Socioeconomic Segregation in European Cities. a Comparative Study of Brussels, Copenhagen, Amsterdam, Oslo, and Stockholm

Rafael Costa, Vrije Universiteit Brussel
Bo Malmberg, Stockholm University
Karen Haandrikman, Stockholm University, Department of Human Geography
Adrian Farner Rogne, University of Oslo
Bart Sleutjes, Netherlands Interdisciplinary Demographic Institute

The purpose of this study is to investigate socioeconomic segregation patterns and levels in Brussels, Copenhagen, Amsterdam, Oslo and Stockholm. Previous studies have shown that socioeconomic segregation is a persistent reality in European cities and that it can have negative effects on social cohesion and individuals’ outcomes. Few studies so far, however, have investigated socioeconomic segregation from an international comparative perspective, mainly due to the lack of appropriate and comparable measures across countries. Yet, comparative studies of cities under different housing policies and economic contexts can help elucidate the underlying processes that produce and sustain segregation. In our study we benefit from innovative geocoded data from Belgium, Denmark, the Netherlands, Norway and Sweden, which allow us to compute comparable measures of socioeconomic segregation in 2011 for the five capitals. First we delimit the five metropolitan areas within 25km-radius circles and divide these areas into small-scale grids. We then expand geographic buffers around each grid using the EquiPop software until we obtain samples of the 200, 1600, 12800 and 51200 nearest neighbours. For each sample of nearest neighbours, we then calculate two indicators of socioeconomic composition: (i) at risk of poverty: the share of people aged 25 and above with disposable income below 60% of the national median; and (ii) high income: share of people aged 25—64 with net earned income in the highest decile. In this way, we obtain comparable measures of the distribution of poverty and affluence in the five cities at different scales—from individuals’ immediate surroundings until urban areas—with a high level of geographic detail and independently of administrative borders. We analyse these measures using maps, segregation indices and percentile plots. The differences in levels and patterns across cities are interpreted in the light of their particular housing systems and territorial processes.

Background and aims

Socioeconomic segregation is a persistent reality inEuropean cities (Cassiers and Kesteloot, 2012; Musterd, 2005; Tammaru et al.,2015) and a subject of concern in urban policy (Bolt, 2009; Galster, 2007;Kleinhans, 2004). An extensive literature has shown that the concentration ofdeprived populations in specific neighbourhoods can represent a threat tosocial cohesion (Cassiers and Kesteloot, 2012), hindering citizen participation(Kühn, 2015), access to the labour market (Andersson, 2004; Dujardin et al.,2008), education attainment (Andersson and Malmberg, 2014) and even lead tourban unrest and riots (Malmberg et al., 2014; Olzak et al., 1996).

Few studies so far have investigated socioeconomicsegregation from an international comparative perspective (see Tammaru et al.,2015). In fact, spatial data are typically available for predefined arealunits—such as municipalities, wards, etc.—, which differ greatly in size,function and distribution from one country to the other. These differences inspatial units make it hard to compare segregation levels and patterns acrosscountries in an accurate way. Yet, comparative studies of cities underdifferent housing policies and economic contexts can help elucidate theunderlying processes that produce and sustain segregation.

The main purpose of our study is to investigatesocioeconomic segregation patterns and levels in Brussels, Copenhagen,Amsterdam, Oslo and Stockholm. To this end, we benefit from harmonised datasetsin the five countries containing geocoded indicators based on anearest-neighbours approach. These datasets allow us to produce comparablemeasures of socioeconomic segregation with a fine level of geographic detailand at multiple scales. Our analyses offer an unprecedented comparison ofpatterns and levels of socio-spatial inequalities in five European capitals,each with its particular housing system and territorial history.

Data and methods

Our analyses focus on the metropolitan areas of Brussels,Copenhagen, Amsterdam, Oslo and Stockholm. For comparability, we use ageometric definition of metropolitan area as a circle of 25-kilometres radiusencompassing each city.

For each metropolitan area we construct individualisedscalable neighbourhoods. First, the urban territory is divided into small-scalegrids. Next, we use individual geocoded data in 2011 available in the fivecountries to identify the population in each grid. The neighbourhoods areconstructed by expanding a geographic buffer around each grid using the EquiPopsoftware (Östh et al., 2015) until we obtain samples of the 200, 1600, 12800and 51200 nearest neighbours. For each sample of nearest neighbours, we thencalculate two indicators of socioeconomic composition (Nielsen et al., 2017):(i) at risk of poverty, or the share of people aged 25 and above withdisposable income below 60% of the national median; and (ii) high income, theshare of people aged 25–64 with net earned income in the highest decile. Inthis way, we obtain comparable measures of the distribution of poverty andaffluence in the five cities at different scales—from individuals’ immediatesurroundings until urban areas—with a high level of geographic detail andindependently of administrative borders.

The two indicators will be analysed in different ways.First, their mapping in the five metropolitan areas will give a detailedpicture of segregation patterns. Segregation levels will be compared usingsynthetic segregation indices (dissimilarity index). Finally, we will usepercentile plots of the distribution of the population in poor and affluentneighbourhoods to compare segregation patterns across cities and neighbourhoodscales. The differences in levels and patterns across cities will beinterpreted in the light of their particular housing systems and territorialprocesses.

Preliminary findings and future work

At this point, we have constructed the individualisedscalable neighbourhoods in the five countries and calculated the two indicatorsof the neighbourhoods’ socioeconomic composition. The first maps produced forBrussels and Stockholm (figures 1 and 2) show the interesting contrasts in thespatial patterns of poverty in these cities. In particular, the maps depict thestriking concentration of poor in Brussels inner city, whereas the segregationof poor in Stockholm is located in off-centre areas. In the light of thesefirst tests, we expect the complete comparative analyses to offer importantinsights into the underlying forces that shape socioeconomic segregation.

Figure 1: Brussels: share ofpeople at risk of poverty among the 1600 nearest neighbours, 2011

Figure 2: Stockholm: share ofpeople at risk of poverty among the 1600 nearest neighbours, 2011

Presented in Session 1220: Internal Migration and Urbanization