Trajectories of neighborhood change

Neighborhoods represent a scale at which inequalities are reflected in the unequal spatial distribution of ethnic and income groups across urban space. In many cities, the rich reside in high-quality neighborhoods in favorable locations while the poor are concentrated in disadvantaged areas (Hulchanski, 2010; Van Eijk, 2010). However, neighborhoods are not static entities and spatial patterns of socioeconomic and ethnic inequality shift over time as a result of processes of neighborhood change. Neighborhoods can develop in different ways: (1) they can gentrify which is characterized by rising house prices and the replacement of lower income groups by higher income groups (Hochstenbach & Van Gent, 2015; Newman & Wyly, 2006; Slater, 2006); (2) neighborhoods can decline which is indicated by physical deterioration and declining house prices and the succession of higher income groups by lower income groups (Grigsby et al., 1987; Prak & Priemus, 1986; Van Beckhoven et al., 2009); (3) neighborhoods can remain stable in their population composition and/or overall status for longer periods of time (Meen et al., 2013; Tunstall, 2016).

There are two empirical gaps in the literature on neighborhood change that this dissertation addresses. First, there has been a lack of longitudinal studies. Many studies on neighborhood change take on a relatively short-term perspective and reduce change to the difference between two points in time. While the literature has been dominated by case-studies on gentrification or decline, fuelling the assumption that gentrification and decline are widespread processes that quickly transform neighborhoods and cities, a growing body of research suggests that neighborhoods are rather ‘slothful’ and that neighborhood change takes time to take effect (Tunstall, 2016; Meen et al., 2013). Overall, we have little insight into the extent to which gentrification and decline are exceptional cases, in addition to, the prevalence and rate of change across all neighborhoods over time (cf. Tunstall, 2016).

Second, residential mobility is often seen as the most important driver of neighborhood change. However, residential mobility is shaped by structural factors such as the housing stock, local housing markets, and government policy (Meen et al., 2013; Nygaard & Meen, 2011). Moreover, researchers have argued that residential mobility should be understood in relation to demographic and in-situ change, which can also play an important role in processes of neighborhood change (Bailey, 2012; Finney & Simpson, 2009; Teernstra, 2014). The relative impact of the housing stock and different population dynamics on neighborhood change has however received little attention in the literature to date.

This dissertation contributes to the literature on longitudinal neighborhood change, both theoretically and methodologically. Theoretically, it provides insight into diverging pathways of neighborhood change over time, illustrating how different mechanisms interact to shape the urban geography along socioeconomic and ethnic lines. The path-dependent role of the housing stock is analyzed, in addition to the extent to which changes to the housing stock as a result of urban restructuring affect residential mobility and neighborhood change. Moreover, this dissertation investigates patterns of ethnic segregation over time and explores the relative impact of residential mobility and demographic change. Methodologically, this dissertation explores innovative methods for the analysis of neighborhood trajectories, broadening the scope of statistical methods for the field of neighborhood change research.

This dissertation uses individual-level administrative data from the System of Social statistical Datasets (SSD) provided by Statistics Netherlands. The SSD contains longitudinal geocoded data on the full Dutch population, as well as information on the built environment. As such, the SSD allows for the analysis of the relationship between the housing stock and population change in processes of neighborhood change. Neighborhoods are operationalized using 500 by 500 meter grids, which are the most consistent low spatial scale over time. Three out of four chapters focused on the 1999 to 2013 time period, while chapter 3 covered the 1971 to 2013 period. This dissertation employed innovative methodologies to analyze trajectories of neighborhood change over time. Chapter 3 presents a combination of sequence analysis and a tree-structured discrepancy analysis that allows for the visualization of neighborhood pathways and its relation to their contexts. Chapter 5 uses a Latent Class Growth Model (LCGM) to categorize neighborhoods based on similarities in the timing and pace of change over time. Both methodologies have proven to be valuable tools for the identification of diverging neighborhood pathways over time.

Summary of chapters

This dissertation is comprised of five separate, but related papers. Chapter 2 presents a literature review of theories and studies on neighborhood decline. Chapters 3 to 6 are empirical research papers that have their own theoretical framework, empirical analyzes, results and discussion section. All papers have either been published in peer-reviewed journals or are currently under review. The content of chapters 2 to 6 is summarized below. The Global Financial Crisis and neighborhood decline

Chapter 2 presents an overview of the literature and theories on the spatial consequences of the Global Financial Crisis (GFC). The impact of the GFC and the economic recession that followed is unevenly distributed between households and individuals, with low-income and vulnerable households being affected the most. As such, it can be expected that the consequences of the GFC are most pronounced in disadvantaged neighborhoods. While many studies have investigated the effects of the GFC on the economy and/or housing markets, only a few studies have focused on the unequal geographical impacts of the GFC (Batson & Monnat, 2015; Foster & Kleit, 2015). This chapter bridges two streams of literature by formulating ten ways in which the GFC might accelerate processes of neighborhood decline. The main goal of this chapter is to further the intellectual debate on neighborhood decline and to call for more longitudinal research on the ways in which the GFC has affected neighborhood trajectories and spatial patterns of increasing inequality.

