Among one's own? : The effect of the neighborhood context on mobility intentions an social capital in Amsterdam.

(1)

UNIVERSITY OF AMSTERDAM

Date: 17th August 2014

Author: Michael Lesch

Student Number: 10601449

Supervisor: dhr. dr. Aslan Zorlu

Second Reader: dhr. Christian Lennartz MA

Program: Human Geography Msc.

Thesis Project: Applying a Quantitative Approach 2013-2014

A

MONG

O

NE

’

S

O

WN

?

T

HE

E

FFECT OF THE

N

EIGHBORHOOD

C

ONTEXT ON

M

OBILITY

I

NTENTIONS AND

(2)

List of Tables and Graphs

Table 1: Mean values of variables used by ethnic group ... 18

Table 2: Multinominal logit estimates by ethnic group, relative risk ratios (base = Stay) ... 25

Graph 1a: Home feelings by neighb. Ethnic composition (APR) ... 34

Graph 1b: Home feelings by neighb. Ethnic composition (MER) ... 35

Graph 2a: Home feelings by change of neighb. Composition (APR) ... 37

Graph 2b: Home feelings by change of neighb. Composition (MER) ... 38

Graph 3a: Home feelings by age (APR) ... 39

Graph 3b: Home feelings by age (MER) ... 40

Graph 4a: Home feelings by occupancy (APR) ... 42

Graph 4b: Home feelings by occupancy (MER) ... 43

Graph 5a: Safety feelings women by neighb. composition (APR) ... 44

Graph 5a: Safety feelings women by neighb. composition (MER) ... 45

Graph 6a: Safety feelings parents by neighb. composition (APR) ... 46

Graph 6a: Safety feelings parents by neighb. composition (MER) ... 47

Graph 7a: Perceived commitment by neighb. composition (APR) ... 49

Graph 7b: Perceived commitment by neighb. composition (MER)... 50

Graph 8a: Neighb. satisfaction by perceived commitment (APR) ... 50

Graph 8b: Neighb. satisfaction by perceived commitment (MER)... 51

Graph 9a: Expected neighb. development by neighb. composition (APR) ... 53

Graph 9b: Expected neighb. development by neighb. composition (MER) ... 54

Graph 10a: Expected neighb. development by change of neighb. composition (APR) ... 55

Graph 10b: Expected neighb. development by change of neighb. composition (MER) ... 56

Graph 11a: Expected neighb. development by population turnover (APR) ... 57

(3)

1. Introduction

This should be the promise of urban life: the city’s diversity of urban life becoming a source of mutual strength rather than a source of mutual estrangement and civic bitterness.

Richard Sennett (2005: 1)

To meet Amsterdam’s increasing demand for affordable housing the city started to develop Ijburg in the year 2000 as a mixed housing area aspiring to build 22,000 units. Following the city’s goal of promoting social inclusion, aside from rental units and homes for sale, 40% of the housing stock was devoted to social housing for low-income groups. After the homeowners moved in, the city began to occupy the housing stock with non-Western minorities, lower income families and disabled people. However, the social mix in the neighborhood led to considerably growing tensions between its heterogeneous residents. Many home buyers claimed that they would not have bought a home in the neighborhood, if they would have known beforehand, that they will share their living spaces with such a diverse, and apparently to some extent undesired, range of people (Loux 2011; see also Essbai 2013).

This brief example stays in stark contrast to how urban life is conceived in science and society: Cities are said to be the places where it all comes together, where people of different ethnicities and classes converge, individuals pursue their life chances and together create something new and exciting. This notion of urbanity promotes those ideas embodying all the virtues associated with good life in the city. Late urban sociologist Hartmut Häußermann even stressed the image of “the city as an

integration machine”, a place naturally integrating a diverse range of people into society

(Häußermann 1995). In modern societies urbanity evokes ideas of openness, equality and diversity and thereby has an utterly positive connotation in public discourse. As urban planner Franz Pesch exemplarily puts it: “Urbanity manifests in multiple dimensions: politically in self-government and

participation, socially in civic identity and public welfare and culturally in exchange and communication” (Pesch 2010: 1).

Nowadays, the latter notion of diversity has become the normative core and guiding principle of planning practices or as Susan Fainstein argues “the new orthodoxy of city planners” (Fainstein 2005: 3). In regard to the Netherlands, the Big City Policies were devised under the assumptions that

(5)

the best opportunities for individual economic progress, while the concentration of social groups was conceived as a problem (Musterd et al. 2011: 86). However, more and more scholars seem to take the functional rhetoric of mixed neighborhoods into question (e.g. Butler et al. 2001; Musterd et al. 2003a; Putnam 2007; Musterd et al. 2009). By looking closer at group dynamics within cities, they appear to draw a more ambiguous picture of urban life.

To get back to the initial example, an entirely new development such as Ijburg is rather the exception than the rule in a city like Amsterdam, but it exemplarily indicates, that certain groups might not like to be exposed to difference in their immediate living environment and might even base their residential choices on this very notion. Generally, the ethnic and social composition of already existing neighborhoods changes more gradually, but still intensely: In the larger cities of the Netherlands one in six people move every year while large-scale urban renewal programs can cause population turnovers which far exceed 30 per cent of a neighborhood’s population (de Groot et al. 2011).

In this context, a deeper understanding of the factors influencing both social capital in the neighborhood as well as residential mobility becomes crucial for adapting urban regeneration policies sensitive to potential conflict. Evidence from the Netherlands already suggests that policies promoting ethnic diversity in the neighborhood to facilitate the integration of ethnic groups into society might have the opposite effect eroding individual social trust in those residential areas (see Lancee & Dronkers 2010, Musterd et al. 2003).

This thesis aims to answer the research question how the neighborhood context affects their mobility

intention in Amsterdam [Q1], and accordingly residents’ social capital in the neighborhood [Q2].

These questions will be investigated by studying the migrants’ and natives’ perceptions of neighborhood social capital and their intentions to move. It is assumed that neighborhood context influences these groups differently, which will be tested along the lines of ethnicity and destination preferences. The Wonen in Amsterdam 2011 survey data provided by Bureau Onderzoek & Statistiek Amsterdam allows to examine these questions on the scale of Amsterdam’s 84 neighborhoods. Therefore, a multinominal logistic regression model as well as a series of binary logistic regression models will be employed respectively controlling for a) individual, household and dwelling characteristics, b) subjective (indicators of) neighborhood social capital, and c) objective indicators of neighborhood change.

