Residential Segregation in a Highly Unequal Society: Istanbul in the 2000s

Abstract

Contrary to trends in many European countries, income inequality in Turkey, measured by the Gini coefficient, has declined between 1994 and 2014, with a small but consistent increase since then. Turkish income inequality is among the highest in OECD countries, with levels not lower than 0.4. This chapter will examine residential socio-economic segregation in Istanbul against the backdrop of this relatively stable and high-income inequality. The chapter shows signs that residential segregation is on the rise. Istanbul has undergone a radical change in the 2000s thanks to active intervention by the state in the real estate market by opening up large pieces of land in the outskirts and gentrifying inner-city areas once occupied by unauthorized settlements that once were home to the poor. Dynamics of urban development, fueled by rapid urban sprawl in peri-urban areas and ceaseless gentrification of inner-city areas, gave way to diverse patterns of segregation depending on the already existing divisions and physical geography of cities. Given the lack of neighbourhood level data on either occupations or income, this chapter analyses segregation through indices based on fertility and educational level, which we know from detailed household microdata are closely correlated with income. On the basis of 2000 and 2017 neighbourhood data, we show that in Istanbul, there is a clearly visible pattern where the poor are progressively pushed further to the city limits, while some parts of built-up areas once home to middle classes, were recaptured by the poor. The result in some parts of the city is a juxtaposition of seemingly conflicting patterns: parts of the inner city were reclaimed by the poor while some parts were gentrified led by the nascent urban elite. The urban periphery was partly occupied by the bourgeoning middle classes and was also home to the urban poor who were displaced by urban transformation projects.

1 Introduction

Turkey is a land of vast and enduring inequalities. Ever since the first reliable studies were made on income inequality in the late 1960s, the Gini coefficient has not fallen below 0.4 except a few years in the early 2000s, making Turkey one of the most unequal countries in the OECD, and placing it somewhere between European countries with relatively low levels of inequality and Latin American countries where income inequality has been notoriously high. In the face of a state unwilling to take active measures to reduce inequality, the society devised many ingenious ways, especially in the urban labour and property markets, to cope with high levels of inequality and to keep at bay the tensions that may arise therefrom. Things went well till the 2000s and the poor were able to find a place to live for themselves in the rapidly bourgeoning urban property market thanks to networks of solidarity they set up, spanning in most cases the boundaries between the formal and informal and authorised and unauthorised. Thanks to the innovative survival strategies of the urban poor, the problems that are often associated with high inequality and residential segregation went unnoticed for most of society.

The 2000s have been frantic years in terms of urban development in all large cities in Turkey, a period when the mechanisms devised by the urban poor proved inefficient, and the state that had been a silent partner became an active agent in the housing and urban property market. Given the voracious desire on the part of the state to appropriate ever-increasing land values in many gargantuan projects, the poor found it increasingly difficult to maintain their positions in the city and were driven to the city limits where they had to compete with many other agents for land and shelter, finding themselves unable to use space as means of power as they had done in previous periods.

This chapter is an attempt to assess how already high levels of residential segregation have changed in line with the massive urban restructuring of the 2000s. The objective is to understand how the construction frenzy and urban sprawl of the 2000s were articulated with the existing pattern of residential segregation inherited from earlier periods of urban development. Starting with a summary of the basic contours of urban development in Turkey, the chapter discusses the problems of studying segregation with limited available data in Istanbul, where the physical geography has a strong impact on almost every urban process, including residential segregation.

2 Background: Frantic years of 2000s

Compared to the 2000s, urban development in Turkey in the pre-2000 period was certainly a calm process, which ensured that some of the big problems encountered by similar countries undergoing a rapid process of urbanisation were solved without causing large-scale unrest. The basic characteristics of this period can be described succinctly in a few statements: An informal job and urban property market which guaranteed that the ceaseless flux of newcomers would easily find a job and a shelter thanks to the networks of solidarity they set up, though the jobs they found were as shaky as the squatters they built up for themselves on vast state land, mostly at the periphery of cities; a populist-clientelist state apparatus, incapable of responding to the needs of these newcomers, that happily turned a blind eye, in exchange for their votes, to people who squat illegally on state land or who run informal businesses (Işık and Pınarcıoğlu 2009). Things did really go well both for the urban poor and the state till the late 1990s; by exploiting the opportunities of the informal labour and real estate markets, the urban poor were able to transfer part of their poverty conditions to newcomers and thus to climb up the social ladder into the ranks of the rapidly growing middle class. This undeclared alliance between the state and the urban poor secured a decent place for the latter while the politicians were able to maintain their position, thus laying the foundations for a more repressive and populist political climate in the 2000s.

