population composition of neighbourhoods

Who wants to leave the
neighbourhood?
The effect on moving wishes of
being different from the
neighbourhood population
Maarten van Ham & Peteke Feijten
CHR – Centre for Housing Research, University of St Andrews
Residential mobility studies
• Residential mobility is mainly explained from
individual and household characteristics +
characteristics of the dwelling:
– age, household composition
– educational level, income
– dissatisfaction with the dwelling
• Indications that also the neighbourhood plays
a role in the decision to move… But the
neighbourhood gets little attention in the
literature.
CHR – Centre for Housing Research, University of St Andrews
Residential mobility and
neighbourhoods
• About 10% of the Netherlands population moves
every year.
• 75% of the movers leave their neighbourhood
(postcode area).
• Flows between neighbourhoods are selective (for
example high income groups leaving certain
neighbourhoods).
• Insight in the role of the neighbourhood in residential
mobility is important to understand the mechanisms
behind segregation and neighbourhood
deterioration.
CHR – Centre for Housing Research, University of St Andrews
Residential mobility on postcode level
25 km
A
U
TH
R
CHR – Centre for Housing Research, University of St Andrews
The role of the neighbourhood in
residential mobility behaviour
• The neighbourhood can play a role in individual
moving decisions.
– physical characteristics: amount of green, noise, air
pollution, building density.
– social characteristics (population composition): ethnic
composition, socio-economic level, age structure.
• If people are unsatisfied about the neighbourhood
or the neighbourhood population, this may trigger
their desire to leave the neighbourhood.
• In our paper we focus on the role of the
population composition of neighbourhoods in
people’s decision to leave the neighbourhood.
CHR – Centre for Housing Research, University of St Andrews
CHR – Centre for Housing Research, University of St Andrews
Schelling’s segregation
hypothesis (1971)
• Behaviour at micro level:
– People prefer to live among others who are like
themselves (income, religion, ethnic background).
– Therefore: people who belong to a minority in a
neighbourhood, will be more likely to have a wish to
leave the neighbourhood than those belonging to a
majority.
• Effect at macro level:
– Increasing (spontaneous) segregation
• Schelling’s model is purely theoretical; first
empirical test by Clark (1991) for US.
CHR – Centre for Housing Research, University of St Andrews
Research questions
To what extent and how does the population
composition of the neighbourhood influence
individuals’ wish to leave their neighbourhood?
Central hypotheses:
1. People are more likely to have the wish to leave their
neighbourhood with an increase of the...
o
o
o
percentage of low income households
percentage of immigrants
percentage of rented dwellings
2. This mechanism is less strong (or is even nullified) for
persons who have such a characteristic themselves
(Schelling).
CHR – Centre for Housing Research, University of St Andrews
Data
• Housing Demand Survey 2002 (WBO)
• Sample: 62,144 respondents, 18-80 years old, living
independently.
• Information on…
–
–
–
–
Individual background characteristics
Residential situation and location (4-digit postcode)
Satisfaction with dwelling
Wish to leave the neighbourhood
• Enriched with neighbourhood characteristics (physical and
social) at postcode level
– % low income households, % immigrants, % rented dwellings
– degree of urbanization (based on address density)
CHR – Centre for Housing Research, University of St Andrews
Variables
• Dependent:
– Wish to leave the neighbourhood (no wish=0; wish=1).
– 18,7% of the sample has the wish to leave the
neighbourhood.
• Independent:
–
–
–
–
Individual and household characteristics
Dwelling characteristics
Neighbourhood characteristics
Interaction terms between individual characteristics
and neighbourhood characteristics
CHR – Centre for Housing Research, University of St Andrews
Method
• Because some characteristics are on individual
level and others are on neighbourhood level, we
have a multilevel structure in the data.