(6)

2. Theoretical and Empirical Framework

Figure 1: Conceptual Scheme

This thesis puts particular emphasis on the ethnic component of the neighborhood context as it is at the core of the potential conflict between the urban ideal and potentially disapproving attitudes of some resident groups. If there is a tendency that residents in Amsterdam want to live among their own kind will be investigated in a twofold manner. The first research question is concerned with the relation between the neighborhood context and respondents’ mobility intentions and is posed as follows:

To answer this question a multinominal logistic regression model is employed using the destination preferences stated by the respondents of the Wonen in Amsterdam 2011 survey as the dependent variable. The model controls for the commonly used predictors of a) individual, household and dwelling characteristics, and adds to them b) subjective (indicators of) neighborhood social capital, as well as a set c) objective indicators of neighborhood change. As it is expected that the neighborhood context affects the different ethnic groups differently, the estimates are calculated for each of the groups individually.

Q1: How does the neighborhood context affect the mobility intentions of natives and migrants in Amsterdam?

(7)

There is international evidence from the US supporting the White Flight Hypothesis which suggests that white Americans are significantly more likely to leave neighborhoods with minority populations, even when controlling for socio-economic factors. However, research in the city of Amsterdam has not confirmed ethnically motivated spatial mobility of its residents, when controlled for the economic conditions of the residential areas (Zorlu 2008: 12). Spatial mobility in Amsterdam also faces certain restraints by a dominant social housing sector and a tense private market segment restricting residential choices. A turning-away from ethnic diversity with a slowly looming wish for a more homogenous setting can therefore be expected to become apparent at the earliest stage of the mobility process, but might not necessarily translate into actual mobility behavior. The distinction between the subjective neighborhood context and the objective neighborhood context is crucial in that regard. While the former has proven to have substantial explanatory value in modeling mobility behavior, the latter’s influence on residents’ mobility is regarded to be “virtually non-existent” (Lu 1998: 1477, see also Lee et al. 1994).

Accordingly, the second research question is not concerned with mobility intentions, but rather focusing on the relation between the (objective) neighborhood context and respondents social capital in the neighborhood and is posed as follows:

The analysis will focus on the differences between natives and migrants by employing binary logit regression models with indicators of neighborhood social capital serving as the dependent variables controlling for the same sets of covariates as in Q1. This proceeding thereby links the objective neighborhood context with the subjective neighborhood context in regard to the ethnic groups. To intuitively present the results adjusted predictions and (average) marginal effects are estimated at representative values (objective neighborhood context: share of non-Westerners; population turnover etc.) are visualized by marginsplots.

In the following, it will be elaborated on the theories and theoretical concepts used in regard to the research questions of how the neighborhood affects context residential mobility (see chapter 2.1) and social capital (see chapter 2.2). Therefore, the corresponding literature will be reviewed and it will be demonstrated how the concepts were operationalized in the context of this research. On the basis of empirical evidence, hypotheses will be posed which will be tested in the analysis.

Q2: How does the neighborhood context affect the social capital of natives and migrants in Amsterdam neighborhoods?

(8)

2.1 Neighborhood Context and Residential Mobility

2.1.1 Residential Stress

Literature overwhelmingly conceives residential mobility as a response to residential stress provoked when a household’s residential needs and preferences are not met by their current living situation. These needs can become apparent in various spheres and range, for instance, from dwelling characteristics (e.g. a societal climbers’ demand for better dwelling quality) to neighborhood characteristics (e.g. elderlies’ need for accessibility of local supply) and were predominantly discussed from the perspective of changes within households in terms of social aspirations or life-cycle stages (Lu 1998: 1474). While the landlord could be asked to improve dwelling by renovation or elderly could make use of ambulant and delivery services, residential stress induced by external factors as changing neighborhoods themselves often goes beyond the residents’ sphere of influence:

“[T]he only way to improve one`s neighborhood is to move to a better one” (Feijten & van Ham 2009:

2103). Mobility can therefore be expected to dissolve residential stress by taking households into a new residential setting that better suits their needs.

Households judge their current residential situation on the basis of their own characteristics as a household, the characteristics of their dwelling and their neighborhood. While most studies on residential mobility are concerned with actual moving behavior, Feijten and van Ham (2009) focus on people’s wish to leave their neighborhood. In fact, de Groot (2011: 62) shows that 68 per cent of people who had an intention to move in the Netherlands did not leave their dwelling within two years after the interviews he has conducted.

The emphasis on moving wishes rather than actual moving behavior is also significant for this thesis as the Amsterdam housing market still has a comparatively huge social housing sector and remains highly regulated. People state moving wishes without taking individual opportunities and external restrictions into account which makes them the ideal indicator for residential stress and thereby an adequate dependent variable for analysis.

2.1.2 Destination Preferences

Forrest and Kearns argue that the neighborhood has become an extension of the home for social purposes and is crucial part for residents’ identity: “´location matters´ and the neighborhood

becomes a part of our statement about who we are” (Forrest & Kearns 2001: 2130). Karsten’s

qualitative study “Housing as a way of life” on middle-class families who deliberately decide to live in the city also emphasizes this notion. While the suburbs are still regarded as the main habitat of middle class families, those who prefer the city particularly stress the identity aspect of it. They value

(9)

diversity highly and refer to themselves as “true urbanites”, “sturdy families” and “real city people” (Kartsen 2007).

Empirical evidence indicates, that different motivations are related to the destinations of residential mobility. Long-distance mobility is commonly more associated with job seeking or job promotion, than with dwelling or neighborhood induced considerations. Inner-city mobility is generally considered as the adjustment of a households’ spatial needs to an appropriate dwelling, while a move to the suburbs is rather being interpreted as mainly triggered by dissatisfaction with the current neighborhood (Clark & Ledwith 2006 cited Zorlu 2008). As the proportion of migrants is particularly high in Amsterdam and disproportionately low in its surrounding suburbs, the choice of moving within the city or to the suburbs has a high potential to further promote a process of ethnic sorting (Zorlu 2008: 6). Preferring the suburbs over the city as a place to reside can certainly be considered a hint towards a lifestyle choice in the sense that the residents are escaping the high diversity and density of the city. However, it also has to be noted that there are also quite homogenous neighborhoods in Amsterdam which have a less urban character.

2.1.3 Influence of Personal, Household and Dwelling Characteristics

There are several known factors related to residents’ personal, household and dwelling characteristics that influence their moving wishes, which need to be controlled in the analysis. A first set of variables will therefore include personal and household characteristics, notably age, gender, ethnicity, income, education, and the household composition (see Feijten & van Ham 2009; van Ham and Feijten 2008). Residential stress induced by those factors often does arise from changes internal to the household such as changes in the life-cycle stages or social aspirations (Lu 1998). People in different life phases show different patterns of residential mobility and also have different needs and expectations towards their living environment as it is conceived in the sociological life-cycle approach (Li & Tu 2011: 4). The corresponding life-course events are often demanding spatial adjustments and are regarded as a decisive factor for residential mobility (Rabe & Taylor 2011). For instance, young families need more space when a baby is born, while elderly people might prefer a quieter environment than the lively city center in the later stages of their life.