This tacit alliance, whereby the urban poor made use of the informal labour and property markets as leverage to cope with the prevailing levels of inequality, came to an end in the late 1990s due to several intertwined processes. The reasons behind the collapse of this alliance and the ensuing changes that Turkish society has gone through in the 2000s are too large to be summarised in a few words. The 2000s can be best described by rising to the power of the Justice and Development Party (JDP), an Islamist-leaning party, and their desire to mould the society in their vision, in a way described by some rightfully as the marriage of Islamism and neo-liberalism (Gürcan and Peker 2015; Karaman 2013a; Atasoy 2009) with strong populist overtones. The JDP and local governments launched an ambitious ‘urban transformation programme’ through public–private partnership schemes in most large Turkish cities, under the strong support of the state-run housing administration (TOKI). TOKI undertook huge housing projects at the outskirts of cities and played a leading role in the clearance of formerly unauthorised housing areas in inner cities where land prices had soared (Türkün 2011; Karaman 2013b). TOKI became the most powerful agent in the real estate market, by acquiring the right to transfer public land to private developers for the construction of housing estates on the outskirts through public–private partnership scheme with titles delivered to prospective buyers. In a country where owner-occupation is the rule (nearly 2/3 of the existing stock is owner-occupied) and public housing ownership is an exception, projects undertaken through public–private partnership programmes such as those by TOKI fuelled the fragmentation of the housing market. It is estimated that some 66 million square meters of state land were transferred to TOKI between 2002 and 2008. In inner-city areas, TOKI helped clear former squatter areas and residents’ property rights were transferred to housing projects far from the city centre. When considered in combination with the privatisation frenzy of the JDP and their obsession with grandiose urban projects (including renovation of former shipyards, ports and railway stations and large-scale urban renewal projects, shopping malls, colossal luxury hotels and gated communities), this heralded the end of a long period in Turkish history of urbanisation where the migrating masses had used their networks of solidarity as a survival strategy under the harsh conditions of persistent inequality. Devoid of any chance and means to secure a place for themselves in the property market, the poor were gradually pushed further out of the city and found only limited space of manoeuvre in inner-city areas where the urban elites had no interest for. The urban middle classes seized every opportunity they found and were the main beneficiaries of new housing schemes in both central and peri-urban areas where the poor had been displaced. The nascent urban elites, on the other hand, found themselves unrivalled in both the peri-urban and inner-city areas and continued to live in now poor-proof neighbourhoods.

Figure 15.1 shows the evolution of the Gini coefficient since the late 1980s. With the exception of 2005 and 2014, the Gini has been above 0.4, exhibiting an upward trend in the last few years. In the case of Istanbul, for which we have reliable data only after 2006, the Gini coefficient has been somewhat lower than the one for Turkey but started to rise sharply after 2014. In Istanbul, income inequality seems to have registered a record jump after 2015 and surpassed the Gini of Turkey.

Fig. 15.1

Sourcea Figures before 1986 from TurkStat Household Income and Consumption Expenditures Survey, 1986

Income inequality in Turkey and Istanbul: Gini Coefficient*.

3 Studying Segregation in Istanbul

The study of segregation in Istanbul is beset by a number of difficulties. The first set of problems originates from the fact that Istanbul is literally a city in flux, a site of a never-ending process of creation and destruction. Its population rose from 7.2 million in 1990 to 11 million in 2000 and finally to 15 million in 2017. This refers to an annual growth rate of no less than 2% on most occasions, a rate which is difficult to imagine for many cities of the West. In a similar fashion, the number of neighbourhoods increased from 550 in 2000 to some 780 in 2017, making the measurement of segregation and comparisons over time an arduous task.

