Segregation, Integration and Neighbourhood Effects Debates and Analyses Sako Musterd University of Amsterdam University of Manchester CCSR Seminar June.

Segregation, Integration and Neighbourhood Effects

Debates and Analyses

Sako Musterd

University of Amsterdam

University of Manchester

CCSR Seminar

June 3, 2008

Central questions

A. Debates: what are the prevailing ideologies, perceptions, assumptions and policy responses regarding segregation and its potential (neighbourhood) effects?

B. Analyses: to what extent are these perceptions, etc. informed by theoretical knowledge and empirical evidence?

Outline (debates and analyses)

1. Key concepts and prevailing ideologies and perceptions

2. Theories on segregation and concentration3. Theories on neighbourhood effects4. Segregation and concentration; levels and

dynamics5. Segregation and neighbourhood effects

6. Implications for urban policies

A

B

1Key concepts and prevailing

ideologies and perceptions

Key concepts• Segregation• Concentration• Neighbourhood effects

– Individual opportunities– Livable neighbourhoods

• Integration (participation)• Assimilation

Prevailing ideologies and perceptions • Vague use of key concepts (segregation, concentration,

integration, assimilation)• Segregation levels are regarded as high and increasing• Lack of integration is seen as key neighbourhood problem• Segregation is seen as the cause and thus as ‘bad’• Segregation would create negative neighbourhood effects• Fear for parallel societies and a strong call for assimilation• Neighbourhood restructuring and housing mix as panacea

1Key concepts and prevailing

ideologies and perceptions

2 Theories of segregation and concentration

• Globalisation

• Economic restructuring• Welfare regime (special attention in next slide)

• Cultural (language, religion, discrimination, identity, level of acceptance of inequality, tolerance towards difference, eagerness to ‘enforce’ integration)

• Historic social, economic and cultural urban paths• Political attitudes towards diversity (ideas regarding

assimilation; multiculturalism and mix)

2 Theories of segregation and concentration

• Welfare regime

– Benefit systems for unemployed, elderly and disabled

– Access to high quality education– Access to housing– Labour market access– Housing benefits– Health care systems access– Income redistribution

2 Theories of segregation and concentration

• Segregation is a strong process, reflecting the relationship between spatial inequality and social inequality, lifestyle differences, and difference in terms of other resources

• Segregation is influenced by global, national, local and group level processes, structural and individual factors; and thus not simply to modify with single sector policies, such as housing policies

3Theories on neighbourhood effects

(theories on segregation effects)

• Socialisation processes (role models)

• Social networks (communication)

• Stigmatisation

• Spatial mismatch

4 empirical findings

1. Key concepts and prevailing ideologies and perceptions

2. Theories on segregation and concentration3. Theories on neighbourhood effects4. Segregation and concentration; levels

and dynamics5. Segregation and neighbourhood effects

6. Implications for urban policies

A

B

4

Levels of ethnic segregation:

most segregated groups per city

12 EU countries 24 cities

0-100 : low-high segregation0 20 40 60 80 100

Munich Foreigners

Frankfurt Turks

Milan non-Italian

Milan Moroccans

Frankfurt Americans

Paris Algerians Dept. 75

Lille non-French

Madrid Moroccans

Oslo 3rd w orld immigrants

Vienna Foreigners

Düsseldorf Turks

Amsterdam Turks

Lisbon Cape Verdian

Rotterdam Turks

The Hague Turks

Stockholm Iranian 14 municip.

