Segregation article new version March20 Roger without corrections

advertisement
Housing policy and Ethnic Spatial Segregation: Comparing Capital
Cities in Four Nordic Welfare States
Hans Skifter Andersen, Roger Andersson, Terje Wessel and Katja Vilkama
(Preliminary paper; Not to be cited)
Abstract
Spatial segregation is related to housing markets and housing policies. In this paper, ethnic
segregation is compared across four Nordic capitals and explanations for the differences are
examined by comparing housing markets and housing policies in the countries. The housing
markets in Denmark, Finland, Norway and Sweden are ethnically segmented with high
concentrations of immigrants in some tenures (especially social/public housing) and low
concentrations in others. The paper shows that the spatial location of immigrants is strongly
connected to the tenure composition in neighbourhoods. Ethnic segmentation of housing tenures
thus contributes to segregation. This effect is however much dependent on how tenures are
distributed across space. Neighbourhood tenure mix policy has in one of the cities resulted in a
relatively low segregation in spite of high concentrations of immigrants in social/public housing.
Introduction
Spatial segregation is connected to housing markets. Segregation affects housing markets
because the market reacts to the spatial distribution of housing demand from different social and
ethnic groups. In neighbourhoods attracting high income groups the market will tend to establish a
supply of large size, high quality housing with owner-occupation. In neighbourhoods having
concentrations of households with lower incomes the supply will tend to be smaller dwellings of
lower quality in rented housing.
The composition of residents and housing in neighbourhoods is, however, a result of historical
developments where the dominant demographic and social groups tend to change over time (van
Kempen and Murie 2009). Especially in some of the central parts of the cities demand has shifted
over time, quite often from high income groups to lower income groups, and later in some
neighbourhoods, due to gentrification the other way around. Compositional shifts have occurred
also in post-war large housing estates on the urban periphery. In many countries such areas
originally housed predominantly native-born working class and lower middle class households
which later on have been substituted by increasing numbers of unemployed residents with
immigrant background. The social composition of urban neighbourhoods has thus often changed
over time while only minor changes have occurred in the housing stock and in housing tenure
composition.
In most countries the location of different kinds of housing is not a simple product of market forces.
Housing markets can be highly regulated and national policies often favour one particular tenure
form over other forms. It is sometimes also the case, not least in the Nordic countries, that local
authorities have the power to influence land use and housing developments in a way that are
subject to political objectives (Atkinson and Kintrea 2000, Galster 2007, Briggs 2005, Musterd and
Andersson 2005).
1
As the housing stock in a neighbourhood has a major influence on the kind of people moving to the
area, the inertia in changes in the housing stock and the intervening of local authorities results in
that the connection between segregation and housing markets also works the other way round.
The actual settlement of different groups is influenced by the actual composition of housing in
neighbourhoods and therefore actual segregation is also affected by the distribution of different
kinds of housing in space.
The research question in the paper is: Can a comparison of four countries tell us something about
the connection between housing policy, housing markets and the spatial distribution of immigrants?
We will examine how housing policies in the countries affect the spatial workings of the housing
market and the segregation of immigrants from natives via their distribution on housing tenures and
the extent to which housing tenures has been separated in space. The hypothesis is that a
stronger ethnic segmentation of the housing market in a country combined with higher spatial
segregation of housing tenures will lead to increased ethnic segregation. If immigrants are
concentrated in a certain housing tenure form, e.g. social/public housing, and housing with this
tenure is located in large isolated housing estates, we expect more pronounced ethnic residential
segregation.
The paper focuses on four Nordic capital regions: Copenhagen, Helsinki, Oslo and Stockholm. We
have compiled comparable sets of neighbourhood data based on population and housing statistics,
and employ both descriptive and multivariate statistics to estimate the relation between housing
segmentation and ethnic residential segregation.
In the next section we provide a short literature review on causes of ethnic spatial segregation and
the importance of housing policy. This is followed by a description of data and methods. The
differences between the cities concerning immigrant populations are described followed by a
comparison of measures of ethnic spatial segregation in the city regions and the spatial distribution
of immigrants. The analyses of the connection between this and housing policy commence with a
description of the differences in housing policies across the countries and a comparison of the
ethnic segmentation of respective housing market. This is followed by a comparison of the spatial
structure of housing markets in the regions: to what extent are tenures spatially mixed or separated
in space? We thereafter estimate the connection between tenure segmentation and segregation
using statistical analyses. In the conclusion we discuss to what extent differences in ethnic
segregation can be attributed to differences in the spatial structure of the housing markets and in
housing policies.