The second set of difficulties arises from the unique geographical setting of the city. Built on two continents (Europe and Asia), with an extremely busy seaway (Bosporus) in between connecting Balkan countries and Russia to the rest of the world, Istanbul is divided into two distinct geographical units. For the study of settlement and segregation patterns, this means that Istanbul is destined to be a polycentric city, with each geographical unit having its core and peripheral areas (Dökmeci and Berköz 1994). The uneven distribution of population between these two parts (nearly two-thirds living on the European side and one-thirds on the Asian side) generates an immense volume of traffic and development pressure for the whole city. Fifteen million people squeezed on a piece of land slightly larger than 5,000 km² and the existence of numerous naturally and historically important sites in and around the city make land the most precious commodity for Istanbul. Fierce competition for land contributes to the emergence of a segregation pattern resembling a patchwork, especially at the periphery.

Though not peculiar to Istanbul but common to all cities of Turkey, the main challenges arise to those interested in the study of segregation, mostly due to the limited availability of geocoded data. Given the absence of appropriate spatial data and the fact that most cities are in a constant process of building and destruction, one has to find some innovative and particularly indirect ways to decipher the pattern of segregation observed in Turkish cities. It is to this problem of data and methodology that I turn briefly in the following section.

4 Data and Method

Although Turkey has gone a long way in the 2000s in assuring the quality of statistical data, there is still a long way to go in what may be called small area statistics. Designation of functional urban areas and NUTS areas has yet to be completed. The only spatial unit where data is available for cities is the ambiguous level of neighbourhoods (mahalle). Mahalle does not have a clear definition in the administrative system, nor are there any known limits to its population and size. In the case of Istanbul, the population of neighbourhoods in 2017 varied between 50 and 93,000, with an average of 15,600. Furthermore, local governments may—and usually do—change the number and boundaries of neighbourhoods without any restriction. This generates problems concerning the comparability of datasets, an issue discussed below in relation to dissimilarity indices.

Neighbourhood level data are gathered through the Address-Based Population Registration System (ABPRS). The ABPRS provides only very basic information about the population-merely the educational level and age distribution by gender. No information is available on the occupational status of the population, or on working conditions. With this very limited data, one has to be creative in analysing segregation patterns in Istanbul. With the purpose of finding out which of the few variables available at the neighbourhood level are correlated with income and social status, and thus drawing out the segregation map of Istanbul, 2016 household budget survey microdata sets were used. This data from TurkStat is used for analysing the income distribution and consumption, and information is available on households and individuals. If any of the available ABPRS data is correlated with income and/or socioeconomic status, we can use it as a proxy for socioeconomic status and use it to study residential segregation.

Of the information accessible through the ABPRS, the most promising defines education data, as there is a strong positive correlation between educational attainment and income level. Figure 15.2 summarises the relationship between income and other sets of variables to find out the extent to which they can be used as a proxy for income. The top left figure shows the educational attainment of the adult population (population over the age of 25) by income deciles. Here the educational level is grouped into four categories: no formal education represents those who have not completed any formal education; the second category represents those who have completed only the compulsory 8-year elementary school; and the other two categories of education are for those who have completed secondary school and university. The figure shows a clear positive correlation between educational level and income level. The share of university graduates is higher than 56% in the 10th decile while it is as low as 1.2% at the lowest. The percentage of those who have completed only compulsory education shows the opposite trend-high in low-income groups and low in high-income groups (43% and 17% respectively). Of course, the relationship between educational status and income or in general socio-economic status is a complex one and is not as straightforward as this figure might imply. But the analyses show that educational level can safely be used as a proxy for the level of income as there is a clear tendency for income to rise with rising education level.

Fig. 15.2

Source Calculated from TurkStat, 2016 Household Budget Survey microdata set

Selected variables and income groups by deciles, 2016.

The other variables that can be used as a proxy for income and socio-economic status are also shown in Fig. 15.2. The top right graph shows the relationship between household size and disposable household income by deciles. The better-off households are usually smaller, while low-income groups tend to live in larger households. One to three-person households make up two-thirds of the highest income decile, but their share drops to one third in the lowest decile. The fact that well-to-do families are smaller in size may be regarded as an indication of differences in fertility between low and high-income households. The bottom right graph shows the relationship between the child–woman ratio (CWR) and income level. CWR is defined as the number of children under the age of 5 per 1000 women of reproductive age (ages 15–49) in a population in a given year. It may be regarded as an indicator of recent fertility net of child mortality. A crude and indirect measure of fertility easily obtainable from census data, CWR is frequently used in fertility studies in the absence of such specific measures as total or age-specific fertility rate. The graph shows that there is a strong negative relationship between CWR and income; the lower the income, the higher the level of fertility. As a matter of fact, CWR in the highest decile is smaller than half its level in the lowest decile (519 vs 224, respectively).