Brussels Moroccan

Bristol Pakistani

London Bangladeshi

Manchester Bangladeshi

Barcelona Filipinos

Birmingham Bangladeshi

Antw erp N. African, Bosnian

Bradford Bangladeshi

Leicester Bangladeshi

Oldham Bangladeshi

Oldham Pakistani

US 6 metrop. areas Blacks

4 Levels of ethnic segregation; impact of

area-size (index 0-100 = low-high segregation)

0 20 40 60 80 100

Amsterdam Turks 1216 grids

Amsterdam Turks 369 neighbourhoods

Amsterdam Turks 93 neighbourhoods

Amsterdam Turks metro area 30 distr

Birmingham Bangladeshi ED

Birmingham Bangladeshi w ards

London Bangladeshi 15300 ED

London Bangladeshi 782 w ards

Levels of ethnic segregation:

group comparison

index (0-100 = low-high segregation)

0 10 20 30 40 50 60 70 80

Amsterdam SurinameseAmsterdam Moroccans

Amsterdam Turks

Rotterdam SurinameseRotterdam Moroccans

Rotterdam Turks

The Hague SurinameseThe Hague Moroccan

The Hague Turks

London Black AfricanLondon Black Caribbean

London PakistaniLondon Bangladeshi

Manchester Black CaribbeanManchester Pakistani

Manchester Bangladeshi

Birmingham Black AfricanBirmingham Black Caribbean

Birmingham PakistaniBirmingham Bangladeshi

Leicester Black CaribbeanLeicester Pakistani

Leicester Bangladeshi

Bradford Black CaribbeanBradford Pakistani

Bradford Bangladeshi

Oldham Black CaribbeanOldham Pakistani

UK

Netherlands

0 10 20 30 40 50 60 70

Barcelona Peruvians

Barcelona Moroccans

Barcelona Filipinos

Madrid Latins American

Madrid Peruvians

Madrid Moroccans

Milan Egyptians

Milan Filipinos

Milan Moroccans

Lisbon Brazilians

Lisbon Africans

Lisbon Cape Verdians

Levels of ethnic segregation:

group comparison

index (0-100 = low-high segregation)

Spain

Portugal

Italy

4 Segregation levels and dynamics

in some Dutch cities (ethnic)

1980 1995 2000 2004 1980 1995 2000 2004 1980 1995 2000 2004

Turks 37.3 40.7 41.2 42.4 - 51.7 47.8 44.1 66.4 54.6 51.3 51.1

Moroccan 38.6 39.1 39.5 40.0 - 46.8 42.6 39.7 64.7 49.9 48.8 48.3

Surinamese 27.8 34.8 33.3 32.9 - 28.6 24.1 21.1 - 40.2 37.0 33.5

Antillean 26.2 34.9 37.1 33.3 - 28.5 30.2 29.7 - 25.5 27.3 28.1

Amsterdam Rotterdam The Hague

increasing/stable; decreasing; decreasing

London

20,0

30,0

40,0

50,0

60,0

70,0

1991 2001

ID

Bangladeshi Indians Pakistani Carribean

birmingham

20,0

30,0

40,0

50,0

60,0

70,0

1991 2001

ID

Bangladeshi Indians Pakistani Carribean

Rotterdam

20,0

30,0

40,0

50,0

60,0

1995 2000 2004

ID

Turks Moroccan Surinamese Antillean

Amsterdam

20

30

40

50

1980 1995 2000 2004

ID

Turks Moroccan Surinamese Antillean

Segregation dynamics

Concentrations in Amsterdam and the Amsterdam metropolitan region

Concentrations of four population categoriesT M

S A

Amsterdam2004

But don’t be mislead by graphs of spatial concentrations

ConcentrationAmsterdam Surinamese 2004> 2sd above the mean> 19.8%In concentrations: 33%Of all Surinamese: 38%

These figures were the same in

1994!

Strong concentrationAmsterdam Surinamese 2004> 4sd above the mean> 27.8%In concentrations: 38%Of all Surinamese: 31%

These figures were the same in

1994!

