global status: beta(-) Rank: 109 Types of divisions: neighborhoods Number of divisions: 145

Income based segregation: Type of data: Yearly income of population over 15 years of age

Population categories: No income, under 1000$, 1000$ to 2999$, 3000$ to 4999$, 5000$ to 6999$, 7000$ to 9999$, 10000$ to 11999$, 12000$ to 14999$, 15000$ to 19999$, 20000$ to 24999, 25000$ to 29999$, 30000$ to 34999$, 35000$ to 39999$, 40000$ to 44999$, 45000$ to 49999$, 50000$ to 59999$, 60000 and over.

Ethnic based segregation: Type of data: Population by visible minority group

Population categories: Not minority, South Asian, Chinese, Japanese, Black, Latino, Arab and West Asian, multiple race, and not classified

Data source: 2001 Census – 2006 Census: Statistics Canada.

Data catalogue | 283

City: Cape Town Country: South Africa Global status: beta Rank: 79 Types of divisions: wards Number of divisions: 100

Income based segregation: Type of data: Population by their individual monthly income Population categories: No income, 1$ to 6400$, 6401$ to 12800$, 12801 to 51200$, 51201$ to 102400$, 102401$ to 204800$, Over 204800

Ethnic based segregation: Type of data: Population by race Population categories: Black, White, Colored, and Indian

Data source: Statistics South Africa: 1996 census - 2001 census.

City: Charlotte Country: USA Global status: gamma(+) Rank: 117 Types of divisions: census tracts Number of divisions: 161

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

284 | Appendix II

City: Chicago Country: USA Global status: alpha (+) Rank: 8 Types of divisions: census tracts Number of divisions: 1912

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Cincinnati Country: USA Global status: gamma(+) Rank: 116 Types of divisions: census tracts Number of divisions: 392 Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/Australia, Caribbean/Central America, South America, North America, and not foreign born population. Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

Data catalogue | 285

City: Cleveland Country: USA Global status: beta(-) Rank: 107 Types of divisions: census tracts Number of divisions: 615 Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/Australia, Caribbean/Central America, South America, North America, and not foreign born population. Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Cologne Country: Germany Global status: beta(-) Rank: 99 Types of divisions: district Number of divisions: 85

Income based segregation: Type of data: No data available.

Ethnic based segregation: Type of data: Population by nationality by continent Population categories: German, African, Asian, other European, North American, Australian, and Turkish

Data source: City of Cologne - Department of City Development and Statistics (Statistical Information System). Kölner Stadtteilinformationen .

286 | Appendix II

City: Columbus Country: USA Global status: gamma Rank: 149 Types of divisions: census tracts Number of divisions: 292

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Copenhagen Country: Denmark Global status: beta(+) Rank: 60 Types of divisions: district Number of divisions: 15

Income based segregation: Type of data:

Households by taxable income

Population categories: Less than 49.999kr, 50000kr to 99999kr, 100000kr to 149999kr, 150000kr to 199999kr, 200000kr to 299999kr, 300000kr to 399999kr, 400000kr to 499999kr, 500000kr to 599999kr, 600000kr to 699999kr, 700000kr and more.

Ethnic based segregation: Type of data: Population by ethnic origin

Population categories: Denmark, other Europe, Africa North America , South and central America, Asia, Oceania, stateless, and unknown

Data source: Statistics Copenhagen, Copenhagen City - www.kk.dk/statistik

Data catalogue | 287

City: Dallas Country: USA Global status: alpha(-) Rank: 38 Types of divisions: census tracts Number of divisions: 901

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Denver Country: USA Global status: beta(-) Rank: 91 Types of divisions: census tracts Number of divisions: 484

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

288 | Appendix II

City: Detroit Country: USA Global status: beta(-) Rank: 90 Types of divisions: census tracts Number of divisions: 1148

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Durban Country: South Africa Global status: gamma(-) Rank: 167 Types of divisions: wards Number of divisions: 103

Income based segregation: Type of data: Population by their individual monthly income Population categories: No income, 1$ to 6400$, 6401$ to 12800$, 12801 to 51200$, 51201$ to 102400$, 102401$ to 204800$, Over 204800

Ethnic based segregation: Type of data: Population by race Population categories: Black, White, Colored, and Indian

Data source: Statistics South Africa: 1996 census - 2001 census.

Data catalogue | 289

City: Dusseldorf Country: Germany Global status: beta(+) Rank: 48 Types of divisions: districts Number of divisions: 49

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by citizenship status Population categories: German, and foreigners.

Data source: Landeshauptstadt Düsseldorf. (city of Dusseldorf). Landesamt für Datenverarbeitung und Statistik NRW. (State Office for Data Processing and Statistics NRW).

City: Edmonton Country: Canada Global status: gamma Rank: 152 Types of divisions: neighborhoods Number of divisions: 210

Income based segregation: Type of data: Yearly income of population over 15 years of age

Population categories: No income, under 1000$, 1000$ to 2999$, 3000$ to 4999$, 5000$ to 6999$, 7000$ to 9999$, 10000$ to 11999$, 12000$ to 14999$, 15000$ to 19999$, 20000$ to 24999, 25000$ to 29999$, 30000$ to 34999$, 35000$ to 39999$, 40000$ to 44999$, 45000$ to 49999$, 50000$ to 59999$, 60000 and over.

Ethnic based segregation: Type of data: Population by visible minority group

Population categories: Not minority, South Asian, Chinese, Japanese, Black, Latino, Arab and West Asian, multiple race, and not classified

Data source: 2001 Census – 2006 Census: Statistics Canada.

290 | Appendix II

City: Frankfurt Country: Germany Global status: alpha Rank: 19 Types of divisions: neighborhoods Number of divisions: 45

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by citizenship status Population categories: German, and foreigners.

Data source: Statistisches Jahrbuch Frankfurt am Main: 2007, and 2011. Stadt Frankfurt am Main Bürgeramt, Statistik und Wahlen.

City: Geneva Country: Switzerland Global status: beta(-) Rank: 88 Types of divisions: municipality Number of divisions: 16

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by citizenship status Population categories: Foreigners, and not foreigners.

Data source: Office cantonal de la statistique – OCSTAT: Statistique Genève.

Data catalogue | 291

City: Gothenburg Country: Sweden Global status: gamma(-) Rank: 158 Types of divisions: district Number of divisions: 94

Income based segregation: Type of data: Individuals disposable income in thousands kr.

Population categories: No income, 0.1 kr to 79 kr, 80 kr to 159 kr, 160 kr to 239 kr, 240 kr, 359 kr and more.

Ethnic based segregation: Type of data: Foreign-born population

Population categories: Swedish, other Scandinavian, other European, Asia, Africa, North America, South American, and others.

Data source: Statistiska centralbyrån (SCB). INK3 Göteborg.

City: Guadalajara Country: Mexico Global status: gamma Rank: 139 Types of divisions: municipalities Number of divisions: 8

Income based segregation: Type of data: Employed population by their income Population categories: Individuals earning less income than the minimum wage, individual earning the minimum wage to double the minimum wages, individual earning more than double the minimum wage, and individual whose income is not specified

Ethnic based segregation: Type of data: No data available

Data source: INEGI: XII Censo general de poblacion y vivienda 2000. INEGI: Censo de Población y Vivienda 2010.

292 | Appendix II

City: Johannesburg Country: South Africa Global status: alpha(-) Rank: 47 Types of divisions: wards Number of divisions: 109

Income based segregation: Type of data: Population by their individual monthly income Population categories: No income, 1$ to 6400$, 6401$ to 12800$, 12801 to 51200$, 51201$ to 102400$, 102401$ to 204800$, Over 204800

Ethnic based segregation: Type of data: Population by race Population categories: Black, White, Colored, and Indian

Data source: Statistics South Africa: 1996 census - 2001 census.

City: Hamburg Country: Germany Global status: beta(+) Rank: 53 Types of divisions: neighborhoods Number of divisions: 98

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by citizenship status Population categories: German, and foreigners.

Data source: Statistisches Amt für Hamburg und Schleswig- Holstein (Statistical Office of Hamburg and Schleswig-Holstein).

Data catalogue | 293

City: Hong Kong Country: China Global status: alpha(+) Rank: 3 Types of divisions: statistical divisions Number of divisions: 128

Income based segregation: Type of data: Monthly income of individuals over 15 years of age

Population categories: No income, under 2000$, 2000$ to 3999$, 4000 to 5999$, 6000$ to 7999$, 8000$ to 9999$, 10000$ to 14999$, 15000$ to 19999$, 20000$ to 24999$, 25000 to 39999$, 40000$ and more.

Ethnic based segregation: Type of data: Population by origin Population categories: Chinese, other Asian, and others.

Data source: Census and statistics department: The Government of the Hong Kong Special Administrative Region.

City: Houston Country: USA Global status: beta(+) Rank: 55 Types of divisions: census tracts Number of divisions: 731

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

294 | Appendix II

City: Indianapolis Country: USA Global status: gamma Rank: 150 Types of divisions: census tracts Number of divisions: 275

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Kansas City Country: USA Global status: gamma Rank: 136 Types of divisions: census tracts Number of divisions: 423

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

Data catalogue | 295

City: London Country: UK Global status: alpha(++) Rank: 1 Types of divisions: boroughs Number of divisions: 32

Income based segregation: Type of data: No comparable data available.

Ethnic based segregation: Type of data: Population by ethnic group Population categories: Whites, Black African, Black Caribbean, other Blacks, Indians, Pakistanis, Bangladeshis, Chinese, other Asians, and others (including mixed ethnicity).

Data source: London Borough Profiles: Greater London Authority.

City: Los Angeles Country: USA Global status: alpha Rank: 17 Types of divisions: census tracts Number of divisions: 2295

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

296 | Appendix II

City: Lyon Country: France Global status: gamma Rank: 140 Types of divisions: commune Number of divisions: 138

Income based segregation: Type of data: No data available.

Ethnic based segregation: Type of data: Population by citizenship status Population categories: Foreigners , and not foreigners.

Data source: Recensement de la population de 1982, 1990, 1999 et 2008: Insee, Institut national de la statistique et des etudes economiques.

City: Madrid Country: Spain Global status: alpha Rank: 15 Types of divisions: barrios Number of divisions: 128

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by citizenship status Population categories: Foreigners, and not foreigners.

Data source: Madrid Datos: Direccion General de Estadistica (Madrid info: General Bureau of Statistics)

Data catalogue | 297

City: Manchester Country: UK Global status: beta Rank: 76 Types of divisions: wards Number of divisions: 32

Income based segregation: Type of data: No comparable data available

Ethnic based segregation: Type of data: Population by ethnic groups

Population categories: Whites, Black African, Black Caribbean, other Blacks, Indians, Pakistanis, Bangladeshis, Chinese, other Asians, and others (including mixed ethnicity).

Data source: Manchester City Council. Office of National Statistics: Neighborhood profile.

City: Marseille Country: France Global status: gamma(-) Rank: 159 Types of divisions: commune Number of divisions: 67

Income based segregation: Type of data: No data available.

Ethnic based segregation: Type of data: Population by citizenship status Population categories: Foreigners , and not foreigners.

Data source: Recensement de la population de 1982, 1990, 1999 et 2008: Insee, Institut national de la statistique et des etudes economiques.