The fertility indicator discussed above brings to the fore, though indirectly, the problem of women’s position in Turkish society. Turkey’s performance has been very poor in terms of gender equality, ranking 130 among the 149 countries for which the 2018 Global Gender Gap Index is available (World Economic Forum 2018). We know that the participation of woman in the labour force has been traditionally low in Turkey (Göksel 2013) compared to countries of a similar income level and female education in Turkey has lagged considerably behind that of men (Işık and Pınarcıoğlu 2006). Not surprisingly, the participation rate of a woman in education exhibits striking variations between income groups, with the percentage of 30+ aged women with a university degree as high as 49% in the wealthiest decile and as low as 0.1% in the poorest decile. Therefore, variables relating to education, especially for women’s education, are adequate proxies to understand segregation along the lines of income and socioeconomic status. For the analyses, university graduates are compared to those without formal education to delve into divisions characterising Istanbul in the 2000s. Other variables referred to above are used for the classification of neighbourhoods.

5 Results

Figure 15.3 shows maps with the location quotients (LQ) for university graduates and the population without formal education for 2000, 2010 and 2017. LQ maps give a cursory idea about the segregation pattern of Istanbul. The wealthy segments of the population have settled along the sea, while the lowest income groups (i.e. those groups with low levels of educational attainment) seem to have found a place for themselves at the periphery. These maps reveal that the proximity to the sea is the basic determinant of the pattern of segregation. Here one can also see the impact of city’s unique physical geography on segregation patterns, where both Asian and European parts of Istanbul have their own rich, poor and middle-class areas. With some exceptions, especially in the historic parts of the city, better-off groups have settled in neighbourhoods along the sea coast, with low-status groups settling in areas at the fringe. Although at first glance, the pattern of the well-to-do living along with the coast, the poor at the periphery and a mixture of middle-class neighbourhoods between the two does not seem to have changed very much between 2000 and 2017, there are some very important differences that need further attention. I turn to this issue below when I discuss the segregation along the lines of socioeconomic status, based on a grouping of neighbourhoods.

Fig. 15.3

Source TurkStat, Address-Based Population Registration System

LQ maps for university graduates and population with no formal education.

Table 15.1 shows the dissimilarity indices (DI) for selected education groups in Istanbul for 2000, 2010 and 2017. The table reveals some crucially important features of segregation. Of the groups considered, the ones that are most unevenly distributed across the city are university graduates. The largest segregation is between university graduates and people with no formal education (0.49 in 2000, rising to 0.52 in 2010 and then dropping to 0.44 in 2017), the main reason for which is the sharp decline in the percentage of people with no formal education. As can be seen from the table, groups with a low level of education tend to live close to each other in space, but they live separated from those with a high level of education. What is also striking is that the indices for upper-status groups are higher in all cases than those for lower status groups, an indication of the stronger tendency on the part of the higher educated to concentrate in areas where their percentage is already high. Another tendency that the DI figures reveal is the fact that segregation along the lines of education seems to have slightly dropped in the 17 years from 2000 to 2017 on both continents, probably an indication that the neighbourhoods that are being gentrified get more mixed, a tendency which is observed on both European and Asian sides of the city.

Table 15.1 Dissimilarity indices for Istanbul, 2000, 2010 and 2017

This conclusion must, however, be treated with caution. Although the dissimilarity index is argued to be compositionally invariant (i.e. not affected by the overall composition of the population), Taylor et al. (2000) and Gorard and Taylor (2002) have reported cases where DI cannot be considered fully composition invariant; there are cases when the index is altered by changing composition of a population over the years considered. We should note the fact that during the period under consideration, the educational categories studied here recorded some great changes with, for instance, the share of university graduates rising from 8% in 2000 to 19% in 2017 and that of people with no formal education falling from 35 to 11% in the same period.