Ethnic neighbourhoodAmsterdam Surinamese 2004 > 50%In concentrations: 57%Of all Surinamese: 2.9%

‘Little Surinam’?2004> 60%In concentrations: 65%Of all Surinamese: 0.6%

Dynamics: % that is Turkish [same for other categories] in so-called Turkish concentrations 1994-2007, Amsterdam

0

10

20

30

40

1993 1998 2003 2008

%

Turkish

Moroccan

Surinamese

Antillean

Dynamics: % of all Turkish in the city [same for other categories] living in so-called Turkish concentrations 1994-2007, Amsterdam

10

20

30

40

50

1993 1998 2003 2008

%

Turkish

Moroccan

Surinamese

Antillean

Dynamic Moroccans 2007

1973

Amsterdam region, ‘non-western’, 2000> 4sd above the mean> 48%In concentrations: 63%Of all non-western: 50%

Amsterdam region, ‘non-western’, 2004> 4sd above the mean> 51.5%In concentrations: 66%Of all non-western: 49%

Ethnic concentrations are unstable 1994-2004 change relative to 1994; Turkish

concentrations in Amsterdam

Ethnic concentrations are unstable 1994-2004 change relative to 1994; Moroccan

concentrations in Amsterdam

Ethnic concentrations are unstable 1994-2004 growth rates in concentrations relative

to the expected growth on the basis of the development in Amsterdam as a whole

4. Levels

Index of SegregationSocio-Economic Categories

0 10 20 30 40 50

Copenhagen 1st quintile

Amsterdam 1st quintile metro

Bern unemployed

Berlin hh income < € 900

Birmingham income support

Milan blue collar w orkers

Manchester income support

Manchester unemployed

Amsterdam 1st quintile

Berlin hh income > € 3500

Lille unemployed vs employed

Rotterdam 1st quintile

Amsterdam 5th quintile metro

Oslo social assistance

Birmingham unemployed

Leeds unemployed

Sheffield unemployed

Milan professionals

Copenhagen 10th decile

Amsterdam 5th quintile

Rotterdam 5th quintile

USA Portland OR MSA poor

Antw erp 'poor'

USA 100 largest cities poor

USA Rochester NY MSA poor

4. Segregation levels (socio-economic) in some Dutch cities

Amsterdam Rotterdam

low incomes (1st quintile) vs rest

16.1 20.1 27.1

low income non Dutch vs rest

31.2 36.5 48.0

5th quintile vs rest

27.2 26.7 34.7

1st quintile vs 5th quintile

37.9 40.9 40.6

The Hague

Social Mix is CommonIncome distribution of the richest (zuid, left) and poorest (westerpark, right) urban districts

of Amsterdam, quintiles, 1996

1st

15%

2nd

18%

3rd

18%

4th

21%

5th

28%

1st20%

2nd31%

3rd23%

4th17%

5th9%

richest poorest

Social Mix is CommonIncome distribution of the poorest

neighbourhoods in the three largest Dutch cities, 2000

0

10

20

30

40

50

60

70

AmsterdamKolenkit

RotterdamSpangen

The HagueSchilderswijk

%

1st quintile

middle

5th quintile

Source: Pinkster 2006

In short:

• Ethnic segregation levels are moderate and generally not increasing

• Ethnic concentrations are still limited (except in UK and B)• Ethnic concentrations are dynamic, due to housing careers• Segregation levels of lowest income categories are moderate• Segregation of low and high social strata is relatively high,

but segregation between low and the middle is low• Social mix is common already

5 Segregation and neighbourhood

effects• Moderate segregation: few effects?

• Small-scale concentrations: few effects?

• Even the poorest areas are mixed: few effects?

• Some experiences in The Netherlands and Sweden: seven large-scale neighbourhood effect studies

a. Longitudinal studies in The Netherlands: Does Neighbourhood Matter?

• Impact of social composition of 500 x 500 m environments on individual’s social mobility (2 mln. cases; 1989-1994; tax income data).

Musterd, S., W. Ostendorf & S. de Vos (2003) Environmental Effects and Social Mobility. Housing Studies. Vol. 18 6. pp. 877-892.

Findings

• There are weak effects of social compositions on social mobility for people without a job

• There are fairly strong effects for people with a stronger position

Neighbourhood effects on ‘socially weaker’ and ‘socially stronger’ individuals in The Netherlands; percentages relative to households not belonging to pensioners.