Causes of ethnic spatial segregation
In the European literature, there is a general consensus verifying the disadvantage faced by
immigrants and minority groups in the housing market (Fonseca et al 2010). The literature on
segregation and housing market positions of ethnic minorities in Western European cities has
shown that minorities typically are confined to the least desirable private or social/public rented
housing in the inner city or on peripheral housing estates (Hamnett and Butler 2008).
The extent to which the housing market affects ethnic spatial segregation in a country depends on
many things but not least the housing policy performed. Based on a comparison of welfare
systems, housing policies and ethnic segregation in cities in eight European countries Arbaci
(2007) concludes that differences in housing policy and the resulting housing markets in the
2
countries, together with planning systems, play a very important role in explaining differences in
ethnic segregation across countries. An earlier Danish study of segregation (Skifter Andersen,
Andersen and Ærø 2000) showed that more than half of the social segregation between different
parts of Copenhagen could be explained by differences in the tenure composition. However, as
pointed out by Andersson, Bråmå and Johansson (2007) and Andersson, Bråmå and Hogdal
(2009), using data on Stockholm and Gothenburg, there is also a distinct ethnic residential
segregation within each tenure form so there is far from a one-to-one relationship between housing
segmentation and segregation processes.
In many countries in Northern Europe, immigration has grown in recent years and there has been a
general tendency among immigrant families to –volunteerily or because of constraints– settle in
certain parts of the housing market and in limited parts of cities. Often they have settled in
social/public housing. In this way some neighbourhoods in the cities have obtained a large
proportion of ethnic minorities and have been transformed into what sometimes is called ‘multiethnic neighbourhoods’, in which the native-born majority population has become a minority.
Previous studies have found marked differences between different ethnic groups concerning the
extent to which they live in neighbourhoods with a high concentration of immigrants (Johnston et
al. 2002, Finney 2002, Fong and Chan 2010).
Some studies explain the housing situation of ethnic minorities primarily by their lack of resources.
Not only economic resources but also cognitive, political and social resources are thought to be
important (Van Kempen 2003). It is particularly these non-economic resources that newly arrived
ethnic minorities often lack. In parts of the housing market, good contacts to persons or institutions
are decisive for access to dwellings. This especially concerns private landlords. It is important to
have relevant knowledge of the possibilities and rules on the housing market, which also often
requires good language skills or good access to advisers. Besides the disadvantage of lower
incomes, immigrants can thus have special difficulties in accessing parts of the housing market. If
the housing market is difficult to understand, it is likely to be more difficult for immigrants with a
limited knowledge of the host society to act on the market and find good solutions to their housing
needs (Søholt 2007, Søholt and Astrup 2009a).
Other studies (Aalbers 2002, Andersson 1998, Skifter Andersen 2008, Søholt and Astrup 2009a,
Molina 2010) point to discriminatory practices in the housing market. Especially social/public and
private landlords might exclude ethnic minorities from housing in some areas while sometimes
steering them to other, less attractive areas (Swedish Equality Ombudsman 2008) . The extent to
which discrimination occurs can depend on the way housing tenures are, among others, regulated
and supported through housing policy. If access to housing is very dependent on decisions made
by administrators of housing and subject to local execution of power, there is greater scope for
discrimination than if there are strict rules for how to allocate vacant dwellings. Moreover, it is
important to what extent the housing in question is subject to a strong excess demand. If there is a
supply deficit, there will likely be queues, which in turn will generate more fertile conditions for
discrimination. Another consequence of discrimination in the private rental market is the fact that
immigrants have to pay more than natives, if there is no rent regulation (Røed Larsen and
Sommervoll 2011). There could also be discriminatory practices among banks or institutions
providing capital for the purchase of housing if, as a result of prejudice and the tendency to see
ethnic minorities as less solvent customers. Discrimination against immigrants by financial
3
institutions can be dependent on the extent of public subsidies for housing and can be reduced by
public loan guarantees.