There are also sufficient grounds to claim that DI performs very poorly in making temporal comparisons. It is very sensitive to changes in the number and sizes of areas compared (Simpson 2007). We know that the number of neighbourhoods rose from 550 to 780, from 2000 to 2017, with the expansion of the city towards the unoccupied and low-density areas. This rise in the number of neighbourhoods is not only due to the expansion of the city towards the periphery but also to further division of existing ones into smaller units; some made for gerrymandering in elections. Therefore, the initial impression that one can get from falling DI might be somewhat misleading since there is enough evidence that changing borders and the increasing number of neighbourhoods do affect the measurement of segregation. In any case, however, there is a considerable spatial division in Istanbul along the lines of educational status. And we must also note that this segregation is not a new phenomenon peculiar to the 2000s, but something Istanbul obviously inherited from earlier periods (for segregation in 1990 see Işık and Güvenç 2002; and for 2000 see Işık and Pınarcıoğlu 2008, 2009). There is, however, an important point that DI cannot reveal—the clear shift in the spatial pattern of segregation.

5.1 Classification and Mapping of Neighbourhoods

Given that neighbourhood level data is scarce for Istanbul, three variables are used to classify neighbourhoods to gain a better understanding of how spatial segregation patterns have changed between 2000 and 2017. (1) The percentage of those with a university degree in the total adult population (population older than 25 years of age) of each neighbourhood; (2) child–woman ratio (CWR) calculated as the number of children under age 5 per 1000 women of ages between 15 and 49; and (3) the percentage of women with a university degree in the female population of 30 years and over in each neighbourhood.

Using these three variables, neighbourhoods were sorted for each variable from the lowest to the highest value and then divided into quintiles. Each quintile was then assigned a score from 1 to 5, depending on its ranking. In the case of university graduates (both total and female), the lowest quintile was given a score of 1 and the highest quintile a score of 5, since both are positively correlated with income. In the case of CWR, the scoring was reversed (1 to the highest quintile and 5 to the lowest quintile) as CWR is negatively correlated with income. Finally, the individual scores were added up without assigning any weight to reach an overall score between 3 and 15, representing the socioeconomic status of each neighbourhood. Depending on their scores, neighbourhoods were then divided into 5 categories named low, low-middle, middle, middle-high and high-status neighbourhoods. The results are shown in Fig. 15.4 for 2000, 2010 and 2017.

Fig. 15.4

Socioeconomic classification of neighbourhoods

5.2 From 2000 to 2017: Exodus of the Poor

The segregation map for 2000 represents a typical pattern of segregation that could be found not only in Istanbul but also in most other Turkish cities. The basic characteristics of this segregation pattern can be summarised in a few words: concentration of affluence in and around the prestigious areas of the city (in the case of Istanbul, along the sea coast) and clustering of the poor at the fringe that is of no interest to the better-off classes. It must be noted that in the case of Istanbul, these areas are mostly unauthorised squatter housing areas built in the 1960s and 1970s. Another important aspect of this segregation pattern is the existence of middle-income groups acting as a buffer between low- and high-income groups. The lower and upper ends of the social ladder do not seem to touch each other in residential neighbourhoods. In few instances where they seem to have a direct contact, there is almost always a physical barrier: D-100 highway on both the Asian and European side is a clear separator between the rich and the poor (Işık and Güvenç 2002). There are very few examples of the rich and the poor living in neighbouring areas. Poor areas next to rich areas can be found on both sides of the Bosporus, but they are more visible on the Asian side of Istanbul. They are the remnants of squatters built at times when these areas were of little value to the wealthy, but their relative position in the city has changed over time due to transformation in the late 1990s.

The 2000 segregation map shows the first signs of what we may call the flight of the upper classes from their traditional residential neighbourhoods at the coastline. This flight of the upper classes started in the 1990s but was until then of sporadic and limited nature. The growing tendency on the part of high-income groups to settle at the urban fringe, particularly in areas known as ‘gated communities’ (Geniş 2007; Tanülkü 2012, 2013) is still invisible in the 2000 map because neighbourhoods on the fringe are too large for the impact of such scattered communities to be observable. Such flight of high-income groups is only visible if it is large enough to make up a significant portion of a neighbourhood. The most notable example of such movements is a new settlement known as Bahçeşehir (Garden City) on the far European edge of the city (see the westernmost part of 2000 map). As a harbinger of what was to come in the 2000s, this settlement was built in the late 1980s with generous credit support from now-abolished state-owned real estate banks and became a symbolic example of the flight of the middle and upper classes from the built-up areas of the city.