The path-dependency of low-income neighborhoods

Chapter 3 presents an innovative longitudinal approach to analyzing neighborhood change and investigates the trajectories of low-income neighborhoods in the 31 largest cities in the Netherlands over the 1971 to 2013 period. Many studies on neighborhood change are limited by relatively short-term perspectives, and/or a focus on specific case-studies of gentrification or decline (e.g. Bailey, 2012; Jivraj, 2013; Hochstenbach & Van Gent, 2015). As such, it is unclear to what extent neighborhoods with similar characteristics experience the same process of change over time — or to what extent gentrification or decline are the exception to the rule. Using sequence analysis and a tree-structured discrepancy analysis, this chapter contributes to the literature by analyzing how housing stock characteristics shape neighborhood trajectories over longer periods of time. The results show that neighborhoods exhibit a high degree of path-dependency. Neighborhoods with high shares of social housing in 1971 display a pattern of increased poverty concentration and neighborhood decline over time. By way of contrast, increases in the share of owner-occupied housing contribute to more upward neighborhood trajectories.

The effects of physical restructuring on neighborhoods

Chapter 4 analyzes the effects of urban restructuring programs on neighborhood change in the 31 largest Dutch cities. Researchers have been critical about the effectiveness of urban restructuring in actually achieving upgrading neighborhoods (e.g. Lawless, 2011; Permentier et al., 2013; Tunstall, 2016; Wilson, 2013). However, many studies have been faced with methodological limitations with respect to measuring urban restructuring, spatial scale, and time periods. Chapter 4 overcomes these limitations by focusing on the effects demolition and new construction on a low spatial scale over a 15- year period. Using propensity score matching, this chapter finds a positive causal effect of demolition and new construction on neighborhood upgrading. The results indicate that large-scale demolition and new construction leads to socioeconomic upgrading of deprived neighborhoods as a result of attracting and maintaining middle- and high-income households. Urban restructuring appears to have negative spillover effects in terms of an increased share of low-income households in other neighborhoods.

Trajectories of ethnic neighborhood change

Chapter 5 focuses on trajectories of ethnic neighborhood change in the four largest Dutch cities, Amsterdam, Rotterdam, The Hague and Utrecht, between 1999 and 2013. As the share of ethnic minorities continues to grow in many cities, this raises concerns about increasing levels of ethnic segregation. The literature has been divided on the methods for analyzing ethnic segregation over time and many researchers have relied on single-number indices or typologies based on arbitrary thresholds (e.g. Duncan & Duncan, 1955; Johnston et al., 2010; Massey & Denton, 1993; Peach, 1996; Poulsen et al., 2001). Chapter 5 presents an innovative alternative for the identification of trends in the ethnic population composition over time. Using LCGMs, this chapter finds that neighborhoods show relative stability in the ethnic population composition over time, despite a substantial growth in the ethnic population. Although ethnic minorities are increasingly moving away from concentration neighborhoods, processes of natural growth play an important role in maintaining levels of ethnic segregation.

Intergenerational continuity of ethnic segregation

Chapter 6 investigates persistent patterns of ethnic segregation over the course of generations. In the literature, it is assumed that ethnic segregation will decrease over the course of generations as later generations will be more socially and economically integrated in society (e.g. Massey, 1985). This assumption is reflected in the official Dutch definition of ethnicity that classifies individuals whose parents are born in the Netherlands, but with one or more immigrant grandparents, as native Dutch. The use of this definition has important empirical consequences and influences conclusions about ethnic neighborhood change. Focusing on the residential patterns of third generation parental home-leavers in the 31 largest Dutch cities between 1999 and 2013, this chapter illustrates that third generation ethnic minorities continue to be overrepresented in ethnic concentration neighborhoods. The intergenerational continuity of socioeconomic disadvantage among ethnic minorities plays an important role in persistent ethnic segregation over time.