The second set of control variables is concerned with characteristics of the dwelling and includes the dwelling type, the satisfaction with the dwelling, and the question whether it is owner-occupied, privately rented or part of the city’s social housing stock. The dwelling size is a crucial predictor for residential stress and the resulting residential mobility, which has to be controlled for (Clark et al. 2006). Residents who show high levels of satisfaction with their dwelling can certainly be expected to be less likely to wish to leave their neighborhood. Homeowners can also be expected to show a

(10)

tendency to stay in the neighborhood they selected as they did so deliberately and chose a place which is likely to suit them according to their needs and preferences.

2.1.4 Influence of Neighborhood Context

Residential stress can also be caused by changes external to the household such as changes in the neighborhood. Empirical research has been predominantly focused on residential stress caused by changing household characteristics, while the effects of the neighborhood context have rarely been taken into account (Feijten & van Ham 2009: 2103-04). As alluded before, for the purpose of this research the neighborhood context is conceived in a twofold manner as subjective and a objective context. There is still an ongoing debate among scholars if the objective neighborhood context in fact significantly influences spatial mobility.

Lee et al. (1994) contended that mobility decisions are significantly influenced by the way residents experience and perceive change in their neighborhood. Moreover, they arrived at the conclusion that there is just limited influence of the objective aspects neighborhood context on residential mobility (see also Lu 1998). In contrast, Feijten and van Ham (2009) were able to show that neighborhood change measured by objective indicators has a significant impact on residents’ mobility intentions. Whereas Lee and colleagues used a small sample of 484 respondents from Nashville, Tennessee, Feijten and van Ham were using data for the Netherlands (N=52403) from the Dutch Housing Demand Survey WoON 2002. However, the lowest spatial entity they were able to analyze on this way were four-digit postcode areas, which they explicitly addressed as problematic as these areas might be too large to be perceived as neighborhoods by the residents. Another important difference is that the former were investigating actual residential mobility, while the latter were inquiring moving wishes and solely testing for one subjective variable (Opinion about neighborhood

development in the last year). The thesis aims to add to that debate by testing if the effects of the

objective indicators on residential mobility can also be confirmed a.) with a sample using smaller entities as neighborhoods and b) in the highly urbanized setting of the Dutch capital. The thereby developed hypotheses are in line with those of Feijten and van Ham (although they did not calculate individual estimates for ethnic groups) and offer the possibility to compare if their findings in the Netherlands can be maintained under the premises of this inquiry.

Subjective neighborhood context

The subjective context is measured by a set of (indicators of) social capital in the residential area including residents’ satisfaction with the neighborhood, the perceived neighborhood commitment, their home feelings their perceptions of safety at night, and eventually their expectation about the

(11)

future development of the neighborhood.1 Residential stress as a perceived lack of social capital in their residential area is expected to have a significant impact on their moving intentions as it has been demonstrated for example by Lee et al. (1994) and Kan (2007), who showed that a household’s possession of local social capital has a negative effect on its residential mobility.

Objective Neighborhood Context

Literature indicates that three aspects of neighborhood change are particularly important for residents’ mobility intentions, notably changes in the socio-economic status of the neighborhood population, changes in the ethnical composition of the neighborhood population, and a high population turnover within the neighborhood

Harris (1999 cited Clark & van Ham 2009: 1444) concludes in his literature review that people show the clear tendency to move out of neighborhoods with low socio-economic status, if they are given the opportunity. Low levels of education, low income and unemployment are generally considered as indicators that the neighborhood deviates from mainstream norms and values by residents (especially those with children) and trigger their desire to leave. Accordingly, neighborhoods with low socio-economic status can be expected to also show high levels of residential mobility.

Social status awareness also plays a significant role in household’s neighborhood choices as Michelson has shown (1977 cited Feijten & van Ham 2009: 2106). Asking people who were moving to a different neighborhood to compare themselves socio-economically with their old and their new neighbors, he found that they thought of themselves as belonging to a higher socioeconomic class than their old neighborhood represented and moved accordingly. This status aware moving behavior can trigger a spiral of selective downward mobility. With the socioeconomic status and the desirability of the neighborhood dropping further and further even more families might consider to leave eventually. Therefore, it is hypothesized:

1

For a more detailed discussion of the employed predictors and their relation to social capital please see chapter 2.2.3

H1: Residents are more likely to wish to seek for a new place of residence when the socio-economic status of their neighborhood declines.

(12)

Semyonov et al. (2007: 438) identify three general lines of argumentation concerning the debate on the socio-psychological underpinnings of ethnic residential preferences. The first line suggests that individuals have a desire to maintain cultural homogeneity and therefore prefer live among their own kind in homogenous communities as they share their values and cultural background (Clark 1986 cited Semyonov 2007). The White Flight Hypothesis claims that a (white) majority group’s preference for homogenous neighborhoods is an expression of racial prejudice towards ethnic minority groups (Bobo & Zubrinsky 1996 cited Semyonov 2007). Thirdly, the Racial Proxy Hypothesis argues that the preference of residential homogeneity is a response of individuals associating a whole range of social problems with ethnic minorities. On the one hand, ethnic minority groups are more likely to be unemployed and poor, and on the other the limited choices these groups have on the housing market make them more likely to be forced to move into a deprived neighborhood (Harris 1999; 2001). Following the Racial Proxy Hypothesis, residents do not want to leave these neighborhoods because of prejudice against ethnic minority groups, but because these neighborhoods are troubled with social problems.

Accordingly, Harris argues that models need to properly control for ethnic and socio-economical composition of neighborhoods as he expects that potential effects of ethnic composition will disappear when a change of the socio-economic status of the neighborhood is controlled for, which will be done in the analysis (Harris 1999: 465). To test if this assumption is valid the second hypothesis is posed as follows:

Van Ham and Clark (2009: 1442) further suggest that intense population turnover (understood as the percentage of people moving within or out) in a neighborhood is often associated with negative implications for its residents: Residential instability is said to be related to residents’ involvement in crime and violence, eroding social capital and weakening identification with the neighborhood. As mentioned before, in the larger cities of the Netherlands one in six people move every year. In neighborhoods which are undergoing large-scale urban regeneration the numbers are significantly higher, which can be expected to trigger the residents’ wish to leave the neighborhood.

Lee et al. (1994) included a measurement of population turnover in his study and arrived at counter-intuitive results indicating that people are actually less likely to have the intention to move in

H2: Residents are more likely to seek for a new place of residence when there is an increase in non-Western ethnic minorities in their neighborhood.