A pattern where the wealthy appropriated the privileged parts of the city, the poor lived exclusively in clusters at the outskirts and the middle classes acted as a buffer between the two was almost a trademark of most Turkish cities in the 1990s and early 2000s (cf. Ataç 2017). But this pattern was subject to a brutal and abrupt change at an unforeseen speed in the 2000s, with some disastrous results for the urban poor and ruining everything we thought we knew about Turkish cities. A glimpse of the changes that took place in the 2000s can be caught even with a cursory comparison of the maps for 2000 and 2010. Although the basic pattern of the rich along the sea coast and the poor at the outskirts still remains, there are important trends that shed light to the defeat of the poor, the zeitgeist of the period. Here we can see the first and undeniable traces of what we may refer to as the (a) displacement of the urban poor from most of their previous locations and their exodus to even more remote parts of the city; (b) consolidation of the urban elites in, around and along with the privileged areas by cleansing the remnants of the poor that had settled there in earlier periods; (c) large-scale gentrification of some inner-city areas with some conflicting results; and (d) occupation by the poor or lower classes of some inner-city areas abandoned by the wealthy, the so-called degradation of some parts of the city.

It is not surprising to see that some of the largest transformation projects took place under the generous support of the state in areas with high accessibility between the TEM and D-100 highways on both sides of the metropolis. Parts of these areas had been subject to unauthorised housing development in the 1970s and 1980s. They were the typical examples of pre-2000 type of unauthorised urban development—built on occupied state land through ad hoc networks working along the lines of ethnicity, religious affiliation and fellow-townsmanship, meaning that they had some degree of heterogeneity within themselves before the onset of large-scale projects of 2002. Though they had lost their initial characteristics and turned into more heterogeneous settlements by time in terms of their inhabitants, they were without exception on pieces of land that were far more valuable than their initial value. This ‘non-correspondence’ between their relative position within the city—hence their soaring land values—and the characteristics of their inhabitants paved the way for a large-scale and devastating intervention on the part of the state and local authorities. Ayazma transformation project for about 20.000 inhabitants on the European side (see Lovering and Türkmen 2011) and Maltepe-Başıbüyük projects for about 15.000 inhabitants on the Asian side (Kuyucu and Ünsal 2010; Karaman 2013b) are among the best documented and best-known examples of such projects. In all these projects, former residents were offered housing units at the outskirts and thus forced to live in newly built residential areas on the far edges of the city. These were, with no doubt, cleansing projects whereby the poor were pushed further away from the city.

A similar process of sterilisation is also visible in areas close to the prestigious areas inhabited by the upper classes, particularly on sites overlooking the Bosporus on both continents. A comparison of 2000 and 2010 maps shows how the upper classes expanded geographically and how pockets of the poor were eradicated along with the areas between the two Bosporus bridges. These areas, too, had been subject to extensive squatter development in earlier periods. The transformation projects in the Beykoz district on the Asian side and the infamous Armutlu project are among the many examples of such projects (Ergun 2004). It must be noted that the cleansing of hills overlooking the Bosporus served more to the upper classes as opposed to the ones described above, such as Ayazma, where middle and upper-middle-income groups made the most use of.

In the historic parts of the city, several renewal projects took place and two processes were at work there; large-scale projects launched by the state and projects of a minor scale initiated by the nascent urban elite. As an example of the latter, we can refer to the Cihangir area (Lovering and Türkmen 2011), a once degraded site, opposite to the historic peninsula in an area known as Pera occupied in the nineteenth and early twentieth century by non-Muslim minorities. This area became the focus of culture-led regeneration (Günay and Dökmeci 2012). The gentrification of the Cihangir area was a slow and self-paced process, while the ones instigated by the state were brutal and devoid of any room for the residents to get their share of rising values. Of these renewal projects, that in Sulukule, famous for being the first neighbourhood in the world to be permanently settled by the Romani people, became the stage of a widespread protest and was even mentioned in EU’s 2006 annual progress report as an example of the displacement of Roma population (CEC 2006). All these protests were in vain, and the neighbourhood was evacuated by police force, and Sulukule residents found themselves exiled to a distant part of the metropolis where they had not seen before (Uysal 2012; Schoon 2014).