In earlier American research the spatial location of immigrants has been analyzed in relation to
their social, cultural and economic integration in the host society. According to the American
‘ecological’ theoretical tradition of the Chicago School (Park 1925), the spatial distribution of
groups is a reflection of their human capital and the state of their social and economic integration,
called ‘assimilation’ (Alba and Nee 1997). A basic assumption in this theoretical tradition is that
especially new immigrants for different reasons prefer to settle in neighbourhoods dominated by
their own ethnic group, sometimes called ethnic enclaves. As stated by Massey (1985:page):
‘Some degree of geographic concentration is an inevitable by-product of immigration, which is
guided by social networks and leads to settlement patterns determined partly by the need of new
immigrants unfamiliar with American society and frequently lacking proficiency in English for
assistance from kin and co-ethnics’. Others are speaking about ‘chain migration processes’
(MacDonald and MacDonald 1964; Johnston et al 2002) where earlier immigrants draw a number
of newer immigrants to the host country and particular neighbourhoods through kinship and other
social networks.
Preferences for living in neighbourhoods with countrymen could be important for what tenures and
dwellings immigrants try to get and what dwellings they can get access to. In different countries,
multi-ethnic neighbourhoods, where immigrants find many countrymen, have been established in
tenures depending on how easy it has been for immigrants to get access to these tenures. In some
countries it has taken place in private rented housing, in others in social/public housing and yet
others in owner-occupied housing. A hypothesis could be formulated that neighbourhoods with less
attractive housing dominated by an easy-to-access tenure provide the basis for an initial influx of
immigrants (Scaffer and Huang 1975, Bleiklie 1997, Søholt 2007, Søholt and Astrup 2009a).
When the presence of ethnic groups become very visible in the city, self-perpetuating segregation
processes called ‘White flight’ and ‘White avoidance’ begin to appear. In the U.S., it has been
observed that Whites ‘flee’ when the number of Black residents in their neighbourhood exceeds a
certain proportion of the population (Ellen 2000; Wright et al. 2005). In recent years, there has
been a tendency to put more weight on ‘White avoidance’, meaning that natives tend to avoid
moving to neighbourhoods with many immigrants or concentration of particular ethnic groups
(Clark, 1992; Quillian, 2002; Bråmå 2006; Bråmå and Andersson 2010, Andersson 2013). As a
consequence of these flight and avoidance processes, it is easier for immigrants to get access to
these neighbourhoods, which in turn are often dominated by a particular tenure form.
As a result of the above mentioned factors, and the segregation processes that follow, especially
more recently arrived or economically less resourceful immigrants tend to be over represented in
certain tenures and in less attractive or low-quality housing often concentrated in certain
neighbourhoods. The segregation outcome in each case can of course vary, not least because the
context is influenced by the structure of the housing market and the national and local housing
policies that shape it.
The importance of housing policies and housing markets for segregation
The functioning of the housing markets is the outcome of policies influencing housing
conditions but also a range of policies that impact on the distribution of welfare and
4
economic resources across households in a given city. The important issue addressed in
this paper is whether housing policy has a different impact on the housing and location of
the immigrant population compared to that of the majority and whether such differences
appear to be different across the Nordic region.
Housing policy can be defined as public initiatives that affect the supply, price and quality
of dwellings but also how dwellings are distributed across different types of households.
Housing policy is often intertwined with urban policy and land use planning, which
influences where and how dwellings are located and the (physical) quality of
neighbourhoods in different parts of a city.
Immigrants’ housing options are of course very dependent on the general options on the
housing market and on the socio-economic position of different immigrant categories. In
the Nordic context much of the debate on ethnic segregation has focused on immigrants
(often refugees) originating in thirld world countries, many of whom show low employment
rates and low income. In such a case one may hypothesize that the general housing
market for low-income groups and the ethnic segmentation of the market much depend on
the degree of income segmentation. Furthermore, this segmentation depends on to what
extent housing policy creates even or uneven opportunities and economic incentives in
different tenures. If housing subsidies or tax incentives are designed in such a way that
high-income groups receive the largest support in owner-occupation and only low-income
groups are supported in rental housing, it can be expected that there will be pronounced
income segmentation. And if this kind of housing also is concentrated in certain parts of
the cities social and ethnic segregation will be high. On the other hand, if housing policy is
more universalistic in the sense that it to a greater extent is aimed at housing for the
whole population and not only for vulnerable low-income groups, income segmentation will
probably be lower.
