Ethnic and socioeconomic segregation

in BelgiumA multi-scalar approach

using individualised neighbourhoods

Rafael Costa∗ and Helga de Valk†

PAA – 2016 Annual MeetingExtended abstract

(Draft: please do not cite or quote)

Background and aims

Residential segregation is a persistent reality in Europe (Bolt, 2009; Cassiers and Keste-loot, 2012). Ethnic minorities and deprived populations are often clustered in specificneighbourhoods, a situation that not only endangers social cohesion, but also lowersindividuals’ outcomes in terms of employment, educational achievement and politicalparticipation (Crane, 1991; Sampson et al., 2002; Murie and Musterd, 2004; Lichteret al., 2012; Andersson and Malmberg, 2014). In Belgium, a considerable share of theurban population lives in disadvantaged neighbourhoods (Musterd, 2005), and urbandeprivation is often associated with a concentration of ethnic origin groups, as in the caseof the inner city of Brussels (Kesteloot and Van der Haegen, 1997; Dujardin et al., 2008).

The vast majority of the existing studies have examined segregation patterns basedon predefined areal units such as municipalities, neighbourhoods and statistical wards.Such areal units often differ in size, function and distribution across different regionsand over time. This problem, known as Modifiable Areal Unit Problem (MAUP), affectsany quantitative analyses of segregation (Openshaw, 1984; Wong, 2004; Reardon andO’Sullivan, 2004) and conceal processes that operate at different scales (Andersson andMusterd, 2010)

Moreover, only few studies so far have investigated the interactions between ethnic andsocioeconomic segregation, especially in the European context (see for example Peach(1999)).

∗Vrije Universiteit Brussel†Netherlands Interdisciplinary Demographic Institute / Koninklijke Nederlandse Akademie van

Wetenschappen / University of Groningen / Vrije Universiteit Brussel

1

The purpose of this paper is twofold. It aims at (i) analysing the interplay betweensocioeconomic and ethnic segregation in Belgium and in Brussels; and (ii) investigatinghow this interplay operates at different scales.

Data and methods

Our study is part of a broader international project – ‘Residential segregation in fiveEuropean countries’ (ResSegr) –, which uses an innovative approach in order to producemore accurate measures of segregation in Belgium, Denmark, the Netherlands, Norwayand Sweden. Based on geocoded data available in these five countries we constructindividualised neighbourhoods by expanding a geographical buffer around grids of 100square metres, until the buffer contains a pre-determined number of neighbours (Östh,2014). The resulting sample of neighbours is then used to compute aggregate statistics.

In this paper we use the geocoded data from the 2011 Belgian Census in orderto produced geocentric neighbourhoods at eleven different scales, ranging from 50 to51,200 nearest neighbours, and compute five indicators of socioeconomic and ethniccharacteristics:

• share of the population with foreign origin in a EU/EFTA country;• share of the population with foreign origin in a non-EU/EFTA country;• share of persons 25–64 years old who have completed tertiary education;• share of persons 30–59 years old in employment;• share of persons who received social assistance in 2011.

Our final dataset contains therefore 55 variables (i.e. each one of the five indicators ateleven different scales) corresponding to the coordinates of each of the 100m2 grids of theBelgian territory.

In a first descriptive step, the data is used to assess segregation patterns using maps,and to study the bivariate correlations between ethnic and socioeconomic indicators.Next, in order to seize the complex interactions between the indicators at the differentscales, we apply factor analyses on the ensemble of variables, in a similar approachproposed by Andersson and Malmberg (2014).

The analyses focus on Belgium as a whole, as well as on the Brussels Agglomerationseparately. We expect the segregation patterns in the capital to be particularly interestingdue to the strong presence of foreign non-European populations on the one hand, andwell-off European expats on the other hand.

Preliminary findings and expected results

The preliminary tests with factor analysis reveal the interesting associations betweenindicators and scales. We obtain seven factors for Belgium (with eigenvalue greater than1), which together account for 87% of the total variance. They reflect different dimensionsof the residential segregation in Belgium, some at the immediate neighbourhood, othersover longer distances.

2

As a matter of illustration, figures 1 and 2 show the first two factors. Both are directlyconcerned with the spatial concentration of foreigners, combined with other socioeconomiccharacteristics.