(13)

residential areas with a high population turnover. On the contrary, Feijten and van Ham (2009) findings demonstrate that residents are more likely to move, if they perceive a high population turnover in their neighborhoods. Hypothesis 3 will follow the suggestion of the latter as it seems to be more in line with intuitive expectations and is the more covered relation in international literature:

H3: Residents are more likely to seek a new place of residence, if they live in a neighborhood with a high population turnover.

(14)

2.2 Neighborhood Context and Social Capital

2.2.1 Ethnic Concentration in Amsterdam

Immigration has always been pivotal for the development of Amsterdam’s urban culture. People born outside the Netherlands accounted constantly for around one-third of the city’s population in the 17th and 18th Century (Lucassen & Penninx 1994: 29). Decreasing to its all-time low at the beginning of the 20th Century, immigration gained momentum again in the 1960s by a substantial influx of guest workers. Turkish, Moroccan and Mediterranean people moved into cheap housing in and around the old city center when their settlement became more permanent. Surinam’s independence in 1975 induced considerable migration to Amsterdam. As the Surinamese intended to stay permanently from the beginning on, their settlement was mainly focusing on the pricier, modernist high-rises in the Bijlmermeer area. The inflow of non-Western migrants into the city’s old areas provoked a considerable portion of the native Dutch population to leave their long-time residential areas. If the indigenous intended to improve their dwelling situation before 1985, they solely had option to move into the suburban towns around Amsterdam. From there on, the migration to Amsterdam can be characterized by three patterns: Firstly, family reunification of the Turkish and Moroccan, secondly, further immigration for educational purposes and family formation by Antilleans and Surinamese and, thirdly, an increasing influx of economic refugees form less developed, non-Western countries (van Heelsum 2007: 11).

The trend towards ethnic contraction continued from the second half of the last century to the current day as recent research indicates. Zorlu and van Latten show that there are indeed segregatory tendencies in the mobility behavior among natives as well as non-Western immigrants in the Netherlands: The native Dutch tend to move to neighborhoods with a high share of natives, while immigrants from non-Western countries are less inclined to move to native neighborhoods. There is no evidence indicating spatial assimilation of non-Western immigrants, whereas second generation Western immigrants show similar moving behavior as natives (Zorlu & van Latten 2009: 1918).

2.2.2 Ethnic Concentration

The unequal distribution of housing of social groups within a city is commonly labeled residential segregation or residential concentration. These terms are in principle non-judgmental and are mostly investigated along the lines of economic, demographic or cultural characteristics. Current sociological and geographical research is overwhelmingly focusing on the nexuses of ethnicity and segregation as well as poverty and segregation, which are considered problematic in their long-term effects (Löw 2008: 43).

(15)

The assessment of this phenomenon depends largely on the society’s self-conception. In the pre-industrial European City the socio-spatial structure unfolded on the basis of the concurrence of corporate principles (ancestry and respectability), the functional arrangement of professional guilds (craftsman and merchants) and religion (Christian and Jewish people): “In hierarchically structured and cooperatively fragmented societies segregation was and is institutionalized as a matter of course” (Häußermann & Siebel 2004: 153). Not until the emergence of open societies, which are guided by the ambition of providing equal rights and life chances to all people, segregation became a problem as it is seemingly opposed to the ideals and openness and equality.

The questions if ethnic concentration is harmful or functional and stays contested in research and policy to the current day. In Dutch political discourse, like in the most countries across Western Europe, the phenomenon has generally been conceived as problematic. Correspondingly, social and “ethnic” mixing policies have been regarded as the suitable political response. Mixed neighborhoods are expected to provide the residents of deprived neighborhoods, where immigrants are often overrepresented, with better opportunities for social mobility and facilitate their integration into society (Musterd 2005). About two decades ago, the first round of the Dutch Big City Policies declared the establishment of mixed neighborhoods as their main objective. Targeting “social

ghettos”, a diversification of the housing stock by demolishing low-cost accommodation and

replacing it with higher quality dwellings was aimed to attract more affluent households to these residential areas. These restructuring policies are continued to the current day with the Big City

Policy III+ particularly focusing on ethnic and social integration (Musterd & Ostendorf 2008).

Musterd and Ostendorf conclude that segregation levels in the Netherlands are not increasing, but rather remain at comparatively low levels. In fact, higher income households tend to live more segregated than lower income households (Pinkster 2006 cited Musterd & Ostendorf 2008), while conclusive scientific evidence for the negative effects of ethnic concentration on life chances is still lacking: “[S]ocial life and social interactions are no longer confined to neighbourhoods, while social

opportunities may not be neighbourhood-related. The ‘community’ may have partly lost its territorial neighbourhood link.”

(16)

2.2.3 Neighborhood Social Capital and Ethnic Diversity

As alluded before, in academic and political debates about social cohesion, for more than two decades attention is increasingly drawn to the neighborhood as a place where the processes which presumably shape life chances and social identity convene. Against this backdrop, a renewed interest in local social relations and notions of community is taken note of (Forrest & Kearns 2001: 2125). Residents’ sense of community is reflected in their stock of social capital. Robert Putnam defines social capital as “features of social organization such as networks, norms, and social trust that

facilitate coordination and cooperation for mutual benefit“ (Putnam 1995: 67). In his controversially

discussed study “E Pluribus Unum: Diversity and Community in the Twenty-first Century” he argues, not without a sense of genuine worriedness, the social capital is challenged by diversity. On the basis of empirical studies using survey data (N= 29739) for neighborhoods all over in the United States of America he claims that more ethnic diversity, in fact, leads to a decrease of social capital and a decline of mutual solidarity in the short to medium run, which are both inclined to increase residential stress (Putnam 2007: 138). In ethnically diverse neighborhoods, residents of all ethnicities show a tendency of “hunkering down”, trust towards other ethnic groups as well as to one’s own group is lower, the social networks are less dense and altruism becomes rarer. Another point he is making tends to get often overlooked in reception: Putnam also expects successful immigrant societies to create new forms of social solidarity mitigating the negative short-term effects of diversity.

His findings were met by ample attention and induced a considerable echo within the academic community, which led many scholars to follow Putnam’s request to investigate if his conclusions have to be confirmed in communities outside of North America. A common notion among researchers at this time has been that in a more polarized US society this relation might be stronger than in Western European neighborhoods which are hallmarked by lower levels of social and spatial inequity (Winship: 2008).

Bram Lancee and Jaap Dronkers replicate his efforts using data from the Netherlands (Lancee & Dronkers 2010). They arrive at the conclusion that Putnam’s findings can also be confirmed in the context of European welfare states. At least in the short run, there is also an existing negative relation between ethnic diversity in the neighborhood and the quality of inter-ethnic contacts for natives and immigrants alike in the Netherlands. Moreover, they particularly address that municipal ethnic mixing policies have an opposed effect as they decrease individual social trust in the targeted residential areas. However, they also find that trust in other ethnic groups than one’s own was not negatively affected by ethnic diversity in the Dutch neighborhoods (Lancee & Dronkers 2010: 97).