In the comparative literature about housing policy a division is made between ‘dual’ and
‘unitary’ housing systems (Allen 2004, Balchin 1996, Kemeny 1995). In dual systems the
housing market is much socially segmented between rented housing and homeownership,
and respective tenure form is dominated by specific social groups while other groups have
difficulties getting access to the tenure. Another division is made between unitary and dual
rental systems (Kemeny 1995). In the unitary rental systems social/public supported
housing is competing on even terms with private renting, while in dual systems public
housing is a restricted sector reserved for low income groups. It is often assumed (Arbaci
2007) that the Nordic countries all have unitary housing systems, which –compared to
more liberal welfare systems– means that social segmentation of the markets is expected
to be relatively low. As is shown below this assumption is not always true
Some other general features of housing policies have been argued to be important for
immigrants (Skifter Andersen, Magnusson Turner and Søholt 2013). One such feature is
whether housing policy makes the market more, or less, transparent. Complicated systems
of economic support and of rules for access to housing make it more difficult for especially
more recently arrived immigrants, who have limited knowledge of the housing systems and
might therefore be more constrained in searching for housing options. A related issue is if
conditions for discrimination exist. If rules for access to rental dwellings are unclear, or if
allocation is entrusted to landlords alone, the risk of discrimination is greater. In more
5
market based rental housing systems, where there is a sufficient supply and landlords
compete to attract tenants, risks of discrimination can be hypothesized to be lower.
Housing policies in the Nordic countries
As stated above, housing policies and housing markets vary much between Nordic countries in
spite of their common background as universal welfare states. In a comparison of housing policy
instruments in the four countries (Lujanen (eds.) 2004), it was stated that 'surprisingly big
differences' was found in implementation and in means used. Similarly, another comparative study
of housing policy in four Nordic countries (Bengtsson et al. 2006, p 12) concludes that the way
housing policy has been implemented in the different countries shows so important differences that
one can talk about quite different housing systems; differences in both policies institutions and
markets. The Danish and Swedish housing policies are characterised by being more general and
universalistic directed towards all groups in society. In Norway and Finland policies are to a greater
extent directed towards special measures for low-income groups regulated by means tests.
Sweden has traditionally been the country with the most unitary system, because it has put most
weight on housing as a social good with even housing possibilities for all in all tenures (Turner and
Whitehead 2002, Magnusson Turner 2010). Denmark also had strong social objectives for housing
but not as pronounced as in Sweden because more weight has been put on the market and less
on state control. Greater tax advantages, which have been more valuable for high-income groups,
have tended to create a more dual market than in Sweden. Policies in Norway have been more
selective for a long time with extensive needs tests and a restricted and residual public housing
sector, creating a more segmented housing market and a less unitary housing system. In Finland,
housing policy to a greater extent than in the other three countries has been seen as social policy
for the weaker groups in society where social/public housing until recently has been reserved for
low-income groups. The rental markets in Norway and Finland can thus be characterised as falling
into the category ‘dual rental systems’.
Table 1. A rough characteristic of the housing systems.
Rental markets
Housing system
More Unitary
More Unitary
Sweden
More Dual
Finland
More Dual
Denmark
Norway
Note: For immigrants the Danish rental market is more dual.
Table 1 roughly illustrates the differences between the housing systems in the four countries.
Sweden and Finland have a better balance between renting and ownership and thus have more
unitary housing systems, but the rental market in Finland is more segmented. In Norway and
Denmark there is a stronger separation between renting and owing and thus represent more dual
systems. But the rental sector in Denmark is more unitary while it is more segmented in Norway.
However, both in Sweden and Norway the large co-operative sectors form a bridge between
renting and owing, which to some extent make the housing systems more unitary.
Table 2 displays the segmentation between home ownership and the rented/co-operative sectors
by comparing the proportion of homeowners in different income quartiles. This is used to calculate
an index of income segmentation (see bottom row). The table shows to what extent households in
different income groups are separated between owning and renting.