298 | Appendix II

City: Melbourne Country: Australia Global status: alpha(-) Rank: 31 Types of divisions: statistical local areas Number of divisions: 57

Income based segregation: Type of data: Gross weekly individual income of population over 15 years of age

Population categories: 0$, 1$ to 159$, 160$ to 399$, 400$ to 599$, 600$ to 799$, 800$ to 999$, 1000$ to 1549$, 1550$ or more

Ethnic based segregation: Type of data: Population by birthplace

Population categories: Individuals born in the UK, surrounding territories, Anglo America, Western Europe, Eastern Europe and former SU, South/East Asia, Middle East, Africa, and elsewhere

Data source: 2001 Census of Population and Housing 2006 Census of Population and Housing Australian Bureau of Statistics

City: Mexico city Country: Mexico Global status: alpha Rank: 20 Types of divisions: municipalities Number of divisions: 50

Income based segregation: Type of data: Employed population by their income Population categories: Individuals earning less income than the minimum wage, individual earning the minimum wage to double the minimum wages, individual earning more than double the minimum wage, and individual whose income is not specified

Ethnic based segregation: Type of data: No data available

Data source: INEGI: XII Censo general de poblacion y vivienda 2000. INEGI: Censo de Población y Vivienda 2010.

Data catalogue | 299

City: Miami Country: USA Global status: alpha (-) Rank: 29 Types of divisions: census tracts Number of divisions: 878

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Milan Country: Italy Global status: alpha(+) Rank: 11 Types of divisions: Nuclei di Identità Locale Number of divisions: 88

Income based segregation: Type of data: No comparable data available

Ethnic based segregation: Type of data: Population by citizenship status

Population categories: Italian, and not Italian

Data source: Comune di Milano - Settore Statistica e S.I.T. - Servizio Statistica. (City of Milan - Statistics Service)

300 | Appendix II

City: Milwaukee Country: USA Global status: gamma(-) Rank: 174 Types of divisions: census tracts Number of divisions: 396

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Minneapolis Country: USA Global status: beta Rank: 82 Types of divisions: census tracts Number of divisions: 643

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

Data catalogue | 301

City: Monterrey Country: Mexico Global status: beta(-) Rank: 92 Types of divisions: municipalities Number of divisions: 11

Income based segregation: Type of data: Employed population by their income Population categories: Individuals earning less income than the minimum wage, individual earning the minimum wage to double the minimum wages, individual earning more than double the minimum wage, and individual whose income is not specified

Ethnic based segregation: Type of data: No data available

Data source: INEGI: XII Censo general de poblacion y vivienda 2000. INEGI: Censo de Población y Vivienda 2010.

City: Montréal Country: Canada Global status: beta(+) Rank: 51 Types of divisions: neighborhoods Number of divisions: 81

Income based segregation: Type of data: Yearly income of population over 15 years of age

Population categories: No income, under 1000$, 1000$ to 2999$, 3000$ to 4999$, 5000$ to 6999$, 7000$ to 9999$, 10000$ to 11999$, 12000$ to 14999$, 15000$ to 19999$, 20000$ to 24999, 25000$ to 29999$, 30000$ to 34999$, 35000$ to 39999$, 40000$ to 44999$, 45000$ to 49999$, 50000$ to 59999$, 60000 and over.

Ethnic based segregation: Type of data: Population by visible minority group

Population categories: Not minority, South Asian, Chinese, Japanese, Black, Latino, Arab and West Asian, multiple race, and not classified

Data source: 2001 Census – 2006 Census: Statistics Canada.

302 | Appendix II

City: Munich Country: Germany Global status: alpha(-) Rank: 34 Types of divisions: districts Number of divisions: 25

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by nationality Population categories: German, Turkish, other EU nationals, East European, and other nationalities

Data source: Statistisches Amt Munchen

City: New York Country: USA Global status: alpha (++) Rank: 2 Types of divisions: census tracts Number of divisions: 4429

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

Data catalogue | 303

City: Orlando Country: USA Global status: gamma(-) Rank: 157 Types of divisions: census tracts Number of divisions: 262

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Oslo Country: Norway Global status: beta Rank: 72 Types of divisions: commune Number of divisions: 16

Income based segregation: Type of data: population by gross income of tax payers 14 years of age and older. Population categories: No income, 100NOK to 49900NOK, 50000 to 99900, 100000 to 149900, 150000 to 199900, 200000 to 249900, 250000 to 299900, 300000 to 349900, 400000 to 449900, 450000 to 499900, 500000NOK or more.

Ethnic based segregation: Type of data: Population by continent of origin Population categories: Nordic, Western European, Eastern European, North America and Oceania, and others including: Asia, Africa, and Latin America.

Data source: Statistisk sentralbyrå/statistisk kontor (Statistics Norway - statistical office)

304 | Appendix II

City: Ottawa Country: Canada Global status: gamma(-) Rank: 160 Types of divisions: neighborhoods Number of divisions: 235

Income based segregation: Type of data: Yearly income of population over 15 years of age

Population categories: No income, under 1000$, 1000$ to 2999$, 3000$ to 4999$, 5000$ to 6999$, 7000$ to 9999$, 10000$ to 11999$, 12000$ to 14999$, 15000$ to 19999$, 20000$ to 24999, 25000$ to 29999$, 30000$ to 34999$, 35000$ to 39999$, 40000$ to 44999$, 45000$ to 49999$, 50000$ to 59999$, 60000 and over.

Ethnic based segregation: Type of data: Population by visible minority group

Population categories: Not minority, South Asian, Chinese, Japanese, Black, Latino, Arab and West Asian, multiple race, and not classified

Data source: 2001 Census – 2006 Census: Statistics Canada.

City: Panama City Country: Panama Global status: beta(-) Rank: 102 Types of divisions: districts Number of divisions: 21

Income based segregation: Type of data: Employed population by their income Population categories: No income, less than 100, 100 to 124, 125 to 174, 175 to 249, 250 to 399, 400 to 599, 600 to 799, 800 to 999, 1000 to1499, 1500 to 1999, 2000 to 2499, 2500 to 2999, 3000 to 3999, 4000 to 4999, 5000 and more, and not stated.

Ethnic based segregation: Type of data: No data available

Data source: Contraloria General de la Republica de Panama: Instituto Nacional de Estadistica y Censo. INEC.

Data catalogue | 305

City: Paris Country: France Global status: alpha(+) Rank: 4 Types of divisions: commune Number of divisions: 134

Income based segregation: Type of data: No data available.

Ethnic based segregation: Type of data: Population by citizenship status Population categories: Foreigners , and not foreigners.

Data source: Recensement de la population de 1982, 1990, 1999 et 2008: Insee, Institut national de la statistique et des etudes economiques.

City: Perth Country: Australia Global status: beta(-) Rank: 105 Types of divisions: statistical local areas Number of divisions: 28

Income based segregation: Type of data: Gross weekly individual income of population over 15 years of age

Population categories: 0$, 1$ to 159$, 160$ to 399$, 400$ to 599$, 600$ to 799$, 800$ to 999$, 1000$ to 1549$, 1550$ or more

Ethnic based segregation: Type of data: Population by birthplace

Population categories: Individuals born in the UK, surrounding territories, Anglo America, Western Europe, Eastern Europe and former SU, South/East Asia, Middle East, Africa, and elsewhere

Data source: 2001 Census of Population and Housing 2006 Census of Population and Housing Australian Bureau of Statistics

306 | Appendix II

City: Philadelphia Country: USA Global status: alpha(-) Rank: 46 Types of divisions: census tracts Number of divisions: 1360

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Phoenix Country: USA Global status: gamma Rank: 137 Types of divisions: census tracts Number of divisions: 630

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

Data catalogue | 307

City: Pittsburgh Country: USA Global status: gamma Rank: 151 Types of divisions: census tracts Number of divisions: 599

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Portland Country: USA Global status: gamma(+) Rank: 129 Types of divisions: census tracts Number of divisions: 369

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

308 | Appendix II

City: Richmond Country: USA Global status: gamma(-) Rank: 164 Types of divisions: census tracts Number of divisions: 228

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Rio de Janiero Country: Brazil Global status: beta(-) Rank: 86 Types of divisions: municipal district Number of divisions: 30

Income based segregation: Type of data: Employed population by their income Population categories: Individuals with no income, individuals earning less income than half the minimum wage, individuals earning 1 to 2 the minimum wage, individual earning 2 to 3 the minimum wage, individual earning 3 to 5 the minimum wage, individual earning 5 to 10 the minimum wage, , individual earning 10 to 15 the minimum wage, individual earning 15 to 20 the minimum wage, individual earning more than 20 the minimum wage.

Ethnic based segregation: Type of data: No data available

Data source: IBGE: Instituto Brasileiro de Geografia e Estatistica

Data catalogue | 309

City: Rome Country: Italy Global status: beta(+) Rank: 52 Types of divisions: municipalities Number of divisions: 20

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by citizenship status

Population categories: Italian, and not Italian

Data source: Ufficio di Statistica e Censimento del Comune di Roma. Bureau of Statistics and Census of the City of Roma

City: Rotterdam Country: The Netherlands Global status: gamma Rank: 147 Types of divisions: neighborhoods Number of divisions: 92

Income based segregation: Type of data: Households by income strata

Population categories: Household in the bottom 40% of the income distribution, household in the middle 40%, and household in the top 20% of the income distribution.

Ethnic based segregation: Type of data: Population by ethnic origin

Population categories: Natives, other European, Surinamese, Turkish, Moroccan, Anthill & Aruba, Cape Verde, other Western, and other non-Western.

Data source: Gemeente Rotterdam, Centrum voor Onderzoek en Statistiek.

310 | Appendix II

City: San Diego Country: USA Global status: beta(-) Rank: 103 Types of divisions: census tracts Number of divisions: 487

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: San Francisco Country: USA Global status: alpha Rank: 27 Types of divisions: census tracts Number of divisions: 724

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

Data catalogue | 311

City: San Jose Country: USA Global status: gamma(+) Rank: 133 Types of divisions: census tracts Number of divisions: 315

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Sao Paulo Country: Brazil Global status: alpha Rank: 14 Types of divisions: neighborhood Number of divisions: 8

Income based segregation: Type of data: Employed population by their income Population categories: Individuals earning less income than 3 the minimum wage, individual earning 3 to 5 the minimum wage, individual earning 5 to 10 the minimum wage, individual earning 10 to 20 the minimum wage, individual earning more than 20 the minimum wage,

Ethnic based segregation: Type of data: No data available

Data source: IBGE: Instituto Brasileiro de Geografia e Estatistica

312 | Appendix II

City: Seattle Country: USA Global status: beta Rank: 68 Types of divisions: census tracts Number of divisions: 608

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Seoul Country: South Korea Global status: alpha Rank: 24 Types of divisions: statistical divisions Number of divisions: 25

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by ethnic origin Population categories: China and Taiwan, North America, Europe and Russia, Central and South Asia, East and Southeast Asia, and others.

Data source: Seoul Metropolitan Government: Seoul Statistical Yearbook.

Data catalogue | 313

City: Shanghai Country: China Global status: alpha(+) Rank: 7 Types of divisions: districts Number of divisions: 12

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population immigration status Population categories: Foreign Not foreign

Data source: Shanghai Statistics Bureau

City: Singapore Country: Republic of Singapore Global status: alpha(+) Rank: 5 Types of divisions: planning zones Number of divisions: 35

Income based segregation: Type of data: Monthly income of population over 15 years of age

Population categories: Under 1000$, 1000$ to 1499$, 1500$ to 1999$, 2000$ to 2499, 2500$ to 2999$, 3000$ to 3999$, 40004 to 4999$, 5000$ to 5999$, 6000$ and over.