(17)

These results have also been found valid in subsequent studies in the Netherlands (Tolsma et al. 2009; Lancee & Dronkers 2011).

Drawing on Putnam’s definition of social capital as “social networks and the associated norms of

reciprocity and trustworthiness” Harell and Stolle (2010 cited Lancee & Dronkers 2011) suggest to

distinguish between structural and cognitive social capital: The structural component is concerned with the “wires” in the network as the frequency and intensity of links between actors. It necessarily involves a behavioral element, unlike the cognitive component. The latter is concerned with the “nodes” in a network as the attitudes and values (i.e. trust, support and reciprocity) facilitating the exchange of resources.

The Wonen in Amsterdam 2011 survey provides an adequate approximation to cognitive social capital by including items regarding the respondents’ perception of their neighborhood. Therefore, a set of variables will be added to the regression models regarding (indicators of) social capital in the respective residential areas. These are variables concerned with the residents’ satisfaction with their neighborhood, the perceived neighborhood commitment, their home feelings, their perceptions of safety at night, and their expectation about the future development of the neighborhood. While home feelings and safety feelings can be considered classical measures of cognitive social capital, the remaining variables are approximations and are therefore rather indicators of social capital in the neighborhood. In the second part of the analysis the effects of objective neighborhood context on social capital for the three ethnic groups will be tested along the lines of the hypotheses as developed in the following. Therefore, binary logistic regression models will be employed using subjective neighborhood context variables as dependent variables.

Home Feelings and Ethnic Diversity

For many the choice of home as “a place to live amongst your own kind” has become a lifestyle decision (Savage et al. 2005: 53 cited Duyvendak 2011: 11). In an increasingly mobile and globalized world the meaning of “home” changed fundamentally over the past decades and demands new strategies of how to make sense of their immediate living environment. Even for people who have spent their entire lives in the same neighborhood all along, the chance of getting exposed to different ethnicities or lifestyles rose substantially, for instance due to the influx of guest workers from non-Western countries after the Second World War. Duyvendak (2011: 14) identified “defensive

localism” as a strategy where people react to the mobility paradigm of our times by withdrawing

themselves into a safe “haven” composed of particular places, goods and like-minded people. The effect of ethnic diversity on resident’s home feelings is first tested in respect to the static

(18)

neighborhood composition measured by the share of non-Western migrants in the neighborhood. It is expected that:

To test if there are also short term effects of ethnic diversity on home feelings, they are secondly estimated with regard to the change in the share of non-Western migrants over the course of four years. It is hypothesized that:

Thirdly, the effects of age and occupancy on residents’ home feelings are estimated. Duyvendak (2011: 38) considers familiarity (“knowing the place”) as the basic precondition on which the two elements (“home as haven” security and comfort; “home as heaven” self-expression and public identity) which constitute people’s home feelings can unfold. This notion implies that time and a basic familiarity with the culture one is surrounded by should not be overlooked in this regard. As mentioned before, literature indicates that the satisfaction with dwelling and neighborhood are growing substantially over the course of a lifetime. Natives can be expected to show a less pronounced effect as they are shaped by the very culture from the cradle onwards, while it might take migrants some time to settle in. Thus, it is hypothesized:

Safety Feelings and Ethnic Diversity

International evidence shows that there is no significant relation between reported crimes and the fear of crime, but in fact a significant and strong relation between actual and perceived levels of ethnic diversity and fear of crime (de Vroome & Hooghe 2014: 1). It can be expected that residents feel less safe during night time as there are fewer people around to help in case of a crime, violence or harassment, and that particular groups are more sensitive towards these concerns than others. Women are expected to feel generally more vulnerable than men in such situations, and people who have children can be expected to take particular care about safety in their immediate living

H4: Natives feel more at home in neighborhoods where only few non-Western migrants reside and less at home in residential areas where non-Western

migrants are concentrated (compared to non-Westerners themselves).

H5: An increase of non-Western migrants in the neighborhood impacts the home feelings of natives negatively, while non-Western migrants remain unfazed.

H6: Natives feel generally more at home in the neighborhood, but show the least pronounced effect over the course of time compared with migrants

(19)

of non-Western migrants on the potentially vulnerable groups’ safety feeling at night are tested. The static neighborhood composition was chosen over the change in non-Western migrants as it is assumed that the total share of migrants is more meaningful in this context. If certain residents associate the presence a minority group with a thread to their safety, the chance of encountering a potential aggressor are certainly higher in neighborhoods where they are concentrated. The hypotheses are posed as follows:

Commitment and Neighborhood Satisfaction

In the survey “Internationals on Amsterdam” migrants (more than two thirds of them from Western countries and 79 percent in the Netherlands less than five years) were asked how they experience life in the city (Bureau O+S 2012). They report that the social network of 59% of respondents is primarily comprised of other internationals. Moreover, internationals complain about a lack of social

contact with friends and others, more than 50 percent state that it is hard or very hard to integrate

into Dutch society, and 60 percent of them indicate that aren’t really or aren’t at all part of the local community. Non-Western migrants, on the other hand are known to values their family and kinship highly (Zorlu 2008). Firstly, the effect the share of non-Western migrants in the neighborhood on how residents perceive the commitment in their residential area is estimated. Secondly, residents’ neighborhood satisfaction in regard to how they rate the commitment in the neighborhood is predicted. The corresponding hypothesis is posed as follows:

H7: Women and parents of non-Western descent show the least pronounced effect of ethnic diversity on their safety feelings at night.

H8: Women and parents of non-Western descent feel considerably safer than natives in neighborhoods with high shares of their ethnic group.

H9: Western migrants value neighborhood commitment lower than natives and are therefore more likely to be satisfied when reporting low levels of commitment.

(20)

Expected Neighborhood Development and Ethnic Diversity

If residents think, their neighborhood either deteriorates or improves in the future, it can be considered an indicator of cognitive social capital in the sense that it is decisive to identify with the neighborhood as a common good, where everyone has his stake in. A neighborhood that is perceived by the residents to constantly worsen, does offer them little space for identification, but might rather facilitate retreat from public life. On the contrary, a residential area which is expected by their residents to improve allows them to identify with the place and might tend to evoke collective commitment towards something common which everyone would like to be a part of.