0

10

20

30

40

50

60

70

0 – 2

2 – 4

4 – 6

6 – 8

8 – 10

10 – 12

12 – 14

14 – 16

16 – 20

20 – 30

30 – 40

% on benefits in the environment 1989

%benefits in 1989 and1994

paid job in 1989, benefitsin 1994

b. Longitudinal studies in Sweden: Does Neighbourhood Matter?

• Impact of social composition of 500 x 500m environments on individual’s employment careers (5.5 mln. cases; 1991-1999; GeoSweden; 16-65 year old).

Musterd, S. & R. Andersson (2006) Employment, Social Mobility and Neighbourhood Effects. International Journal of Urban and Regional Research 30 (1), pp. 120-140.

Findings

• Neighbourhood effects exist also after controlling for a range of variables

• Those who were able to improve their educational level during recession were not affected by the environment

Percentage of unemployed in 1991 staying unemployed in 1995 and 1999, per environment type 1991, per educational attainment category 1991-1995 and both years (1991, 1995) living in one of the three big cities in Sweden

0

20

40

60

80

0-2 2-4 4-6 6-8 8-10 10-12 12-14 14-16

% unemployed in the environment 1991

%

low stable (<=10yrs)

medium stable (11-12yrs)

medium high stable(13-14 yrs)

high (15+)

upward

c. Longitudinal studies in Sweden: Does Neighbourhood Matter?

• Impact of social and physical composition of 9,200 SAMS environments on individual’s employment careers (5.5 mln. cases; 1991-1999; GeoSweden; 16-65 year old).

• Focus on housing mix, social mix and social opportunities.

Musterd, S. & R. Andersson (2005) Housing Mix, Social Mix and Social Opportunities. Urban Affairs Review, Vol. 40, No. 6, pp. 761-790.

Key-concepts

• Housing mix: from absolutely homogeneous to highly heterogeneous (mixed) (9 types, entropy measures)

• Social mix: clusters on the basis of scores in classes of income deciles (low, mixed-low, mixed, mixed-high, high)

• Ethnic mix (based on nationalities and share of refugees)

• Socio-ethnic clusters (all combined)• Social mobility: change in employment position

Findings on housing mix and social mix/ethnic mix

• Housing mix and social mix association is not very strong

• Same holds for housing mix and ethnic mix

• ~25% of homogeneous housing areas are relatively homogeneous low income areas

• ~20% of the most heterogeneous housing areas are homogeneous low income areas

Findings regarding impact of mix on social opportunity (also next slide)

• There is limited difference in opportunities of low educated in homogenous low social status areas and mixed low and highly mixed areas.

• In these three types of area the lowest share of people that stays employed is found in both physically homogeneous and heterogeneous areas

• A shift to mixed high and homogenous high areas would help, but is difficult to realise

• There are clear effects of education and of country of origin of self and parents

Perc. individual staying employed in 91,95,99 in various social and housing environments

per educational attainment level 91-95

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

ho

mo

ge

ne

ou

s

he

tero

ge

ne

ou

s

ho

mo

ge

ne

ou

s

he

tero

ge

ne

ou

s

ho

mo

ge

ne

ou

s

he

tero

ge

ne

ou

s

ho

mo

ge

ne

ou

s

he

tero

ge

ne

ou

s

ho

mo

ge

ne

ou

s

he

tero

ge

ne

ou

s

all low stable medium medium high high

Mixed lowHighly mixed Mixed high Homogeneous low Homogeneous high

social

physical

education

0.0

20.0

40.0

60.0

80.0

100.0

low

sta

ble

med

ium

1

med

ium

2

high

(15+

)

low

sta

ble

med

ium

1

med

ium

2

high

(15+

)

low

sta

ble

med

ium

1

med

ium

2

high

(15+

)

low

sta

ble

med

ium

1

med

ium

2

high

(15+

)

entirely swedish swedish one parentforeign

swedish bothparents foreign

not swedish bothparents foreign

% e

mpl

oyed

91,

95,

99

Perc. individuals staying employed in 91,95,99 living in a poor refugee area per country of origin, per

educational attainment level 91-95

d. Longitudinal studies in Sweden: What Mix Matters?