Ethnic based segregation: Type of data: Population by ethnic group

Population categories: Chinese, Malay, Indian, and others.

Data source: Census of Population 2000 - 2010 Singapore Department of Statistics.

314 | Appendix II

City: St Louis Country: USA Global status: beta(-) Rank: 100 Types of divisions: census tracts Number of divisions: 446

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Stockholm Country: Sweden Global status: beta(+) Rank: 49 Types of divisions: district Number of divisions: 128

Income based segregation: Type of data: Individuals disposable income in thousands kr.

Population categories: No income, 0.1 kr to 120 kr, 120 kr to 240 kr, 240 kr to 360 kr, 360 kr and more.

Ethnic based segregation: Type of data: Foreign-born population

Population categories: Swedish, other Scandinavian, other European, Asia, Africa, North America, South American, and others.

Data source: Stockholms stads Utrednings- och statistikkontor. (Stockholm office for Research and Statistics).

Data catalogue | 315

City: Stuttgart Country: Germany Global status: beta(-) Rank: 97 Types of divisions: statistical zones Number of divisions: 23

Income based segregation: Type of data: No data available.

Ethnic based segregation: Type of data: Population by citizenship status Population categories: Foreigners , and not foreigners.

Data source: Landeshauptstadt Stuttgart, Statistisches Amt (City of Stuttgart, Statistical Office).

City: Sydney Country: Australia Global status: alpha(+) Rank: 10 Types of divisions: statistical local areas Number of divisions: 24

Income based segregation: Type of data: Gross weekly individual income of population over 15 years of age

Population categories: 0$, 1$ to 159$, 160$ to 399$, 400$ to 599$, 600$ to 799$, 800$ to 999$, 1000$ to 1549$, 1550$ or more

Ethnic based segregation: Type of data: Population by birthplace

Population categories: Individuals born in the UK, surrounding territories, Anglo America, Western Europe, Eastern Europe and former SU, South/East Asia, Middle East, Africa, and elsewhere

Data source: 2001 Census of Population and Housing 2006 Census of Population and Housing Australian Bureau of Statistics

316 | Appendix II

City: Tallinn Country: Estonia Global status: gamma(-) Rank: 153 Types of divisions: districts Number of divisions: 8

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by ethnic groups Population categories: Natives, Russian, Ukrainian, Belarus, Finnish, Jews, Tatars, and others

Data source: Tallinna Linnavalitsus (Tallinn City Government). Estonian Ministry of the Interior: Population Register.

City: Tampa Country: USA Global status: gamma Rank: 148 Types of divisions: census tracts Number of divisions: 491

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

Data catalogue | 317

City: Tokyo Country: Japan Global status: alpha(+) Rank: 6 Types of divisions: wards Number of divisions: 23

Income based segregation: Type of data: No data available on ward level.

Ethnic based segregation: Type of data: Population by ethnic minority Population categories: population belongs to minority group, and native population (data does not include undocumented migrants)

Data source: Statistics Bureau, Ministry of Internal Affairs and Communications. Bureau of General Affairs: Tokyo Metropolitan Government.

City: Toronto Country: Canada Global status: alpha Rank: 13 Types of divisions: neighborhoods Number of divisions: 874

Income based segregation: Type of data: Yearly income of population over 15 years of age

Population categories: No income, under 1000$, 1000$ to 2999$, 3000$ to 4999$, 5000$ to 6999$, 7000$ to 9999$, 10000$ to 11999$, 12000$ to 14999$, 15000$ to 19999$, 20000$ to 24999, 25000$ to 29999$, 30000$ to 34999$, 35000$ to 39999$, 40000$ to 44999$, 45000$ to 49999$, 50000$ to 59999$, 60000 and over.

Ethnic based segregation: Type of data: Population by visible minority group

Population categories: Not minority, South Asian, Chinese, Japanese, Black, Latino, Arab and West Asian, multiple race, and not classified

Data source: 2001 Census – 2006 Census: Statistics Canada.

318 | Appendix II

City: Valencia Country: Spain Global status: gamma Rank: 135 Types of divisions: districts Number of divisions: 19

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by nationality by continent. Population categories: Spanish, other European, African, North and Central America, South America, Asia, and others.

Data source: Padró Municipal d’Habitants 2001 y 2010: Oficina d’Estadística. Ajuntament de València. (Municipal Population Census 2001 and 2010: Statistical Office. City of Valencia.)

City: Vancouver Country: Canada Global status: beta(+) Rank: 63 Types of divisions: neighborhoods Number of divisions: 375

Income based segregation: Type of data: Yearly income of population over 15 years of age

Population categories: No income, under 1000$, 1000$ to 2999$, 3000$ to 4999$, 5000$ to 6999$, 7000$ to 9999$, 10000$ to 11999$, 12000$ to 14999$, 15000$ to 19999$, 20000$ to 24999, 25000$ to 29999$, 30000$ to 34999$, 35000$ to 39999$, 40000$ to 44999$, 45000$ to 49999$, 50000$ to 59999$, 60000 and over.

Ethnic based segregation: Type of data: Population by visible minority group

Population categories: Not minority, South Asian, Chinese, Japanese, Black, Latino, Arab and West Asian, multiple race, and not classified

Data source: 2001 Census – 2006 Census: Statistics Canada.

Data catalogue | 319

City: Washington Country: USA Global status: alpha Rank: 28 Types of divisions: census tracts Number of divisions: 865

Income based segregation: Type of data: yearly earnings for population 16 years old and over Population categories: 1$ to 9999$, 10000$ to 19999$, 20000$ to 29999$, 30000$ to 39999$, 40000$ to 49999$, 50000$ to 64999$, 65000$ to 74999$, 75000$ to 99999$, 100000$ or more

Ethnic based segregation: Type of data: Population by birthplace Population categories: North/West Europe, South/East Europe, East Asia, South/Central Asia, South/East Asia, Middle East, East/Middle Africa, South/West Africa, Oceania/ Australia, Caribbean/Central America, South America, North America, and not foreign born.

Data source: 2000 US census - 2009 American Community Survey: US Census Bureau

City: Wellington Country: New Zealand Global status: gamma(-) Rank: 175 Types of divisions: wards Number of divisions: 158

Income based segregation: Type of data: Total Personal Income, for the Census Usually Resident Population Count Aged 15 Years and over Population categories: 5000$ or less, 5001$ to 10000, 10001$ to 20000$, 20001$ to 30000$, 30001$ to 50000$, 50001$ or more

Ethnic based segregation: Type of data: Population by ethnic group Population categories: European, Maori, Pacific, Asian, MELLA, and other

Data source: 1996 Census, 2000 Census, 2006 Census, Statistics New Zealand.

320 | Appendix II

City: Zurich Country: Switzerland Global status: alpha(-) Rank: 32 Types of divisions: neighborhoods Number of divisions: 34

Income based segregation: Type of data: No data available

Ethnic based segregation: Type of data: Population by citizenship status Population categories: Foreigners, and not foreigners.

Data source: Statistik Stadt Zürich.

References and bibliography

Aalbers, M. B., & Holm, A. (2008). Privatising social housing in Europe: The cases of Amsterdam and Berlin. In K. Adelhof, B. Glock, J. Lossau, & M. Schulz (Eds.), Urban trends in Berlin and Amsterdam (pp. 12-23). Berlin: Berliner Geographische Arbeiten, Humboldt Universität zu Berlin. Abrahamson, M. (2004). Global cities. New York: Oxford University Press. Andersen, H. T. (2002). Globalization, spatial polarization, and the housing market. Geografisk Tidsskrift: Danish Journal of Geography, 102, 93-102. Anderson, J. (1996). The shifting stage of politics: new medieval and postmodern territorialities?. Environment and Planning D, 14, 133-154. Andersson, R. (2007). Ethnic residential segregation and integration processes in Sweden. In K. Schönwälder (Ed.), Residential segregation and the integration of immigrants Britain, the Netherlands and Sweden (pp. 61-90). Berlin: WZB. Arthurson, K. (1998). Redevelopment of public housing estates: The Australian experience. Urban Policy and Research, 16(1), 35-46. Arthurson, K. (2012). Social mix and the city: Challenging the mixed communities consensus in housing and urban planning policies. Collingwood, Vic.: CSIRO Publishing. Atkinson, A. (1970). On the measurement of inequality. Journal of Economic Theory, 2(3), 244-263. doi: 10.1016/0022-0531(70)90039-6 Atkinson, R., & Bridge, G. (2005). Gentrification in a global context: The new urban colonialism. London: Routledge. Audirac, I., Cunningham-Sabot, E., Fol, S., & Moraes, S. T. (2012). Declining suburbs in Europe and Latin America. International Journal of Urban and Regional Research, 36(2), 226-244. doi: 10.1111/j.1468-2427.2011.01093.x Badcock, B. (1997). Recently observed polarising tendencies and Australian cities. Australian Geographical Studies, 35(3), 243-259. doi: 10.1111/1467- 8470.00025 Badcock, B., (1999): Doing more with less: public housing in the 1990s. In O’Connor, K. (ed.) Houses and Jobs in Cities and Regions: Research in Honour of Chris Maher. University of Queensland Press, Brisbane, 81–95. Badcock, B. (2000). The imprint of the post-fordist transition on Australian cities. In P. Marcuse & R. V. Kempen (Eds.), Globalizing cities: A new spatial order? (pp. 211-227). Oxford: Blackwell.

322 | References and bibliography

Barbosa, N. (2010). Latin America: Counter-cyclical policy in Brazil: 2008-09. Journal of Globalization and Development, 1(1). doi: 10.2202/1948- 1837.1052 Baum, S. (2008). Suburban scars: Australian cities and socio-economic deprivation (Research Paper 15). Urban Research Program, Griffith University. Retrieved November 9, 2012, from http://www.griffith.edu.au/__data/assets/pdf_file/0017/53009/urp-rp15- baum-2008.pdf Bauman, Z. (1998). On glocalization: or globalization for some, localization for some others. Thesis Eleven, 54(1), 37-49. Beall, J. (2002). Globalization and social exclusion in cities: Framing the debate with lessons from Africa and Asia. Environment and Urbanization, 14(1), 41-51. Beall, J., Crankshaw, O., & Parnell, S. (2000). Victims, villains and fixers: The urban environment and Johannesburg’s poor. Journal of Southern African Studies, 26(4), 803-855. Beauregard, R. A., & Deitrick, S. (1995). From front runner to also-run: The transformation of once-dominant region, Pennsylvania, USA. In P. Cooke (Ed.), The rise of the rustbelt (pp. 52-71). New York: St. Martin's Press. Beauregard, R. A., & Haila, A. (2000). The unavoidable continuities of the city. In P. Marcuse & R. Van Kempen (Eds.), Globalizing cities: A new spatial order? (pp. 22-36). Oxford: Blackwell. Beaverstock, J. V., Smith, R. G., & Taylor, P. J. (1999). A roster of world cities. cities, 16(6), 445-458. Beaverstock, J. V., Smith, R. G., & Taylor, P. J. (2000). World-city network: A new metageography? Annals of the Association of American Geographers, 90(1), 123-134. doi: 10.1111/0004-5608.00188 Beaverstock, J. V., Faulconbridge, J. R., & Hoyler, M. (2011). Globalization and the city. The SAGE Handbook of Economic Geography, 189-201. Beaverstock, J. V. (2011). Review of global urban analysis: a survey of cities in globalization, edited by Peter J. Taylor: Pengfei Ni, Ben Derudder, Michael Hoyler, Jin Huang and Frank Witlox, London, Earthscan, 2011, Urban

Research & Practice, 4(2), 219-221. Beck, U. (2000). The brave new world of work. Cambridge: Polity Press. Behr, J. G. (2004). Race, ethnicity, and the politics of city redistricting: Minority- opportunity districts and the election of Hispanics and Blacks to city councils. Albany: State University of New York Press. Bell, W. (1954). A probability model for the measurement of ecological segregation. Social Forces, 32, 337-364.