Findings and conclusions

The findings of this dissertation contribute to the field of neighborhood change research in four ways. First, this dissertation has demonstrated that neighborhoods tend to be relatively stable in their socioeconomic and ethnic status over time and that neighborhood change takes several decades to take effect. Second, this dissertation underlines the determining role of the housing stock in processes of neighborhood change. Neighborhoods exhibit a high degree of path-dependency where the initial quality of the built environment is reinforced over time. Chapter 3 has illustrated that the share of social housing is an important determinant of future processes of neighborhood decline. Changes to the housing stock, however, have the ability to alter the trajectories of neighborhoods. Chapter 4 has shown that large-scale demolition and new construction as a result of urban restructuring programs has led to neighborhood upgrading by attracting and maintaining higher income groups. Third, this dissertation has illustrated how different population dynamics interact to maintain the status quo. Chapter 5 and 6 have identified persistent patterns of ethnic segregation over time as a result of socioeconomic disadvantage among ethnic minorities which leads to high residential mobility rates into ethnic concentration neighborhoods. Although residential mobility is often seen as the most important driver of neighborhood change, this dissertation adds to the growing literature on the role of demographic change. The effects of ethnic residential mobility out of concentration neighborhoods on ethnic segregation are mitigated by processes of natural growth. Fourth, this dissertation has explored innovative methods for the analysis of longitudinal patterns of neighborhood change. Sequence analysis in combination with a tree-structured discrepancy analysis allows for a detailed analysis of neighborhood trajectories and the relationship with their contexts. LCGMs enable the identification of diverging neighborhood patterns of change based on timing and pace.

Challenges and limitations

Despite the contributions to the literature, this dissertation is also faced with several limitations, three of which are highlighted below. First, this dissertation has analyzed patterns of neighborhood change, but has not directly focused on gentrification. While some view urban restructuring as a form of state-led gentrification (e.g. Uitermark & Bosker, 2014), this dissertation sees urban restructuring as fundamentally different from more natural processes of gentrification. The term gentrification has become widely used (and abused) for a wide variety of different and, sometimes unrelated, processes leading to neighborhood upgrading. Future research would benefit from clearly defining gentrification and for analyzing gentrification over longer periods of time. Currently, we have very little insight in the prevalence, rate, and extent of gentrification across neighborhoods and cities and it is unclear to what extent its effects are temporary or long-lasting.

Second, this dissertation has limited its focus on the four largest ethnic groups in the Netherlands. However, the spatial distribution of these four ethnic groups is likely to be related to the residential behavior and distribution of other ethnic groups in the Netherlands. Future research would benefit from comparing patterns of segregation across different ethnic groups and the ways they interact to shape the urban geography along ethnic lines.

Third, the innovative methods employed in this dissertation enable the analysis of patterns of neighborhood change, however, they are not without limitations. Both methods allow for the identification of groups of neighborhoods that follow similar trajectories over time. However, these methods are faced with a degree of uncertainty around the true number of groups. In addition, a tree-structured discrepancy analysis uses the most significant values of the predictor variables as cut-off points, however, it is unclear to what extent these values can be interpreted as threshold values in processes of neighborhood change. Overall, these limitations reflect the nature of the modelling process and underlines the need to string theoretical reasoning beneath the models.

Policy implications

This dissertation has underlined the relative stability of neighborhoods over time. Policy makers should keep in mind that neighborhood change takes time to take effect, often exceeding standard policy time periods. Large-scale changes to the housing stock in the context of urban restructuring programs have the ability to generate neighborhood change by stimulating selective residential mobility. However, the positive effects of urban restructuring are limited to the restructured neighborhood. Other neighborhoods appear to suffer from negative spillover effects, illustrated by an increase in the share of low-income households as a result of displacement.

The GFC has accelerated the shift towards the marketization of social housing. Some cities aim to stimulate gentrification through the sales of social housing which reduces the size and quality of the social housing stock. The spatial consequences of such policies are however unclear and may take time to take effect. Policy makers should be aware that reducing the size and quality of the social housing stock in large cities complicates the accessibility of cities for low-income groups and can have a major impact on the urban geography of cities and regions.

This dissertation has found persistent patterns of ethnic segregation which can be explained by intergenerational ethnic disadvantage. The question remains to what extent spatial patterns of ethnic disadvantage should be targeted by urban (re)development. As studies have shown that ethnic socioeconomic mobility tends to lead to more residential opportunities and spatial dispersal, it could be more beneficial to invest in education and labor market participation.

Last, this dissertation has illustrated that official definitions of ethnicity can influence empirical conclusions. Ethnic origin is based on the country of birth of the parents, however, this indicator ignores other aspects of ethnic origin. Later generations of ethnic minorities might still be characterized by other aspects of ethnic origin that play an important role in group inequalities. As society is becoming increasingly diverse, policy makers should be sensitive to ethnic differences and group inequalities that are not directly reflected in official statistics.