• Neighbourhood incomes (lowest and highest 3 income deciles; and overall diversity, via entropy measure)

• Educational level (share of low and share of high educated and diversity based on 7 categories)

• Ethnic composition (similarly)• Housing tenure structure (similarly)

Andersson, R., Musterd, S., Galster, G. and Kauppinen, T. (2007) “What Mix Matters?”. Exploring the relationships between individual’s incomes and different measures of their neighbourhood contexts. Housing Studies 22 (5), pp. 637-660.

FindingsThe share of adult males with earnings in the lowest 3 income deciles in 1995 holds greatest explanatory power for the later income earned, after controlling for:

• personal characteristics that can vary over time (e.g. marital or fertility status, educational attainment)

• personal characteristics that do not vary after 1995 (e.g., year and country of birth, experiences prior to 1995)

• municipality of residence in 1995 • characteristics of local labour market(s) in which individual resides

in 1995 and 1999 (e.g., mean earnings)

e. Longitudinal studies in Sweden: Are ethnic enclaves good or bad?

• Multiple measures of immigrant environments• For 1995-2002, residing in one of the three big Swedish

metropolitan areas in at least one of the years 1995, 1999, 2002

• Seven immigrant ethnic groups

Musterd, S., R. Andersson, G. Galster & T. Kauppinen (2008) Are Immigrant’s Earnings Influenced by the Characteristics of their Neighbours? Environment and Planning A, pp. 785-805.

Findings

• Own group ethnic concentrations can initially pay dividends for immigrants, but these benefits turn into disadvantages over time, after approx. two years

• The impact of other immigrants is positive only if unemployment levels are very low

f. Longitudinal studies in Sweden: Does neighbourhood income mix affect earnings of adults?

• Controls for omitted variable bias

• Controls for selection bias

• Differences equations 1991-1995 and 1996-1999

Galster, G., R. Andersson, S. Musterd and T. Kauppinen (2008) Does neighbourhood income mix affect earnings of adults? Journal of Urban Economics 63, pp. 858-870

Findings

• Males not employed full time benefit from middle-income neighbours and not from either high- or low-income neighbours.

• Full-time employed males benefit from high-income neighbours

• Even in comprehensive welfare states role models and interpersonal networks shape economic opportunities

g. Longitudinal studies in Sweden: What Scale Matters?• Different spatial scales were compared• 100m x 100m• SAMS• Municipality• Metropolitan region• Multi-level modelling for 1995-2002

Andersson, R. & S. Musterd (fc) What Scale Matters? problematizing scale in neighbourhood effect studies. Under review.

Findings

• Large impacts of environments on earnings at smallest scale (100x100m)

• Low income environments have largest impacts

• Share of unemployed in the neighbourhood has most impact at slightly higher level (SAMS)

Conclusions• Segregation is a strong process driven by objectives to

translate social inequality and lifestyle differences into spatial inequality

• Segregation is influenced by global, national, local and group level processes

• Moderate effects for weakest households in the Netherlands, but stronger effects for stronger households

• Clear effects of neighbourhood compositions in Swedish contexts

• Many ‘problematic neighbourhoods’ are mixed already• Housing mix and social mix are not 1-1 related

6 Implications for Urban Policy • Social mix (if carefully targeted) may contribute

to personal economic success

• But may be difficult to obtain.– Spatial dispersal: discrimination– Forced mix in meritocratic societies is difficult;

contra residential choice and lifestyle homogeneity preferences

– Mixed housing policies may be counter-productive when stronger households move away

Other strategies to consider

• Combating stigmatisation through direct interventions in stigmatised areas

• Improve integration by rising the level of education of all individual residents (not via ABI’s)

• Improve integration by assisting all unemployed residents in getting a full-time job (not via ABI’s)

Finally• Judgment of integration requires process studies

and detailed analysis of the type of association (linear, non-linear relations; thresholds; multilevel)

Sako Musterd

Department of Geography, Planning and International Development StudiesUniversity of Amsterdam