References and bibliography | 323

Bluestone, B., & Harrison, B. (1982). The deindustrialization of America: Plant closings, community abandonment, and the dismantling of basic industry. New York: Basic Books. Boal, F. W. (2005). Urban ethnic segregation and scenarios spectrum. In D. P. Varady (Ed.), Desegregating the city: Ghettos, enclaves, and inequality (pp. 62-78). Albany, NY: State University of New York Press. Bolt, G., Van Kempen, R., & Van Weesep, J. (2009). After urban restructuring: Relocations and segregation In Dutch cities. Tijdschrift Voor Economische En Sociale Geografie, 100(4), 502-518. doi: 10.1111/j.1467- 9663.2009.00555.x Bornschier, V. (2008, June). Income Inequality in the World–Looking Back and Ahead. In conference on “Inequality Beyond Globalisation” organised by the World Society Foundation and the RC02 of the International Sociological Association, University of Neuchatel (Vol. 28). Boter, I., Comabella, J., Lles, C. & Tobio,C. (1988). Socially transversal processes in a context of an increasingly differential social structure, Madrid, 1975- 1986. Paper presented at the Conference of the Research Committee on the Sociology of Urban and Regional Development, of the International Sociological Association, Rio de Janeiro, September. Brenner, N. (1998). Global cities, glocal states: Global city formation and state territorial restructuring in contemporary Europe. Review of International Political Economy, 5(1), 1-37. doi: 10.1080/096922998347633 Brenner, N. (2002). Decoding the Newest “metropolitan regionalism” in the USA: A critical overview. Cities, 19(1), 3-21. Brenner, N. (2004). New state spaces: Urban governance and the rescaling of statehood. Oxford: Oxford University Press. Briggs, X. (2005). Conclusion: Desegregating the city. In D. P. Varady (Ed.), Desegregating the city: Ghettos, enclaves, and inequality (pp. 233-257). Albany, NY: State university of New York press. Bråmå, Å. (2008). Dynamics of ethnic residential segregation in Göteborg, Sweden, 1995–2000. Population, Space and Place, 14(2), 101-117. doi: 10.1002/psp.479 Brandsen, T., Van de Donk, W., & Putters, K. (2005). Griffins or chameleons? Hybridity as a permanent and inevitable characteristic of the third sector. International Journal of Public Administration, 28(9-10), 749-765. doi: 10.1081/PAD-200067320 Bridge, G., & Watson, S. (2003). City differences. In G. Bridge & S. Watson (Eds.), A companion to the city (pp. 251-260). Oxford, UK: Blackwell.

324 | References and bibliography

Brussels Instituut voor STATISTIEK en Analyse. (2005). Buurtatlas van de bevolking van het Brussels Hoofdstedelijk Gewest bij de aanvang van de 21 eeuw (Nr.42) (Belgium, Ministerie van het Brussels Hoofdstedelijk Gewest). Burgers, J., & Engbersen, G. (1996). Globalization, migration, and undocumented migrants. New Community, 22(4), 619-635. Burgers, J., & Musterd, S. (2002). Understanding urban inequality: A model based on existing theories and an empirical illustration. International Journal of Urban and Regional Research, 26(2), 403-413. doi: 10.1111/1468- 2427.00387 Burgers, J. (2002). De gefragmenteerde stad. Amsterdam: Boom. Burgess, E. W., & Newcomb, C. S. (1933). Census data of the city of Chicago, 1930,. Chicago, IL: The University of Chicago press. Burke, R. J., & Cooper, C. L. (2000). The organization in crisis: Downsizing, restructuring, and privatization. Oxford, UK: Blackwell. Butler, T. (2005). Globalization and gentrification: The emergence of a middle range theory. Cities and Territories, 14, working paper series. Cairncross, F. (1997). The Death of distance: how the communications revolution will change our lives. Cambridge, MA: Harvard Business School. Caldeira, T.P.R. (2000). City of walls: crime, segregation, and citizenship in São Paulo. University of California Press, Berkeley. Carr, S., Francis, M., Rivlin, L. G., & Stone, A. M. (1992). Public space. Cambridge, Angleterre: Cambridge University Press. Castells, M., & Mollenkopf, J. H. (1991). Conclusion: Is New York a dual city? In J. H. Mollenkopf & M. Castells (Eds.), Dual city: Restructuring New York. New York: Russell Sage Foundation. Castells, M. (1989). The informational city: Information technology, economic restructuring, and the urban regional process. Oxford: Blackwell. Castells, M. (1996). The rise of the network society. Malden, MA: Blackwell. Castells, M. (1997). End of millenium. Oxford: Blackwell Scientific Publications. Castells, M. (1998). The Rise of Network Society. Malden, MA: Blackwell. Castells, M. (1999). Information technology, globalization and social development. Discussion paper-United Nations Research Institute for Social Development, (114), 1-15. Castells, M. (2000). The information age: Economy, society and culture. Cambridge, MA: Blackwell. Castles, S. (2002). Migration and community formation under conditions of globalization. International Migration Review, 36, 1143-1168.

References and bibliography | 325

Chapple, K., & Lester, T. W. (2010). The resilient regional labour market? The US case. Cambridge Journal of Regions, Economy and Society, 3(1), 85-104. doi: 10.1093/cjres/rsp031 Cheung, C., & Leung, K. (2012). Chinese migrants’ class mobility in Hong Kong. International Migration IOM. doi: 10.1111/j.1468-2435.2012.00778.x Cho, H. (2011, April). Immigration policy and settlement patterns of migrants in South Korea and Singapore: Understanding ethnic and socio-economic class settlement behaviors in Asia. In PSA Conference 2011. Retrieved October 15, 2012, from http://www.psa.ac.uk/journals/pdf/5/2011/1258_669.pdf Clark, E. (2005). The order and simplicity of gentrification - a political challenge. In R. Atkinson & G. Bridge (Eds.), Gentrification in a global context: The new urban colonialism (pp. 256-264). London: Routledge. Clark, W. A. (1991). Residential preferences and neighborhood racial segregation: A test of the Schelling segregation model. Demography, 28(1), 1-19. Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2013). Applied multiple regression/correlation analysis for the behavioral sciences. Routledge. Coldwell Banker. (1987). National survey of international investment ownership of major office buildings in 19 largest United States. Boston: Coldwell Banker. Cortes, C., Mohri, M., Riley, M., & Rostamizadeh, A. (2008, January). Sample selection bias correction theory. In Algorithmic Learning Theory (pp. 38-53). Springer Berlin Heidelberg. Coy, M. and M. Pohler (2002) Gated communities in Latin American mega cities: case studies in Brazil and Argentina. Environment and Planning B: Planning and Design 29.3, 335–70. Cross, M. (1992). Ethnic minorities and industrial change in Europe and North America. Cambridge, England: Cambridge University Press. Cutler, D. M., Glaeser, E. L., & Vigdor, J. L. (2008). Is the melting pot still hot? explaining the resurgence of immigrant segregation. Review of Economics and Statistics, 90(3), 478-497. doi: 10.1162/rest.90.3.478 De Ruijter, A. (1996). Hybridization and governance. The Hague: ISS. Dodge, M., & Kitchin, R. (2007). Virtual places. In I. Douglas, R. J. Huggett, & C. R. Perkins (Eds.), Companion encyclopedia of geography: From local to global (pp. 519-536). London: Routledge. Duncan, G. J. (2009). New lessons from the Gautreaux and moving to opportunity residential mobility program. Symposium of housing after the fall: reassessing the future of the American dream, 19-20 Feb 2009, The University of California Irvine and The Paul Merage School of Business Center for Real Estate.

326 | References and bibliography

Duncan, O. D., & Duncan, B. (1955). A methodological analysis of segregation indexes. American Sociological Review, 20, 210-217. Dussel, E. (2005) The implications of China’s entry into the WTO for Mexico. Global Issue Papers 24, 1–41. Elander, I., & Blanc, M. (2001). Partnerships and democracy: A happy couple in urban governance? In H. T. Andersen & R. Van Kempen (Eds.), Governing European cities: Social fragmentation, social exclusion and urban governance (pp. 93-124). Aldershot: Ashgate. Eriksen, T. H. (2007). Globalization: The key concepts. Oxford: Berg. Espino, N. A. (2005). Inequality, segregation, and housing market: The US case. In D. P. Varady (Ed.), Desegregating the city: Ghettos, enclaves, and inequality (pp. 145-157). Albany, NY: State University of New York Press. Fainstein, S. S., & Campbell, S. (1996). Readings in urban theory. Cambridge, Mass., USA: Blackwell. Fainstein, S. S., & Harloe, M. (2003). Ups and downs in the global city: London and New York at the millennium. In G. Bridge & S. Watson (Eds.), A companion to the city (pp. 155-167). Malden: Blackwell. Fainstein, S. S. (2001). Inequality in global city-regions. In A. J. Scott (Ed.), Global city-regions: Trends, theory, policy (pp. 285-298). Cambridge, UK: Oxford University Press. Fainstein, S. S. (2010). The just city. Ithaca: Cornell University Press. Fainstein, S. S., Gordon, I., & Harloe, M. (1992). Divided cities: New York & London in the contemporary world. Oxford, UK: Blackwell. Farley, R., Fielding, E. L., & Krysan, M. (1997). The residential preferences of blacks and whites: A four
metropolis analysis. Housing Policy Debate, 8(4), 763-800. Feitosa, F. F., Câmara, G., Monteiro, A. M. V., Koschitzki, T., & Silva, M. P. S. (2004). Spatial measurement of residential segregation. VI Brazilian Symposium on GeoInformatics, 70-83. In VI Brazilian Symposium on GeoInformatics, 22-24 Nov 2004, Campos do Jordão, Geneve: IFIP. Feitosa, F. F., Câmara, G., Monteiro, A. V., Koschitzki, T., & Silva, M. S. (2007). Global and local spatial indices of urban segregation. International Journal of Geographical Information Science, 21(3), 299-323. Feitosa, F. F., Le, Q. B., & Vlek, P. L. (2011). Multi-agent simulator for urban segregation (MASUS): A tool to explore alternatives for promoting inclusive cities. Computers, Environment and Urban Systems, 35(2), 104-115. doi: 10.1016/j.compenvurbsys.2010.06.001 Fernandez, M. (2007, October 15). Study finds disparities in mortgage by race. New York Times. Retrieved September 9, 2011, from