How residents expect their neighborhood to develop is predicted in regard to firstly the ethnical composition of the neighborhood, secondly the change of the ethnical composition and, thirdly the population turnover. Residents associating the presence or the decrease of non-migrants in their residential area with a negative development can be considered problematic and point to negative stereotypes. Moreover, it can be expected that not all ethnic groups assess low or high levels population turnover alike: Some might prefer to keep their peers within the neighborhood and thus prefer the status quo, while others with weaker social networks rather remain unfazed. Accordingly, following hypotheses have been posed:

H10: Non-Western migrants are more likely to expect the neighborhood to improve than natives in neighborhoods with high shares of their ethnic group and

respectively in neighborhoods with high growth rates of their ethnic group.

H11: As they have weaker ties in the neighborhood, Western migrants are more likely to associate high rates population turnover with an improvement of the

(21)

3. Data and Descriptive Statistics

The hypotheses will be tested using data on individual as well as neighborhood level. The individual level data (17900 respondents) was assembled from the Wonen in Amsterdam 2011 survey, which is undertaken every two years since 1995 by the Dienst Wonen, Zorg en Samenleven and the Bureau Onderzoek & Statistiek of the municipality Amsterdam. The data is organized in 84 neighborhoods (“buurtcombinaties”) which makes an average of 213 respondents per neighborhood. The Bureau Onderzoek & Statistiek furthermore guarantees that the minimum amount of respondents from ethnic minority groups is met allowing to make substantiated statements (Bureau O+S 2012: 161). The neighborhood level data was partially obtained through the website of Bureau Onderzoek & Statistiek and is presented in detail in their yearly report “Amsterdam in cijfers: Jaarboek 2011” and more comprehensively in the study “Ontwikkeling in Amsterdamse buurten 2005-2011” (Bureau O+S 2013). The individual survey data and the neighborhood statics have been merged on the basis of the variable marking the neighborhoods, which was identifiable in both sets. For the analysis, the three ethnic groups of natives, Western migrants and non-Western migrants, which often emigrated from less developed countries, have been distinguished. The descendants of the migrant groups have been included in the migrant category, even if they gained Dutch citizenship as it can be expected that some cultural patterns still persist. For instance, the second and third generation of non-Western migrants from Turkish and Moroccan families still show lesser educational achievements and are in weaker socio-economical position than the other ethnic groups.

Table 1: Mean values of variables used

Variable Native Non-Western Western

Preferred moving destination (dependent variable)

Stay (base) 0.47 0.41 0.47

Amsterdam 0.29 0.42 0.28

Suburbs 0.06 0.04 0.05

Long-distance 0.06 0.03 0.07

Inclined to move, but no destination yet 0.12 0.10 0.13

Age 18-30 years (base) 0.14 0.13 0.12 31-60 years 0.61 0.74 0.65 61-84 years 0.25 0.14 0.22 Gender Male (base) 0.61 0.63 0.58 Female 0.39 0.37 0.42

(22)

Education

Low: max. VSO, VBO, VMBO, MBO-short (base) 0.14 0.42 0.12

Medium: HAVO, VWO, MBO-long 0.17 0.21 0.18

High: HBO, WO 0.66 0.30 0.67

Individual income (in €) 2.861.76 2.005.15 2.870.67

Household composition

Single (base) 0.46 0.30 0.47

Couple without kids 0.30 0.16 0.28

Single parent 0.06 0.17 0.08

Couple with kids 0.17 0.36 0.16

Other households 0.01 0.02 0.01

Housing type

Owner occupied (base) 0.51 0.23 0.47

Social housing 0.32 0.68 0.33 Private rent 0.17 0.09 0.20 Dwelling size <40m² (base) 0.09 0.09 0.12 40-60m² 0.40 0.44 0.39 60-80m² 0.26 0.31 0.25 80-100m² 0.12 0.10 0.12 >100m² 0.08 0.04 0.08

Dwelling satisfaction (base = satisfied)

unsatisfied 0.04 0.26 0.08

Neighborhood satisfaction (base = satisfied)

unsatisfied 0.07 0.18 0.09

Neighborhood commitment (base = strong)

weak 0.24 0.32 0.25

Feeling at home in neighborhood (base = strongly)

weakly 0.06 0.15 0.09

Safety feelings at night (base = safe)

unsafe 0.13 0.26 0.16

Expected neighborhood development (base = improve)

deteriorate 0.14 0.24 0.17

% income change 2007-2011 10.56 10.33 10.34

% non-Western migrants change 2007-2011 0.49 0.34 0.30

% population turnover 2011 10.33 10.45 10.57

(23)

Respondents, who stated that they want to move within the next two years, were further asked for their destination preference. The variable regarding people’s preferred moving destination (mdest5) is accordingly employed as the dependent variable in the model comprising of the categories “stay” (reference category), “Amsterdam”, “Suburbs”, “long-distance” and “no destination yet”. The category “Suburbs” includes the municipalities outside Amsterdam but still in 25 kilometers distance, while “no destination yet” consists of the people who are inclined to move but are undecided or could not make up their mind yet where to.

While natives and Western migrants show similar patterns in their destination preferences, non-Western migrants depart to some degree. They are more inclined to leave their dwelling than the other groups with a 13% percent higher share in comparison to natives prefer to move within Amsterdam. In contrast, higher shares of natives and Western migrants are inclined to move to the suburbs.

The first set of control variables comprises of respondents’ age, gender, education, individual income and their household composition. A coding into categorical variables is chosen over dummies to test if certain effects also chance gradually in regard to the low base category with the increasing factors of the independent variables (as for age, individual income, education and dwelling size). The continuous age variable is thus coded into three categories representing three distinct stages in people’s life cycle loosely following O’Rand and Krecker (1990: 243). The first stage (18-30 years, reference) is related to education, the second stage (31-60 years) represents people’s working life and potential family founding, while the last stage (61-84 years) is expected to be framed by retirement and pension. Within these three different categories it is expected that people do have different needs regarding their dwelling and their location. The sample contains a higher share of non-Western migrants in the middle age group, while a lower proportion of them than natives and Western migrants are represented in the oldest age category. This is not surprising as not all of the guest workers settled in Amsterdam permanently, but have also moved back to their home countries. Gender is coded into a dummy variable (0=male, 1=female) and shows an underrepresentation of woman among all groups of respondents in the sample compared to the actual distribution in the Amsterdam population. Gender could be expected to play role in regard to the feeling of safety in the neighborhood, as woman are known to be more sensitive towards these issues. The socio-economic status of responds is accounted for by their educational achievements and their individual income. Education is measured in terms of peoples’ highest diploma and coded in three categories: low as base category (max. VSO, VBO, VMBO, MBO-short), medium (HAVO, VWO, MBO-lang) and high (HBO, WO). The individual income is similarly coded in three representing tertiles from low (<1700) to high (>3000). As it can be expected, non-Western migrants are significantly overrepresented in the low education and income groups, while being underrepresented

(24)

in the high groups compared to natives and Western migrants. This adds up to more than triple the share of them being within the category for low education, and at the same time just half of them in the category of higher university education. Household composition comprises of “singles” (reference category), “couples without kids”, “single parents”, “couples with kids” and the remaining “other” types of households. “Single” is chosen as a reference category as this group can be expected to be the most mobile and less settled and is thereby likely to be hallmarked by characteristics which deviate the most from the other categories. Non-western migrants live proportionately less often in single households and in couple households without kids, but more often in households with kids, which illustrates the higher fertility rates among non-Western migrants and in particular those of people of Turkish and Moroccan origin (see Garssen & Nicolaas 2008).