Amidst this chaos, parts of the historic peninsula, once the home of middle and upper-income groups, were reclaimed by low-income groups (Dinçer 2011). The part played by Syrian refugees fleeing from the civil war must be stressed in this context. According to official figures, Turkey is home to 3.6 million Syrian refugees living mostly in big cities with many unsettling impacts not just for Istanbul but for many other cities. Of these refugees, more than 15% (560,000) live in Istanbul, usually in areas where rents and living costs are low and accessibility is high (Kılıç et al. 2019). Those parts of the historic peninsula for which the upper classes had no interest for were thus appropriated by a mixture of low-income groups coalescing with Syrian refugees.

All these huge projects of transformation at the periphery and gentrification projects of various scales left behind a patchwork metropolis in some parts of the city. Especially in the peri-urban areas where land is scarce and competition is high, the transformation of the fringe was sporadic and remained incomplete, producing a patchy pattern in some parts of the metropolis resembling the one emphasised by Florida (2017) as an important component of the new urban crisis. An examination of the 2017 segregation map in Fig. 15.4 would reveal that the tendencies that started in the early years of the first decade of the 2000s produced a significantly different pattern. Although the parts of the city close to the most valuable asset—namely, to the sea coast—are still colonised by the wealthy, the new Istanbul at the end of the second decade of the 2000s can be best described by the almost total absence of the poor in inner-city areas close to or easily accessible from the city centre. The poor have now been driven away to the edges of the city, with little contact to the rest of Istanbulites. The heavy-handed attitude of the state and local governments in most of the transformation projects and their determination to break any resistance using even police force when they felt necessary must be regarded as an indication of insatiable desire to redistribute urban rents, which in previous periods had been a lifesaver for newcomers and the urban poor. It is to this issue of redistribution of urban rents and consequences flowing out of it that I turn in the concluding remarks.

6 Concluding Remarks

Turkey was and still is a country of high inequality. Although the level of inequality did not change significantly on the surface in the last few decades, the underlying class equations and the mechanisms whereby society coped with these inequalities did. Up until the 2000s, the urban poor found the means to appropriate soaring property values in prospering metropoles and make use of these opportunities as leverage for social climbing. It is this informal redistribution mechanism in both urban labour and property markets that guaranteed that all social tensions related to high levels of inequality were kept at bay and masses joining the ranks of the urban poor maintained their hopes for reaching the upper echelons of social divisions. In this game of inequality, the state acted as an accomplice, turning a blind eye to the existence of markets in the undefined shady zone between the legal and illegal, the formal and informal. The geography of inequality changed as well, with the rich concentrating along the sea coast as they did in previous periods, plus air-tight pockets in gated communities on the edges and creating spaces for themselves in parts of the centre. The final pattern is a patchwork metropolis as opposed to the previous one characterised by clear lines of demarcation in space between social groups.

The resulting pattern of segregation that we see for the year 2017 is the outcome of several intertwined processes that had their origin in the late 1990s and culminated in the 2000s:

• (a) State-led eradication of former squatter areas especially in peripheral parts of the city where accessibility and thus property values are high;

• (b) Selective gentrification of inner-city areas, some under the auspices of the state, others self-initiated and infiltration of the nascent urban elite into former middle-class areas; and,

• (c) In any case, apathy and even brutality on the part of the state toward the losers, that is the former residents of project areas; and thus, the cleansing parts of the periphery and the inner city from the urban poor, a process I referred to as the exodus of the urban poor to the remote parts of the metropolis where they can have little contact with the city.

The extent to which this change in segregation pattern can be attributed to income inequality is a question for which I have no ready-made answer as high inequality has been an integral aspect of Turkish society for a long time. It is, however, certain that a metropolis where the poor have been driven out of sight, a process by which the ability of the poor to use space as means of power has been castrated, is a new phenomenon even for Turkey. This is something we all have to think about.