References and bibliography | 327

http://www.nytimes.com/2007/10/15/nyregion/15subprime.html?pagewanted =all Fernandez, S., Brudney, J. L., & Ryu, J. E. (2005, August). Variations in local service delivery: Examining the effects of state-level factors on local government contracting for services. In annual meeting of the American Political Science Association, Washington, DC. Fernandez, S., Smith, C. R., & Wenger, J. B. (2007). Employment, privatization, and managerial choice: Does contracting out reduce public sector employment? Journal of Policy Analysis and Management, 26(1), 57-77. doi: 10.1002/pam.20227 Fincher, R., & Shaw, K. (2009). The unintended segregation of transnational students in central Melbourne [Abstract]. Environment and Planning A, 41(8), 1884-1902. doi: 10.1068/a41126 Findlay, A. (1995). Skilled transients: The invisible phenomenon. In R. Cohen (Ed.), The Cambridge survey of world migration (pp. 515-522). Cambridge: Cambridge University Press. Fischer, C. S. (1976). The urban experience. New York: Harcourt Brace Jovanovich. Fong, E., & Shibuya, K. (2000). The spatial separation of the poor in Canadian cities. Demography, 37(4), 449. doi: 10.2307/2648071 Foreign Policy. (2010, August 16). Global cities index. Retrieved October 7, 2012, from http://www.foreignpolicy.com/articles/2010/08/11/the_global_cities_index_2 010 Forrest, R., La Grange, A., & Yip, N. M. (2004). Neighborhood in a high-rise, high- density city: Some observations on contemporary Hong Kong. The Sociological Review, 50(22), 215-240. Forster, C. (2006). The challenge of change: Australian cities and urban planning in the new millennium. Geographical Research, 44(2), 173-182. doi: 10.1111/j.1745-5871.2006.00374.x

Freund, R. J., Wilson, W. J., & Sa, P. (2006). Regression analysis. Academic Press. Frey, W. H. (2002). Census 2000 reveals new native-born and foreign-born shifts across U.S. (Rep. No. 02-520). Population Studies Centre, the Institute for Social Research, University of Michigan. Friedman, T. L. (2005). The world is flat: A brief history of the twenty-first century. New York: Farrar, Straus and Giroux. Friedmann, J., & Wolff, G. (1982). World city formation: An agenda for research and action. International Journal of Urban and Regional Research, 6(3), 309-344.

328 | References and bibliography

Friedmann, J. (1986). The world city hypothesis. Development and Change, 17(1), 69-83. doi: 10.1111/j.1467-7660.1986.tb00231.x Friedmann, J. (1995). The world city hypothesis. In P. L. Knox & P. J. Taylor (Eds.), World cities in a world system (pp. 317-331). Cambridge: Cambridge University Press. (Original work published 1986) Friedrichs, J., Galster, G., & Musterd, S. (2003). Neighbourhood effects on social opportunities: The European and American research and policy context. Housing Studies, 18(6), 797-806. Gardner, A. (2010). City of strangers: Gulf migration and the Indian community in Bahrain. Ithaca: Cornell/ILR Press. Geniş, S. (2007). Globalization of cities: Towards conceptualizing a new politics of place-making in a transnational era. University of Gaziantep Journal of Social Sciences, 6(1), 59-77.

Gentle, J. E. (2002). Elements of computational statistics. New York: Springer. Gibbons, J. D., & Chakraborti, S. (2003). Nonparametric statistical inference. New York: Marcel Dekker. Goldsmith, W. W. (2000). From the metropolis to globalization: The dialectics of race and urban form. In P. Marcuse & R. V. Kempen (Eds.), Globalizing cities: A new spatial order? (pp. 37-55). Oxford: Blackwell. Goodman, S. N. (1999). Toward evidence-based medical statistics. Annals of internal medicine, 130(12), 995-1004. Gray, J. (1996). After social democracy: Politics, capitalism and the common life. London: Demos. Gregory, D., & Urry, J. (1985). Introduction. In D. Gregory & J. Urry (Eds.), Social relations and spatial structures (pp. 1-8). New York: St. Martin's Press. Gwyther, G. (2005). Paradise planned: Community formation and the master planned estate. Urban Policy and Research, 23(1), 57-72. doi: 10.1080/0811114042000335304 Haan, M. (2005). Summary of: Are immigrants buying to get in?: The role of ethnic clustering on the homeownership propensities of 12 Toronto immigrant groups, 1996 - 2001. Ottawa, Ontario: Statistics Canada. Heckman, J. J. (1979). Sample Selection Bias as a Specification Error. Econometrica, 47, 151–161. Hackworth, J. (2002). Postrecession gentrification in New York city. Urban Affairs Review, 37(6), 815-843. Haila, A. (1991). Four types of investment in land and property. International Journal of Urban and Regional Research, 15(3), 343-365.

References and bibliography | 329

Hamnett, C., & Cross, D. (1998). Social polarisation and inequality in London: The earnings evidence, 1979 - 95. Environment and Planning C: Government and Policy, 16(6), 659-680. Hamnett, C. (1994). Social polarisation in global cities: Theory and evidence. Urban Studies, 31(3), 401-424. Hamnett, C. (1996). Social polarisation, economic restructuring and welfare state regimes. Urban studies, 33(8), 1407-1430. Hamnett, C. (2001). Social segregation and social polarization. In R. Paddison (Ed.), Handbook of urban studies (pp. 162-176). London: SAGE. Hamnett, C. (2003). Gentrification and the middle-class remaking of inner London, 1961-2001. Urban Studies, 40(12), 2401-2426. Hancher, L., & Moran, M. (1989). Introduction: Regulation and deregulation. European Journal of Political Research, 17(2), 129-136. Harloe, M., & Fainstein, S. S. (1992). Conclusions: The divided cities. In S. S. Fainstein, I. Gordon, & M. Harloe (Eds.), Divided cities: New York and London in the contemporary world. Oxford: Blackwell. Harvey, D. (1978). The urban process under capitalism: A framework for analysis. International Journal of Urban and Regional Research, 2(1-4), 101-131. Harvey, D. (1985a). The geopolitics of capitalism. In D. Gregory & J. Urry (Eds.), Social relations and spatial structures (pp. 128-163). London: MacMillan. Harvey, D. (1985b). The urbanization of capital. Oxford: Blackwell. Harvey, D. (1989). The condition of postmodernity (Vol. 14). Oxford: Blackwell. Harvey, D. (2001). Globalization and the spatial fix. Geographische revue, 2(3), 23-

31. Harvey, D. (2005). From globalization to the new imperialism. In R. P. Appelbaum & W. I. Robinson (Eds.), Critical globalization studies (pp. 91-100). New York: Routledge. Harvey, D. (2010). The enigma of capital: And the crises of capitalism. Oxford [England: Oxford University Press. Haussermann, H., & Haila, A. (2005). The European city: A conceptual framework and normative project. In Y. Kazepov (Ed.), Cities of Europe: Changing contexts, local arrangements, and the challenge to urban cohesion (pp. 43- 63). Malden, MA: Blackwell Publishing. Healey, P., Cameron, S., Davoudi, S., Graham, S., & Madanipour, A. (1995). Managing cities: The new urban context. Chichester: Wiley. Healy, E., & Birrell, B. (2003). Metropolis divided: The political dynamic of spatial inequality and migrant settlement in Sydney. People and Place, 11, 65-87.

330 | References and bibliography

Hedin, K., Clark, E., Lundholm, E., & Malmberg, G. (2012). Neoliberalization of housing in Sweden: Gentrification, filtering, and social polarization. Annals of the Association of American Geographers, 102(2), 443-463. Held, D. (2004). Global covenant: The social democratic alternative to the Washington Consensus. Cambridge: Polity. Hoffmann, J. P. (2010). Linear regression analysis: applications and assumptions. Brigham Young University, Provo. Huchzermeyer, M. (2011). Cities with 'slums': From informal settlement eradication to a right to the city in Africa. Cape Town: University of Cape Town Press. Hugo, G. J. (1995). Understanding where immigrants live. Canberra: Australian Government Publishing Service. Hunter, B. H. (2003). Trends in neighbourhood inequality of Australian, Canadian, and United States of America cities since the 1970s. Australian Economic History Review, 43(1), 22-44. doi: 10.1111/0004-8992.00039 Iceland, J., & Scopilliti, M. (2008). Immigrant residential segregation in U.S. metropolitan areas, 1990–2000. Demography, 45(1), 79-94. doi: 10.1353/dem.2008.0009 Ismail, A. (2010). Spatial segregation in global cities: Global pressures and local changes in housing market. In 22nd International Housing Research Conference. Istanbul: ENHR. Ismail, A. (2013). The hybrid outcome of urban change: Global city, polarized city? Glocalism, Journal of Culture, Politics and Innovation., (1). doi: 10.12893/gjcpi.2013.1.5 Jacobs, J. (1992). The death and life of great American cities. Random House LLC.

Originally published in (1961). Janoschka, M. (2002) Urbanizaciones privadas en Buenos Aires: ¿hacia un nuevo modelo de ciudad latinoamericana? [Gated communities in Buenos Aires: towards a new Latin American urban model?] In L.F. Cabrales Barajas (ed.), Latinoamérica: países abiertos, ciudades cerradas [Latin America: open countries, closed cities], Pandora, Guadalajara and Paris. Jargowsky, P. A. (1997). Poverty and place: Ghettos, barrios, and the American city. New York: Russell Sage Foundation. Johnston, R., Poulsen, M., & Forrest, J. (2007). The geography of ethnic residential segregation: A comparative study of five countries. Annals of the Association of American Geographers, 97(4), 713-738. doi: 10.1111/j.1467- 8306.2007.00579.x

Kahane, L. H. (2008). Regression basics. Sage. Kazepov, Y. (2005). Cities of Europe: Changing contexts, local arrangements, and the challenge to urban cohesion. Malden, MA: Blackwell Publishing.

References and bibliography | 331

Keane, D., & McGeehan, N. (2008). Enforcing Migrant Workers' Rights in the United Arab Emirates. International Journal on Minority and Group Rights, 15(1), 81-115. Kesteloot, C., & Cortie, C. (1998). Housing Turks and Moroccans in Brussels and Amsterdam: The difference between private and public markets. Urban Studies, 35(10), 1835-1853. doi: 10.1080/0042098984178 Kesteloot, C. (2000). Post-fordist polarization in a fordist spatial canvas. In P. Marcuse & R. Van Kempen (Eds.), Globalizing cities: A new spatial order? (pp. 186-210). Oxford: Blackwell. Kesteloot, C. (2005). Urban socio-spatial configurations and the future of European cities. In Y. Kazepov (Ed.), Cities of Europe: Changing contexts, local arrangements, and the challenge to urban cohesion (pp. 123-148). Malden, MA: Blackwell Publishing. Kesteloot, C., De Decker, P., & Manço, A. (1997). Turks and their housing conditions in Belgium, with special reference to Brussels, Ghent and Visé. In A. S. Özüekren & R. Van Kempen (Eds.), Turks in European cities: Housing and urban segregation (pp. 67-97). Utrecht: European Research Centre on Migration and Ethnic Relations. Kirkman, T.W. (1996) Statistics to Use. http://www.physics.csbsju.edu/stats/ Kirkwood, B. R. (1988). Essentials of medical statistics. Blackwell Scientific Publications. Kirsch, S. (1995). The incredible shrinking world? Technology and the production of space. Environment and Planning: Society and Space, 13, 529–55. Kleinhans, R. (2004). Social implications of housing diversification in urban renewal: A review of recent literature. Journal of Housing and the Built Environment, 19(4), 367-390. doi: 10.1007/s10901-004-3041-5 Kloosterman, R. C. (1996). Double dutch: Polarization trends in Amsterdam and Rotterdam after 1980. Regional Studies, 30(5), 467-476. Knox, P. L. (1991). The restless urban landscape: Economic and sociocultural change and the transformation of metropolitan Washington, DC. Annals of the Association of American Geographers, 81(2), 181-209. Kraidy, M. M. (1999). The global, the local, and the hybrid: A native ethnography of glocalization. Critical studies in media communication, 16(4), 456-476. Kulish, M., Richards, A., & Gillitzer, C. (2011). Urban structure and housing prices: Some evidence from Australian cities (Research Discussion Paper, RDP2011-03). Reserve Bank of Australia. Retrieved November 9, 2012, from http://www.rba.gov.au/publications/rdp/2011/pdf/rdp2011-03.pdf Larbi, G. A. (1999). The new public management approach and crisis states. Geneva: United Nations Research Institute for Social Development.