-0.5

-0.3

-0.1

0.1

0.3

0.5

0.7

0.9

0 10000 20000 30000 40000 50000 60000

Foreigners EU

Employment

(a) Indicators’ factor loadings by number of nearest neighbours

(b) Spatial distribution of factor scores

Figure 1: Belgium, factor 1 (19% variance): European migration and employment

The first factor (19% of the total variance) expresses the spatial clustering of foreignpopulations from European origin, associated to some extent with unemployment. Thisis namely the case of the old industrial belt in the south of Belgium, which attractedan important number of foreign workers in the past – namely from Italy – before the

3

-0.5

-0.3

-0.1

0.1

0.3

0.5

0.7

0.9

0 10000 20000 30000 40000 50000 60000

Foreigners non-EU

Social assistance

Employment

(a) Indicators’ factor loadings by number of nearest neighbours

(b) Spatial distribution of factor scores

Figure 2: Belgium, factor 1 (18% variance): Non-European migration and social assistance

4

industrial decline. Brussels, the European capital, also scores high in this factor, as wellas the areas located close to the Dutch border. The map of the factor scores presents aninteresting contrast between most of Flanders in the north – where employment rates arehigher and there is less concentration of foreigners – and the French-speaking region inthe south.

The second factor (18% of the total variance) is dominated by the concentration ofpeople of foreign origin in a non-European country. The urban areas clearly stand outin this factor - namely Brussels and its outskirts. The factor is also related, althoughin a lesser extent, with high proportions of unemployment and of people who benefitfrom social assistance. It can therefore be interpreted as the clustering of both ethnicminorities and poor socioeconomic conditions in the Belgian cities, especially at a shortdistance (the factor loadings peak around the 3,000 nearest neighbours).

The different factors obtained for Belgium and for Brussels will be thoroughly analysedin our paper. We expect that our approach combining individualised neighbourhoods andfactor analysis will produce a good overview of the intricate interactions between ethnicand socioeconomic segregation operating at different levels. Furthermore, the mapping offactor scores will offer a detailed picture of segregation patterns in a multi-dimensionaland multi-scale perspective. Not only does our approach give compelling results forBelgium – as the preliminary tests suggest – but it may just as well provide an interestingway of assessing segregation patterns in other countries for comparison.

References

Andersson, E. K. and Malmberg, B. Contextual effects on educational attainment inindividualised, scalable neighbourhoods: Differences across gender and social class.Urban Studies, pages 1–17, 2014.

Andersson, R. and Musterd, S. What Scale Matters? Exploring the Relationships betweenIndividuals’ Social Position, Neighbourhood Context and the Scale of Neighbourhood.Geografiska Annaler: Series B, Human Geography, 92(1):23–43, 2010.

Bolt, G. Combating residential segregation of ethnic minorities in European cities.Journal of Housing and the Built Environment, 24:397–405, 2009.

Cassiers, T. and Kesteloot, C. Socio-spatial Inequalities and Social Cohesion in EuropeanCities. Urban Studies, 49(9):1909–1924, 2012.

Crane, J. The Epidemic Theory of Ghettos and Neighborhood Effects on Dropping Outand Teenage Childbearing. American Journal of Sociology, 96(5):1226–1259, 1991.

Dujardin, C.; Selod, H., and Thomas, I. Residential Segregation and Unemployment:The Case of Brussels. Urban Studies, 45(1):89–113, 2008.

Kesteloot, C. and Van der Haegen, H. Foreigners in Brussels 1981–1991: SpatialContinuity and Social Change. Tijdschrift voor Economische en Sociale Geografie, 88(2):105–119, 1997.

5

Lichter, D. T.; Parisi, D., and Taquino, M. C. The Geography of Exclusion: Race,Segregation, and Concentrated Poverty. Social Problems, 59(3):364–388, 2012.

Murie, A. and Musterd, S. Social Exclusion and Opportunity Structures in EuropeanCities and Neighbourhoods. Urban Studies, 41(8):1441–1459, 2004.

Musterd, S. Social and Ethnic Segregation in Europe: Levels, Causes and Effects. Journal

of Urban Affairs, 27(3):331–348, 2005.

Openshaw, S. The modifiable areal unit problem. Concepts and Techniques in Modern

Geography, 38(Geo Books, Norwich), 1984.

Peach, C. London and New York: Contrasts in British and American Models of Segrega-tion. International Journal of Population Geography, 5:319–351, 1999.

Reardon, S. F. and O’Sullivan, D. Measures of Spatial Segregation. Sociological Method-

ology, 34(1):121–162, 2004.

Sampson, R. J.; Morenoff, J. D., and Gannon-Rowley, T. “Neighborhood Effects": SocialProcesses and New Directions in Research. Annual Review of Sociology, 28:443–478,2002.

Wong, D. W. S. Comparing Traditional and Spatial Segregation Measures: A SpatialScale Perspective. Urban Geography, 25:66–82, 2004.

6