The next set of variables is considered with people’s dwelling situation and comprises of their type of housing the dwelling size and their satisfaction with their dwelling. The housing type distinguishes between privately owned dwellings (reference category), people living in social housing, and privately rented dwellings. Owner occupied housing is chosen as reference category as those residents are expected to possibly show different attitudes than people who are renting and are thereby less tied to a neighborhood. The Amsterdam dwelling stock is composed 30,2% privately owned homes, 47,2% social housing and 22,6% tenements in 2011 (Bureau O+S 2012b: 6). The weaker socio-economic position of non-Western migrants does indeed translate to the housing market with an overrepresentation in lowest segment of the housing market (over 30% higher share than the two other groups) and an underrepresentation in the highest segment (close to 30% lower share than the two other groups). Considering the dwelling size the coding of the Bureau Onderzoek en Statistiek in five categories (1-40m²=reference category; 40-60m²; 60-80m²; 80-100m²; >100m²) is adopted. It shows a relatively even distribution of the ethnic groups across the segments with the exception of dwellings larger than 100m² where only 4% of the Non-Western migrants stay compared to respectively 8% of the native and Western migrants. The comparatively equal representation in the four lower segments does not contradict the assumption that non-Western migrants are in a weaker position on the housing market. Considering their household structure (see above: less singles, more families with more children), they share a comparable space with more people. As the first attitudinal variable the dwelling satisfaction is added to the model which is concerned with this particular relation. The Wonen in Amsterdam 2011 survey was asking the respondents to rate their satisfaction with their current dwelling on a scale from one (“very

unsatisfied”) to ten (“very satisfied”). As the analysis is concerned with residential stress, being

satisfied with the dwelling (ratings from 6 to 10 = 0) is employed as reference category of a dummy variable tested against being unsatisfied (ratings from 1 to 5 = 1). While only 4% of natives and 8% of Western migrants are unsatisfied with their dwelling the share of non-Westerners adds up to 26%.

(25)

The twofold share of unsatisfied Westerners compared to the Dutch could possibly be attributed to them being used to more spacious dwellings in their home countries. The tense housing market of the extremely densely populated capital of the Netherlands might not offer enough room for maneuver to respond to growing spatial needs as the individual spatial consumption in Western European cities increases steadily. The high share of unsatisfied non-Western migrants is most likely a result to their more traditional household structure and them overwhelmingly living in social housing of lesser quality.

The following set of attitudinal variables is concerned with subjective indicators of social capital in the residential area representing the subjective neighborhood context. They revolve around the respondents’ perceptions and feelings towards their neighborhood, their fellow residents and the abidance of social norms by measuring the neighborhood satisfaction, how they perceive the commitment of their neighbors, if they feel at home there, their safety feelings at night time and how they expect their neighborhood to develop. In the survey these points of concern have been rated on a scale from one ( “very unsatisfied”/”very weak”/”very unsafe”/”much worse”) to ten (“very satisfied”/”very strong”/”safe”/”very unsafe”/”much improvement”). As the lack of social capital in the neighborhood is expected to contribute to residential stress, they have further been coded into dummy variables. The category comprising of positive ratings (from 6 to 10 = 0) therefore serves as reference category in relation to the negative ratings (from 1 to 5 = 1). Higher levels of overall dissatisfaction can be concluded for non-Western migrants across the variables regarding the perception of social capital in the neighborhood. The shares of Western migrants in categories with negative ratings are just by fine margins higher than those of natives. Not surprisingly, the biggest difference can be found in their representation in the category of residents who do not feel at home in the neighborhood (6% natives; 9% of Western migrants). On the one hand, Western migrants might be partially in Amsterdam for professional reasons and might not wish to stay permanently, and on the other the first generation of Western immigrants was socialized in another country, which they might still have a strong attachment to or consider it their home in some respects.

The next set of variables is regarding the objective neighborhood context and is measured by the average percentage change in individual income, non-Western population between 2007 and 2011 as well as the population turnover in 2011 in the respective neighborhoods. The mean values might appear surprisingly even on first glance, but do make sense when the absolute numbers are taken into consideration. As non-Western migrants are overrepresented in the lower income categories, the results indicate that the neighborhoods they reside in were neither catching up nor got further left behind economically between those years. They rather benefited of the positive trend equally, but started out on a considerably lower level. The percentage of non-Western migrants is also shown to just increase slightly in all neighborhoods distinguished by ethnic groups. Again, you need smaller

(26)

numbers of non-Westerners to move into an overly native neighborhood to arrive at the same rates of change as for a neighborhood with an above average non-Western population. This indicates no signs of increasing ethnic spatial concentration between the years 2007 and 2011. The almost even levels of population turnover are surprising considering the overrepresentation of non-Western migrants in the social housing sector, which is still undergoing immense renewal efforts. In this regard, it could suggest that the turnover in non-Western migrants neighborhoods could rather be driven by restraint due to urban redevelopment and relocation, while the higher socio-economic status of natives and Western migrants gives them more options to deliberately choose where to reside.

(27)

4. Method and Analysis

According to the research questions Q1 and Q2, the analysis is divided into two sections. The first section is dedicated to the question, how the neighborhood context is influencing mobility intentions in Amsterdam and will test H1-H3. This question is investigated on the basis of a multinominal logistic regression using the destination preferences stated by the respondents as the dependent variable. The outcome concerned with people who answered that they want to move, but have not decided on a destination yet, has been estimated but will not be discussed in the analysis because the results were neither significant nor meaningful. The multinominal model controls for a) individual, household and dwelling characteristics, and add to them b) subjective (indicators of) neighborhood social capital, as well as a set of c) objective indicators of neighborhood change. As it is expected that the neighborhood context affects the different ethnic groups differently, the estimates are calculated for each of the groups individually. Relative risk ratios will be reported and were chosen over odds ratios as they facilitate a more natural interpretation.