332 | References and bibliography

Lattes, A., J. Rodríguez and M. Villa (2002) Population dynamics and urbanization in Latin America: concepts and data limitations. Paper presented at the International Union for the Scientific Study of Population Expert Meeting, Bellagio, 11–19 March. Lee, H. (1997). The employment of foreign workers in Korea: Issues and policy suggestion. International Sociology, 12(3), 353-371. Lee, S. H. (2006). SARS in China and Hong Kong. New York: Novinka. Lefebvre, H. (1991). The production of space (Vol. 142). Blackwell: Oxford. Levy, E. (2001). Decentralization and governance in São Paulo, Brazil. In B. A. Ruble, R. E. Stren, J. S. Tulchin, & D. H. Varat (Eds.), Urban governance around the world (pp. 22-25). Washington, DC: Woodrow Wilson International Center for Scholars, Comparative Urban Studies Project. Levy, F. (1998). The new dollars and dreams: American incomes and economic change. New York: Russell Sage Foundation. Li, Z., & Wu, F. (2008). Tenure-based residential segregation in post-reform Chinese cities: A case study of Shanghai. Transactions of the Institute of British Geographers, 33(3), 404-419. doi: 10.1111/j.1475- 5661.2008.00304.x Little, R. J. A., & Rubin, D. B. (1986). Statistical analysis with missing data. New York, NY, USA: John Wiley & Sons, Inc. Loehnert, S. (2010). About Statistical Analysis of Qualitative Survey Data. Journal of Quality and Reliability Engineering. Hindawi Publishing Corporation. Logan, J. R. (2000). Still a global city: The racial and ethnic segmentation of New York. In P. Marcuse & R. V. Kempen (Eds.), Globalizing cities: A new spatial order? (pp. 158-185). Oxford: Blackwell. (Original work published 1997). MacLeod, G. (2002). From urban entrepreneurialism to a “revanchist city”? On the spatial injustices of Glasgow’s renaissance. Antipode, 34(3), 602-624. Madanipour, A. (2003). Social exclusion and space. In R. T. LeGates & F. Stout (Eds.), The city reader (pp. 181-188). London: New York. Madanipour, A. (2004). Marginal public spaces in European cities. Journal of Urban Design, 9(3), 267-286. doi: 10.1080/1357480042000283869 Madanipour, A., Cars, G., & Allen, J. (1998). Social exclusion in European cities: Processes, experiences, and responses. London: Jessica Kingsley. Maloutas, T. (2007). Segregation, social polarization and immigration in Athens during the 1990s: Theoretical expectations and contextual difference. International Journal of Urban and Regional Research, 31(4), 733-758. doi: 10.1111/j.1468-2427.2007.00760.x

References and bibliography | 333

Marceau, D. J. (1999). The scale issue in social and natural sciences. Canadian

Journal, 25(4), 347-356. Marcuse, P., & Van Kempen, R. (2000). Introduction. In P. Marcuse & R. Van Kempen (Eds.), Globalizing cities: A new spatial order? (pp. 1-21). Malden, MA: Blackwell. Marcuse, P., & Van Kempen, R. (2002). Of states and cities: The partitioning of urban space. Oxford: Oxford University Press. Marcuse, P. (1989). Dual city: A muddy metaphor for a quartered city. International Journal of Urban and Regional Research, 13(4), 697-708. Marcuse, P. (1993). What's so new about divided cities?. International Journal of Urban and Regional Research, 17(3), 355-365. Marcuse, P. (1996). Space and race in the post-Fordist city: The outcast ghetto and advanced homelessness in the United States today. Urban poverty and the underclass, 176-216. Marcuse, P. (2002). The shifting meaning of the black ghetto in the United States. In P. Marcuse & R. V. Kempen (Eds.), Of states and cities: The partitioning of urban space (pp. 109-142). Oxford: Oxford University Press. Marcuse, P. (2003). Cities in quarters. In G. Bridge & S. Watson (Eds.), A companion to the city (pp. 270-281). Malden, Mass: Blackwell. Marcuse, P. (2005). Enclaves yes, ghettos no: Segregation and the state. In D. P. Varady (Ed.), Desegregating the city: Ghettos, enclaves, and inequality (pp. 15-30). Albany, NY: State University of New York Press. Martin, P. L., & Miller, M. (2000). Employer sanctions: French, German and US experiences (International Migration Papers, Publication). Geneve: International Labour Organisation. Martinotti, G. (2005). Employer sanction: French, German and US experience. In Y. Kazepov (Ed.), Cities of Europe: Changing contexts, local arrangement and the challenge to urban cohesion (pp. 90-108). Oxford: Blackwell publishing. Massey, D. (1984). Spatial divisions of labor: Social structures and the geography of production. New York: Methuen. Massey, D. (1985). New directions in space. In D. Gregory & J. Urry (Eds.), Social relations and spatial structures (pp. 9-19). London: MacMillan. Massey, D. S., & Denton, N. A. (1993). American apartheid: Segregation and the making of underclass. Cambridge, MA: Harvard University press. Massey, D. S., & Eggers, M. L. (1990). The ecology of inequality: Minorities and the concentration of poverty, 1970-1980. American Journal of Sociology, 95(5), 1153-88. doi: 10.1086/229425

334 | References and bibliography

Massey, D. S., & Fischer, M. J. (1999). Does rising income bring integration? new results for Blacks, Hispanics, and Asians in 1990. Social Science Research, 28(3), 316-326. doi: 10.1006/ssre.1999.0660 Massey, D. S., & International Union for the Scientific Study of Population. (2008). Worlds in motion: Understanding international migration at the end of the millennium. Oxford: Clarendon Press. Massey, D. S. (1990). American Apartheid: Segregation and the Making of the Underclass. American Journal of Sociology, 96(2), 329-357. doi: 10.1086/229532 Massey, D. S. (1996). The age of extremes: Concentrated affluence and poverty in the twenty-first century. Demography, 33(4), 395-412. Massey, D. S. (2008). Globalization and inequality: Explaining American exceptionalism. European Sociological Review, 25(1), 9-23. Massey, D. S., Arango, J., Hugo, G., Kouaouci, A., Pellegrino, A., & Taylor, J. E. (2009). Worlds in motion: Understanding international migration at the end of the millennium. Oxford: Clarendon Press. Massey, D. S., Gross, A. B., & Shibuya, K. (1994). Migration, segregation, and the geographic concentration of poverty. American Sociological Review, 59(3), 425-445. Matsumoto, Y. (2009, July 12). Living together: segregation and the norm of multicultural urbanism in Tokyo. Lecture presented at Workshop on enclave urbanism as problem or solution in University of Utrecht, Utrecht, the Netherlands. Retrieved October 13, 2012, from http://www.rikkyo.ne.jp/web/ymatsumoto/LivingTogether.pdf McCarthy, E. M. (2006). Handbook of avian hybrids of the world. Oxford: Oxford University Press. McCarthy, J. (1999). Chicago: A case study of social exclusion and city regeneration. Cities, 16(5), 323-331. doi: 10.1016/S0264-2751(99)00030-X McGeary, M. G., & Lynn, L. E. (Eds.). (1988). Urban change and poverty. National Academies. McLuhan, M. (1962). The Gutenberg Galaxy. London: Routledge. McLuhan, M. (1964). Understanding Media: The Extensions of Man. New York: Signet. Mingione, E. (2005). Urban social change: A socio-historical framework of analysis. In Y. Kazepov (Ed.), Cities of Europe: Changing contexts, local arrangements, and the challenge to urban cohesion (pp. 67-89). Malden, MA: Blackwell Publishing.

References and bibliography | 335

Ministry of BZK. (2012). De starter op de Amsterdamse woningmarkt [The starter on the Amsterdam housing] (Publication). General Directorate for Housing, Communities and Integration. Ministry of Interior. Ministry of VROM. (2003). Dossier kwalitatieve woningregistratie 2000 [file qualitative housing register] (Publication). Den Haag: Ministerie van VROM. MMF. (2011). GPCI Gobal Power City Index 2011 (Rep.). Institute for Urban Strategies The Mori Memorial Foundation. Retrieved September 9, 2012, from http://www.mori-m- foundation.or.jp/english/research/project/6/pdf/GPCI2011_English.pdf Mollenkopf, J. H., & Castells, M. (1991). Dual city: Restructuring New York. New York: Russell Sage Foundation. Monkkonen, P., & Zhang, X. (2011). Socioeconomic segregation in Hong Kong: Spatial and ordinal measures in a high-density and highly unequal city (Working paper No. 2011-3). Berkeley: Institute of Urban and Regional Development IURD. Monkkonen, P. (2011). Housing finance reform and increasing socioeconomic segregation in Mexico. International Journal of Urban and Regional Research, 63(4), 757-772. doi: 10.1111/j.1468-2427.2011.01085.x Morgan, B. S. (1975). The segregation of socioeconomic groups in urban areas: A comparative analysis. Urban Studies, 12(1), 47-60. doi: 10.1080/00420987520080041 Morgan, K. (2004). The exaggerated death of geography: Learning, proximity and territorial innovation systems. Journal of Economic Geography, 4(1), 3-21. doi: 10.1093/jeg/4.1.3 Munck, R. (2005). Globalization and social exclusion: A transformationalist perspective. Bloomfield, CT: Kumarian Press. Murray, W. (2006). Geographies of globalization. Routledge. Musterd, S., & Andersson, R. (2005). Housing Mix, Social Mix, and Social Opportunities. Urban Affairs Review, 40(6), 761-790. doi: 10.1177/1078087405276006 Musterd, S., & Fullaondo, A. (2008). Ethnic segregation and the housing market in two cities in northern and southern Europe: The cases of Amsterdam and Barcelona. Architecture, City and Environment, 3(8), 93-115. Musterd, S., & Ostendorf, W. J. (1998). Urban segregation and the welfare state: Inequality and exclusion in western cities. London: Routledge. Musterd, S., & Ostendorf, W. J. (2005). Social exclusion, segregation, and neighborhood effect. In Y. Kazepov (Ed.), Cities of Europe: Changing contexts, local arrangements, and the challenge to urban cohesion (pp. 170- 189). Malden, MA: Blackwell.