• Logistic multilevel regression with cross-level
interactions
– level 1 = respondent
– level 2 = neighbourhood
• In formula:
f(πij) = β0 + β1xij + β2zj + β3xijzj + eij + u0j
CHR – Centre for Housing Research, University of St Andrews
Model 3
Model 4
Multilevel
logistic
regression
of
the
wish
to leave
Neighbourhood characteristics
Urbanization (ref = not urbanized)
the neighbourhood
(N = 62,144)
Weakly urbanized
0.043
0.049
0.030 0.048
Urbanized
0.169 Model
0.048
***
0.137 Model
0.048
3
4 ***
Strongly urbanized
0.204
0.049 ***
0.160 0.049 ***
Neighbourhood
characteristics
Very strongly(ref
urbanized
-0.017
0.062
-0.047 0.061
Urbanization
= not urbanized)
Percentage
of low income HH
-0.003
0.002
-0.002
Weakly urbanized
0.043
0.049
0.030 0.002
0.048
Percentage
0.016
0.002
0.022 0.002
Urbanized of ethnic minorities
0.169
0.048 ***
0.137
0.048 ***
Percentage
of rented dwellings
0.000
0.001
0.002 0.001
Strongly urbanized
0.204
0.049 ***
0.160
0.049 **
***
% Surinamese
Very
strongly urbanized
-0.017
0.062
-0.047 0.061
% Antilleans
Percentage
of low income HH
-0.003
0.002
-0.002 0.002
% Turks of ethnic minorities
Percentage
0.016
0.002 ***
0.022 0.002 ***
% Moroccans
Percentage
of rented dwellings
0.000
0.001
0.002 0.001 **
Non-Western ethnic minorities
% Surinamese
Cross-level
interaction terms
% Antilleans
Low
income * % low-income
-0.002 0.001 **
% Turks
households
in the neighbourhood
% Moroccans
Belonging
to ethnic
minority
*%
-0.014 0.002 ***
% Non-Western
ethnic
minorities
ethnic minorities
in the terms
neighb.h
Cross-level
interaction
Renter
* % rented
dwellings in the
-0.004
Low
income
* % low-income
-0.002 0.001 ***
**
neighbourhood
households
in the neighbourhood
Belonging to ethnic minority * %
-0.014 0.002 ***
ethnic
CHRminorities
– Centrein the
forneighb.h
Housing Research, University of St Andrews
Renter * % rented dwellings in the
-0.004 0.001 ***
Estimated probability to have a moving wish by
ethnic background; by percentage of people
from ethnic minorities in the neighbourhood
0.60
0.50
0.40
not ethnic minority
ethnic minority
0.30
0.20
0.10
0
10
20
30
40
50
60
70
80
90
100
CHR – Centre for Housing Research, University of St Andrews
Estimated probability to have a moving wish for
renters and homeowners; by percentage of
rental dwellings in the neighbourhood
0.30
0.25
homeowner
0.20
renter
0.15
0.10
0
10
20
30
40
50
60
70
80
90
100
CHR – Centre for Housing Research, University of St Andrews
Conclusion
• Main effect hypothesis is partly confirmed:
– People are more likely to have a wish to move when there are
more rented dwellings and more immigrants in the
neighbourhood.
• Interaction-effect hypothesis fully confirmed:
– Those with a low income, immigrants, and people in a rented
dwelling are less likely to have the wish to leave when there are
more low income households, immigrants, and rented dwellings in
the neighbourhood.
• The effects are relatively small...... BUT they suggest that
segregation comes about partly through individual
preferences regarding population composition in the
neighbourhood.
CHR – Centre for Housing Research, University of St Andrews
Discussion
• Why do people want to leave concentration neighbourhoods?
– Do they not feel at ease among people who are different
from themselves? Or do they disapprove of people with
different characteristics than their own?
– Or is it that the presence of (mainly) ethnic minorities
correlates with other elements in the neighbourhood that
are not measured in our data?
• Results cast doubt on the success of mixed tenure policies
and other policies aimed at mixing populations at the
neighbourhood level…. People are free to move!
• Present Netherlands housing market is tight and people do
not have many options. But as the housing market relaxes,
people will have more choice in where to live and this will
probably result in more segregation.
CHR – Centre for Housing Research, University of St Andrews