The second section of the analysis is concerned with research question Q2 regarding the relation between the (objective) neighborhood context and respondents social capital in the neighborhood. The analysis will focus on the differences between natives and migrants by employing binary logit regression models with indicators of neighborhood social capital serving as the dependent variables controlling for the same sets of covariates as in Q1. This proceeding thereby links the objective neighborhood context with the subjective neighborhood context in regard to the ethnic groups. To intuitively present the results adjusted predictions and (average) marginal effects are estimated at representative values (objective neighborhood context: share of non-Westerners; population turnover etc.) and accordingly visualized by marginsplots.

4.1 Neighborhood Context and Mobility Intentions

Table 2: Multinominal logit estimates by ethnic group, relative risk ratios (base = Stay) [on the

(28)

Amsterdam: Suburbs: long-distance:

Native NW Western Native NW Western Native NW Western

Age b = 1-30 31-60 0.39*** 0.51*** 0.46*** 0.54*** 0.71 0.45* 0.45*** 0.82 0.49** 61-100 0.22*** 0.40*** 0.22*** 0.23*** 0.25** 0.21*** 0.22*** 0.42 0.20*** Gender b= ma l e fema l e 1.01 1.29* 1.18 1.22* 1.29 0.97 0.80* 1.21 1.26 Education b = l ow medi um 1.27** 1.05 1.38 1.02 2.01** 1.87 1.87** 1.57 1.90 hi gh 1.67*** 1.52*** 1.70** 1.05 2.73*** 1.25 2.84*** 2.93*** 2.79** Individual Income b= <1700 1700-2999 1.21** 1.18 1.29 1.34* 1.39 1.74 1.26 0.81 0.83 >=3000 1.36*** 1.28 1.55* 1.63** 2.28* 1.73 1.34 0.88 1.31 HH-composition b=s i ngl e coupl e wi thout ki ds 0.97 1.16 1.40* 1.60*** 1.36 1.86* 1.07 1.26 1.69* s i ngl e pa rent 1.16 1.36* 1.36 0.98 1.60 2.32* 0.80 0.53 0.45 coupl e wi th ki ds 1.09 1.35* 1.75** 1.78*** 1.32 2.62** 1.15 0.61 1.22 Housing type b=owner

s oci a l hous i ng 1.43*** 1.64*** 1.63** 0.82 0.85 0.98 0.81 1.38 0.89 pri va te rent 1.89*** 2.47*** 2.55*** 1.20 1.59 1.36 1.34* 2.27* 2.13*** Dwelling size b = <40 40-60 0.59*** 0.85 0.75 0.82 1.09 1.38 0.76 0.60 1.13 60-80 0.46*** 0.56*** 0.46*** 0.69* 0.74 1.27 0.59** 0.65 0.94 80-100 0.37*** 0.30*** 0.40*** 0.49*** 0.56 0.83 0.42*** 0.68 1.21 >100 0.29*** 0.35*** 0.21*** 0.34*** 0.83 0.21* 0.29*** 0.33 0.49 Dwelling satisfaction b=s a ti s fi ed uns a ti s fi ed 3.40*** 4.00*** 3.38*** 2.58*** 2.60*** 1.53 2.47*** 1.68 2.91** Neighborhood satisfaction b=s a ti s fi ed uns a ti s fi ed 1.85*** 1.75*** 1.24 2.42*** 2.46** 3.47*** 2.25*** 0.82 1.25 Neighborhood commitment b=s trong

wea k 1.44*** 1.12 1.38* 1.51*** 1.26 0.84 1.45*** 1.79* 1.97** Feeling at home in neighb. b=s trong

wea k 2.43*** 2.47*** 3.64*** 3.89*** 4.07*** 5.85*** 2.73*** 4.34*** 5.19*** Safety (at night) b=s a fe

uns a fe 1.27** 0.93 0.85 1.45** 1.46 1.13 1.21 0.90 0.91 Expected neighb. developement b=i mprove

detori a te 1.31*** 1.08 1.52* 1.50** 1.20 2.14** 1.88*** 0.90 1.85* % income change 2007-2011 1.00 1.00 1.00 1.00 1.00 1.01 1.00 1.00 1.00 % non-western migrants change 2007-2011 0.96*** 0.98 0.98 1.01 0.97 0.96 0.97 0.96 0.97 % population turnover 2011 1.03*** 1.00 1.01 1.00 0.95 0.97 1.00 0.96 0.99 Constant 0.74 0.70 0.40* 0.14*** 0.04*** 0.06*** 0.12*** 0.09** 0.06***

N 11921 3397 2464 11921 3397 2464 11921 3397 2464 * p<.05; ** p<.01; *** p<.001

(29)

Individual and Household Characteristics

The probability of people wanting to move within Amsterdam decreases with age and is an important factor across all ethnicities. The risk for native Dutch to state a moving wish within the city (versus staying in their current dwelling) is 61 percent lower for middle aged people and 78 percent lower for old people compared to the young (controlling for the other variables). Interestingly, old non-Western migrants are more inclined to move within Amsterdam than their native counterparts with a 60 percent lower risk to move (versus “stay”) relative to young non-Westerners. This could possibly be attributed to their overrepresentation in the social housing sector and the resulting higher level of dissatisfaction with their housing situation. At the same time older natives and Western migrants could already have found housing that suits their needs by renting in pricier segments or buying a house or an apartment.

Whether the respondents are male or female is statistically insignificant for almost all the groups and their preferred moving destinations with a few exceptions. Notably, there are low levels of significance indicating that non-Western women are more inclined to move within Amsterdam (relative to “stay”), native woman have a stronger wish to move to the suburbs while having lower preference for long-distance moves than their respective male counterparts.

With higher levels of education people are increasingly inclined to move. There is 67 percent higher chance for highly educated native Dutch of wanting to move within Amsterdam compared with lower educated natives. Interestingly, the effect of education on natives preferring the suburbs as place of residence is insignificant. An urban lifestyle might especially appeal to highly educated natives (van Diepen & Musterd 2009) while less educated natives are in equal parts tied to suburban living (Karsten 2007). For non-Western migrants, on the other hand, education increases the propensity to move to the suburbs significantly as highly educated non-Westerners are 173 percent more likely to state their inclination to move there than their lower educated counterparts. This pinpoints to the interpretation that living in the suburbs is a particularly aspired goal for the social climbers among non-Western migrants in contrast to natives and western migrants where such an effect cannot be observed. Highly educated people among all ethnic groups are about three times more likely (relative to the lower educated) to state a preference for long-distance moves, which are commonly considered to be job related.

There is little to no significance indicating that higher income increases the moving propensity of Western and non-Western migrants. In contrast, native Dutch in the highest income category are 36 percent more likely to move within Amsterdam and 63 percent more likely to move to the suburbs than natives in the lowest income category. Long-distance moves are not more likely for the high-earners in relation to the lowest income categories across all ethnic groups.