336 | References and bibliography

Musterd, S. (1994). A rising European underclass?. Built Environment, 20(3), 185- 191. Musterd, S. (2005). Social and ethnic segregation in Europe: Levels, causes, and effects. Journal of Urban Affairs, 27(3), 331-348. doi: 10.1111/j.0735- 2166.2005.00239.x Nederveen Pieterse, J. (1993). Globalization as hybridization. ISS Working Paper Series/General Series, 152, 1-18. Nederveen Pieterse, J. (2001). Hybridity, so what?: The anti-hybridity backlash and the riddles of recognition. Theory, Culture & Society, 18(2-3), 219-245. doi: 10.1177/026327640101800211 Nederveen Pieterse J. (2009). Globalization and culture: Global mélange. Lanham, MD: Rowman & Littlefield. Ohmae, K. (2005). The next global stage: Challenges and opportunities in our borderless world. Upper Saddle River, NJ: Wharton School Pub. Oliver, A. L., & Montgomery, K. (2000). Creating a Hybrid Organizational Form from Parental Blueprints: The Emergence and Evolution of Knowledge Firms. Human Relations, 53(1), 33-56. doi: 10.1177/0018726700531003 Pacione, M. (2009). Urban geography: A global perspective. London: Routledge. Pahl, R.E. (1988). Some remarks on informal work, social polarization and social structure. International Journal of Urban and Regional Research, 12: 247– 267. Panitch, L. (1998). The state in a changing world: Social-democratizing global capitalism? Monthly Review, 50(5), 11-22. Parakatil, S. (2012, July 11). Defining quality of living. Retrieved October 7, 2012, from http://www.mercer.com/articles/quality-of-living-survey-report-2011 Park, R. E. (1926). The urban community as a spatial pattern and a moral order. In E. W. Burgess (Ed.), The urban community: Selected papers from the Proceedings of the American Sociological Society, 1925. Chicago: University of Chicago Press. Patel, V. R., & Mehta, R. G. (2011). Impact of Outlier Removal and Normalization Approach in Modified k-Means Clustering Algorithm. International Journal of Computer Science Issues (IJCSI), 8(5). Peace, R. (2001). Social exclusion: A concept in need of definition? Social Policy Journal of New Zealand, 16. Peach, C. (1981). Conflicting interpretations of segregation. In P. Jackson & S. Smith (Eds.), Social interaction and ethnic segregation. London: Academic Press. Peach, C. (1996). Good segregation, bad segregation. Planning Perspectives, 11(4), 379-398.

References and bibliography | 337

Peach, C. (2002). Ethnic diversity in the city. In M. Martiniello & B. Piquard (Eds.), Diversity in the city (pp. 21-42). Bilbao: University of Deusto. Peach, C. (2007). Sleepwalking into ghettoisation? the British debate over segregation. In K. Schönwälder (Ed.), Residential segregation and the integration of immigrants Britain, the Netherlands and Sweden (pp. 7-40). Berlin: WZB. Pendall, R. (2005). Does density exacerbate income segregation? evidence from US metropolitan areas. In D. P. Varady (Ed.), Desegregating the city: Ghettos, enclaves, and inequality (pp. 175-199). Albany, NY: State University of New York Press. Piketty, T., & Saez, E. (2003). Income Inequality in the United States, 1913–1998*. The Quarterly journal of economics, 118(1), 1-41. Prescott-Allen, R. (2001). The wellbeing of nations: A country-by-country index of quality of life and the environment. Washington, DC: Island Press. Preteceille, E. (1994). Cidades globais e segmentação social. In L. C. Ribeiro & O. A. Santos Junior (Eds.), Globalização, fragmentação e reforma urbana: O futuro das cidades brasileiras na crise. (pp. 65-89). Rio de Janeiro, RJ: Civilização Brasileira. Qadeer, M. A. (2005). Ethnic segregation in a multicultural city. In D. P. Varady (Ed.), Desegregating the city: Ghettos, enclaves, and inequality (pp. 49-61). Albany, NY: State University of New York Press. Randolph, B., & Holloway, D. (2005). Social disadvantage, tenure and location: An analysis of Sydney and Melbourne. Urban Policy and Research, 23(2), 173- 201. doi: 10.1080/08111470500135136 Raskall, P. (1995, October). Who gets what where? Spatial inequality between and within Australian cities. Paper presented to The Commonwealth Department of Housing and Regional Development Seminar on Aspects of Spatial Inequality, Canberra. Ray, L. J. (2007). Globalization and everyday life. London: Routledge. Reardon, S. F., & Firebaugh, G. (2002). Measures of Multigroup Segregation. Sociological Methodology, 32(1), 33-67. doi: 10.1111/1467-9531.00110 Reardon, S. F., & O'Sullivan, D. (2004). Measures of spatial segregation. Sociological Methodology, 34(1), 121-162. doi: 10.1111/j.0081- 1750.2004.00150.x Reichl, A. J. (2007). Rethinking the dual city. Urban Affairs Review, 42(5), 659- 687. Ribeiro, L. C., & Telles, E. E. (2008). In a historically unequal city. Globalizing Cities, 78.

338 | References and bibliography

Ritzer, G., & Ryan, J. M. (Eds.). (2011). The concise encyclopedia of sociology. Chichester, West Sussex, U.K.: Wiley-Blackwell. Ritzer, G. (Ed.). (2004). Handbook of social problems: A comparative, international perspective. Thousand Oaks, CA: Sage Publications. Roberts, B., & Wilson, R. (2009). Spatial differentiation, inequality and urban policy: The findings. Urban segregation and governance in the Americas, 205-222. Robinson, J. (2002). Global and world cities: A view from off the map. International Journal of Urban and Regional Research, 26(3), 531-554. doi: 10.1111/1468-2427.00397 Robinson, J. (2005). Urban geography: World cities, or a world of cities. Progress in Human Geography, 29(6), 757-765. doi: 10.1191/0309132505ph582pr Rodriguez, A., Swyngedouw, E. and Moulaert, F. (2003). Urban restructuring, social-political polarization and new urban policies, in: F. Moulaert, R. Arantxa and E. Swyngedouw (Eds) The Globalized City: Restructuring and Social Polarization in European Cities, pp. 28– 45. Oxford: Oxford University Press. Rofe, M. W. (2009). Globalisation, gentrification and spatial hierarchies in and beyond New South Wales: The local/global nexus. Geographical Research, 47(3), 292-305. doi: 10.1111/j.1745-5871.2009.00574.x Rolnik, R. (2011). Democracy on the edge: Limits and possibilities in the implementation of an Urban reform agenda in Brazil. International Journal of Urban and Regional Research, 35, 239-255. doi: 10.1111/j.1468- 2427.2010.01036.x Ross, R. J. (2011, July 6). New Orleans as a Rust Belt City? Retrieved from http://www.metropolitiques.eu/New-Orleans-as-a-Rust-Belt-City.html Rubinowitz, L. S., & Rosenbaum, J. E. (2000). Crossing the class and color lines: From public housing to white suburbia. Chicago: University of Chicago Press. Sabatini, F. (2003) The social spatial segregation in the cities of Latin America. Social development strategy document, Inter-American Development Bank, Washington, DC. Sabatini, F., Caceres, G., Cerda, J., 2001, Residential segregation pattern changes in main Chilean cities: Scale shifts and increasing malignancy. In International Seminar on Segregation in the City, 26-28 July 2001, Lincoln Institute of Land Policy, Cambridge. Available online at: www.lincolninst.edu/pubs/dl/615_sabatini_caceres_cerda.pdf Sahlin, I. (2008, June 12). ‘Social housing’ som bostadspolitiskt spoke (’Social housing’ as housing policy ghost). Alba.nu. Retrieved from http://www.alba.nu/?showArticle=19296

References and bibliography | 339

Sakoda, J. (1981). A generalized index of dissimilarity. Demography, 18, 45-50. Sassen, S. (1988). The mobility of labor and capital: A study in international investment and labor flow. Cambridge: Cambridge University Press. Sassen, S. (1990). Economic restructuring and the American city. Annual Review of Sociology, 16(1), 465-490. Sassen, S. (1991). The global city: New York, London, Tokyo. Princeton, NJ: Princeton University Press. Sassen, S. (1994). Cities in a World Economy. Pine Forge Press: London. Sassen, S. (1996). Cities and communities in the global economy: Rethinking our concepts. American Behavioral Scientist, 39(5), 629-639. Sassen, S. (1999). Guests and Aliens. The New Press: New York. Sassen, S. (2000). The state and the new geography of power. In D. Kalb (Ed.), The ends of globalization: Bringing society back in. Lanham, MD: Rowman & Littlefield. Sassen, S. (2001). The global city: New York, London, Tokyo (2nd ed., Original work published 1991). Princeton, NJ: Princeton University Press. Sassen, S. (2002). Locating cities on global circuits. Environment and Urbanization, 14(1), 13-30. doi: 10.1630/095624702101286034 Sassen, S. (2006). Cities in a world economy (3rd ed., Original work published 1994). Thousand Oaks, CA: Pine Forge Press. Sayer, A. (1985). The difference that space makes. In D. Gregory & J. Urry (Eds.), Social relations and spatial structures. New York: St. Martin's Press. Scott, A. J. (2008). Inside the city: on urbanisation, public policy and planning.

Urban Studies, 45(4), 755-772. Sennett, R. (1970). The uses of disorder: Personal identity & city life. New York: A.A. Knopf. Sheppard, E. (2002). The spaces and times of globalization: Place, scale, networks, and positionality. Economic Geography, 78(3), 307. doi: 10.2307/4140812 Short, J.R. & Kim, Y. (1999). Globalization and the city. Addison Silver, H. (1993). National conceptions of the new urban poverty: Social structural change in Britain, France and the United States. International Journal of Urban and Regional Research, 17(3), 336-354. Simon, P. (2005). Gentrification of old neighborhoods and social integration in Europe. In Y. Kazepov (Ed.), Cities of Europe: Changing contexts, local arrangements, and the challenge to urban cohesion (pp. 210-232). Oxford: Blackwell Publishing.

340 | References and bibliography

Sin, C. H. (2002). The interpretation of segregation indices in context: The case of P in Singapore. The Professional Geographer, 54(3), 422-437. doi: 10.1111/0033-0124.00340 Sin, C. H. (2003). The politics of ethnic integration in Singapore: Malay 'regrouping' as an ideological construct. International Journal of Urban and Regional Research, 27(3), 527-544. doi: 10.1111/1468-2427.00465 Singapore Department of Statistics. (1996). General household survey 1995: Socio- demographic and economic characteristics, statistical release 1 (Singapore, Department of Statistics). Siqueira, M. T. (2012). Urban operations: public–private partnerships globalizing São Paulo. Advances in Education in Diverse Communities: Research, Policy and Praxis, 8, 389-413. Skop, E. (2009). Fueling Austin’s boom: The new 21st century immigrant metropolis. In A. Singer, C. Brettell, & S. Hardwick (Eds.), America’s twenty-first century immigrant gateways: Immigrant incorporation in suburbia. Brookings Institution Press. Smith, M. P. (2005). Power in place: retheorizing the local and the global. The urban sociology reader, 241-250. Smith, N. (1996). Spaces of vulnerability: the space of flows and the politics of scale. Critique of Anthropology, 16, 63– 77. Soja, E. (1985). The spatiality of social life: Towards a transformative retheorization. In D. Gregory & J. Urry (Eds.), Social relations and spatial structures (pp. 90-127). London: MacMillan. Soja, E. W. (1989). Postmodern geographies: The reassertion of space in critical social theory. London: Verso. Squires, G. D., Friedman, S., & Saidat, C. (2005). Experiencing residential segregation: A contemporary study of Washington D.C. In D. P. Varady (Ed.), Desegregating the city: Ghettos, enclaves, and inequality (pp. 127- 144). Albany, NY: State University of New York Press. Stanley, B. (2003). "Going global" and wannabe world cities: (Re)conceptualizing regionalism in the Middle East. In W. A. Dunaway (Ed.), Emerging issues in the 21st century world-system: Crisis and resistance in the 21st century world-system (Vol. 1, pp. 151-170). Westport: Greenwood Publishing Group. Statistics Canada. (2006, May 16). Retrieved from http://www12.statcan.gc.ca/census-recensement/2006/index-eng.cfm

Stigler, S. (2008). Fisher and the 5% level. Chance, 21(4), 12-12. Stimson, R. (2001). Dividing societies: The socio-political spatial implications of restructuring in Australia. Australian Geographical Studies, 39(2), 198-216. doi: 10.1111/1467-8470.00140

References and bibliography | 341

Sullivan, H. J. (1987). Privatization of public services: A growing threat to constitutional rights. Public Administration Review, 47(6), 461-467. Swyngedouw, E. C., & Kesteloot, C. (1991). Le passage sociospatial du fordisme à la flexibilité : Une interprétation des aspects spatiaux de la crise et de son issue. Espace Et Sociétés, 54(2), 243-268. Swyngedouw, E., Moulaert, F., & Rodriguez, A. (2002). Neoliberal urbanization in Europe: Large-scale urban development projects and the new urban policy. Antipode, 34(3), 542-577. doi: 10.1111/1467-8330.00254 Swyngedouw, E. (1997). ‘Neither global nor local: “glocalization” and the politics of scale’, in K. R. Cox, (ed.) Globalization: reasserting the power of the local, New York and London: Guilford, pp. 137–66. Taylor, P. J., Catalano, G., & Walker, D. R. (2002). Exploratory analysis of the world city network. Urban Studies, 39(13), 2377-2394. Taylor, P., & Walker, D. R. (2001). World cities: A first multivariate analysis of their service complexes. Urban Studies, 38(1), 23-47. Taylor, P. J., Ni, P., Derudder, B., Hoyler, M., Huang, J., Lu, F., ... Shen, W. (2010, April 13). Measuring the World City Network: New Results and Developments. Retrieved from http://www.lboro.ac.uk/gawc/rb/rb300.html Taylor, P. J. (2000). World cities and territorial states under conditions of contemporary globalization. Political geography, 19(1), 5-32. Taylor, P. J. (2001). Specification of the world city network. Geographical Analysis, 33(2), 181-194. doi: 10.1111/j.1538-4632.2001.tb00443.x Taylor, P. J. (2004). World city network: A global urban analysis. London: Routledge. Taylor, P. J., Ni, P., Derudder, B., Hoyler, M., Huang, J., Lu, F., ... Shen, W. (2011). Global urban analysis: A survey of cities in globalization. London: Earthscan. Telles, E. E. (1995). Structural sources of socioeconomic segregation in Brazilian metropolitan areas. American Journal of Sociology, 100(5), 1199-1223. doi: 10.1086/230636 The State of Texas. (2011). Foreign investment in Texas: The industries and countries leading current growth (Rep.). Retrieved October 10, 2012, from Office of the Governor - Economic Development & Tourism website: http://www.governor.state.tx.us/files/ecodev/Foreign_Investment.pdf Thierstein, A., Lüthi, S., Kruse, C., Gabi, S., & Glanzmann, L. (2008). Changing value chain of the Swiss knowledge economy: spatial impact of intra-firm and inter-firm networks within the emerging mega-city region of Northern Switzerland. Regional studies, 42(8), 1113-1131.

342 | References and bibliography

Thompson, H. (2011). International economics: Global markets and competition (3rd ed.). Singapore: World Scientific. Tickell, A., & Peck, J. (2003). Making global rules: Globalization or neoliberalism? In J. Peck & H. W. Yeung (Eds.), Remaking the global economy: Economic- geographical perspectives (pp. 163-181). London: SAGE. Tiesdell, S., & Oc, T. (1998). Beyond 'fortress' and 'panoptic' cities -- towards a safer urban public realm. Environment and Planning B: Planning and Design, 25(5), 639-655. Timberlake, M., Sanderson, M. R., Ma, X., Derudder, B., Winitzky, J., & Witlox, F. (2012). Testing a global city hypothesis: an assessment of polarization across US cities. City & Community, 11(1), 74-93.

Tomlinson, J. (1999). Globalization and culture. University of Chicago Press. Tu, Y. (1999). Public homeownership, housing finance and socioeconomic development in Singapore. Review of Urban and Regional Development Studies, 11(2), 100-113. doi: 10.1111/1467-940X.00009 Ullah, A. (2012). Bangladeshi migrant workers in Hong Kong: Adaptation strategies in an ethnically distant destination. International Migration IOM. doi: 10.1111/j.1468-2435.2012.00779.x UN-Habitat (2003) Global Report on Human Settlements 2003, The Challenge of Slums, Earthscan, London; Part IV: 'Summary of City Case Studies', pp195- 228. Urry, J. (1985). Social relations, space and time. In D. Gregory & J. Urry (Eds.), Social relations and spatial structures. New York: St. Martin's Press. Van Criekingen, M., & Decroly, J. (2003). Revisiting the diversity of gentrification: Neighbourhood renewal processes in Brussels and Montreal. Urban Studies, 40(12), 2451-2468. doi: 10.1080/0042098032000136156 Van der Waal, J. (2010).“Unravelling the Global City Debate. Economic Inequality and Ethnocentrism in Contemporary Dutch Cities” (Doctoral dissertation, PhD thesis, Erasmus University Rotterdam). Van der Waal, J., & Burgers, J. (2009). Unravelling the global city debate on social inequality: a firm-level analysis of wage inequality in Amsterdam and Rotterdam. Urban Studies, 46(13), 2715-2729. Van Grunsven, L. (2000). Singapore: The changing residential landscape in a winner city. In P. Marcuse & R. Van Kempen (Eds.), Globalizing cities: A new spatial order? (pp. 95-126). Oxford: Blackwell. Van Kempen, R., & Marcuse, P. (1997). A new spatial order in cities? American Behavioral Scientist, 41(3), 285-298.

References and bibliography | 343

Van Kempen, R., & Murie, A. (2009). The new divided city: Changing patterns in European cities. Tijdschrift Voor Economische En Sociale Geografie, 100(4), 377-398. doi: 10.1111/j.1467-9663.2009.00548.x Van Kempen, R., & Özüekren, A. S. (1998). Ethnic segregation in cities: New forms and explanations in a dynamic world. Urban Studies, 35(10), 1631- 1656. Van Kempen, R. (2000). Big cities policy in the Netherlands. Tijdschrift voor economische en sociale geografie, 91: 197–203. doi: 10.1111/1467- 9663.00107 Van Kempen, R. (2005). Segregation and housing conditions of immigrants in Western European cities. In Y. Kazepov (Ed.), Cities of Europe: Changing contexts, local arrangements, and the challenge to urban cohesion (pp. 190- 209). Malden, MA: Blackwell Publishing. Van Kempen, R. (2007). Divided cities in the 21st century: Challenging the importance of globalisation. Journal of Housing and the Built Environment, 22(1), 13-31. Vandell, K. D. (1995). Market factors affecting spatial heterogeneity among urban neighborhoods. Housing Policy Debate, 6(1), 103-139. Varady, D. P. (2005). Desegregating the city: Ghettos, enclaves, and inequality. Albany, NY: State University of New York Press. Wade, R. (1996). Globalization and its limits: Reports of the death of the national economy are exaggerated. In S. Berger & R. P. Dore (Eds.), National diversity and global capitalism (pp. 60-88). Ithaca, NY: Cornell University Press. Walker, R. A. (1985). Class, division of labour, and employment in space. In D. Gregory & J. Urry (Eds.), Social relations and spatial structures (pp. 164- 189). London: MacMillan. Wassmer, R. A. (2005). An economic view of the causes as well as the costs and some of the benefits of urban spatial segregation. In D. P. Varady (Ed.), Desegregating the city: Ghettos, enclaves, and inequality (pp. 158-174). Albany, NY: State University of New York Press. Waters, M. (1995). Globalization. London: Routledge. Wegener, M., Gnad, F., & Vannahme, M. (1986). The time scale of urban change. In B. Hutchinson & M. Batty (Eds.), Advances in urban systems modelling (pp. 175-197). Amsterdam: North Holland. Weinberg, S. L., & Abramowitz, S. K. (2002). Data analysis for the behavioral sciences using SPSS. Cambridge University Press.

Weisberg, S. (2014). Applied linear regression. John Wiley & Sons.

344 | References and bibliography

Weller, S., & Van Hulten, A. (2012). Gentrification and displacement: The effects of a housing crisis on Melbourne's low-income residents. Urban Policy and Research, 30(1), 25-42. Wessel, T. (2000). Social Polarisation and Socioeconomic Segregation in a Welfare State: The Case of Oslo. Urban Studies, 37(11), 1947-1967. doi: 10.1080/713707228 WHO | Glossary of globalization, trade and health terms. (n.d.). Retrieved September 09, 2011, from http://www.who.int/trade/glossary/en/ Wills, J., Datta, K., Evans, Y., Herbert, J., May, J., & McIlwaine, C. (2010). Global Cities at Work. New Migrant Divisions of Labour. London: Pluto. Wilson, W. J. (1987). The truly disadvantaged: The inner city, the underclass, and public policy. Chicago: University of Chicago Press. Wilson, W. J. (1996). When work disappears: The world of the new urban poor. New York: Knopf. Wirth, L. (1928). The ghetto. Chicago: The University of Chicago Press. Wong, D. D. (2004). Changing local segregation of selected U.S. metropolitan areas between 1980 and 2000 (Rep.). Retrieved http://gesg.gmu.edu/seg/change_locseg.pdf Wong, D. W. (1993). Spatial indices of segregation. Urban Studies, 30(3), 559-572. doi: 10.1080/00420989320080551 Wong, D. W. (1998). Measuring multi-ethnic spatial segregation. Urban Geography, 19(1), 77-87. Wong, D. W. (2002). Modeling local segregation: A spatial interaction approach. Geographical and Environmental Modelling, 6(1), 81-97. doi: 10.1080/13615930220127305 Wong, D. W. (2003). Implementing spatial segregation measures in GIS. Computers, Environment and Urban Systems, 27(1), 53-70. doi: 10.1016/S0198-9715(01)00018-7 Wong, D. W. (2005). Formulating a general spatial segregation measure. The Professional Geographer, 57(2), 285-294. doi: 10.1111/j.0033- 0124.2005.00478.x Wong, D. (2009). The modifiable areal unit problem (MAUP). In A. Fotheringham, & P. Rogerson (Eds.), The SAGE handbook of spatial analysis. (pp. 105- 125). London: SAGE Publications Ltd. doi: http://dx.doi.org/10.4135/9780857020130.n7 Xu, J., Yeh, A., & Wu, F. (2009). Land commodification: Mew land development and politics in China since the late 1990s. International Journal of Urban and Regional Research, 33(4), 890-913.

References and bibliography | 345

Yeung, H. W. (2002). The limits to globalization theory: A geographic perspective on global economic change. Economic Geography, 78(3), 285. doi: 10.2307/4140811 Zlotnik, H. (1993). South-to-north migration since 1960: The view from the south. In General Population Conference, Montreal 1993 (pp. 3-14). Liege: International Union for the Scientific Study of Population.