6
Table 2. Homeownership among households in different income quartiles 2006.
Denmark
Finland
Norway
Sweden
Household income quartile
Homeownership, %
1
50
53
42
51
2
52
65
54
60
3
66
71
74
62
4
82
74
88
71
All
62
67
62
59
Index for income segmentation
29
18
33
14
Source: First European Quality of Life Survey: Social dimensions of housing.
http://www.eurofound.europa.eu/pubdocs/2005/94/en/1/ef0594en.pdf and Norway: Levekårsundersøkelsen 2007 and Aarland and
Norvik 2008.
Index of segmentation = ½*sum(abs(proportion of homeowners in quartile - proportion of homeowners for the whole
population))/100
It should be stressed that data in the table are subject to some uncertainties. The Norwegian data
is from another source than those from the three other countries. The data from the EU report is
based on a survey with a limited number of respondents and the average homeownership rate
does not correspond fully to what is known from national statistics. Moreover, it is unclear if cooperatives are included as homeownership, but judging from the average homeownership rates
they are not.
The table indicates that a lower proportion of households in the lowest household income quartile
in Norway are homeowners compared with the other countries. In the highest income quartile, the
proportion is higher. The calculated index of segmentation is much higher in Norway than in the
other three countries. The housing market in Denmark is not quite as income segmented as the
Norwegian one, but, as expected, more segmented than in Sweden and Finland. The figures from
Sweden indicate that the Swedish political goal of even opportunities in different tenures has to
some extent been realized. It should however be pointed out that geographical differences are big
and that the situation in major urban areas differs from the national average estimates displayed in
table 2.
One of the key elements in housing policy, the provision of a social/public housing sector, has
been carried out somewhat differently in the countries. Earlier, Sweden had the largest
social/public housing sector, but conversion into cooperatives –in particular in Stockholm– has
reduced the sector to 14 per cent of the housing stock. Finland has about the same amount of
social/public housing, while Denmark now has the largest sector (21 per cent). In Norway, the
social/public housing sector is very small, only about five per cent of the stock.
Immigrants’ access to social/public housing is determined by the general supply and demand
balance for housing but also by rules for the allocation of vacant dwellings and the extent to which
local authorities can influence or determine allocation policies. In all countries, local authorities do
7
have an obligation to provide housing for new refugees, which most often results in settlement in
social/public housing.
In Denmark and Sweden, access to social/public housing is in principle based on an open
allocation system with waiting lists, which either are controlled by a municipal body or operated by
each housing company or association. In recent years, however, access to social/public housing in
Denmark has become somewhat more difficult for immigrants due to new rules of allocation that
have been introduced in estates with many immigrants in order to avoid further concentration of
minorities. In Sweden we find a range of different local solutions. While for instance Stockholm city
operates a waiting list where both private and public housing companies in many of the region’s
municipalities have agreed to allocate 50 percent of all vacant dwellings to households on the list,
other cities do not organize a joint waiting list so individual households need to seek housing via
each separate rental housing company. It has been found that ethnic discrimination is a problem
and transparency –i.e. who gets which dwelling and why– is often lacking (Solid, Andersson,
Molina 2007). In Finland like in Norway, access is based on urgent housing needs, which favours
the most vulnerable immigrants, but not the more successful.
Instead of social/public housing, Norway has historically focused on cooperative housing as
housing for middle and low-income groups. The main socio-political means are to supply lowincome households with housing allowances in order to be able to buy and keep a dwelling,
regardless of kind of ownership. The cooperative sector is of very little importance in Finland and is
also only a small sector in Denmark, but has some importance in Copenhagen. Sweden has the
largest cooperative sector and its importance is growing in most larger cities.It can to some extent
be seen as a substitution for owner-occupied flats, which yet to a very limited extent exist in
Sweden, and the average incomes in the sector are above average (Andersson and Magnusson
Turner 2011, Skifter Andersen et. al 2013, Skifter Andersen 2012). The history of cooperatives in
Norway and Sweden is somewhat similar and in both countries the sector is dominated by large
housing associations. Cooperatives in Denmark are quite different. They have mostly been
converted from private renting and they have been subject to a strict price control, which has to
some extent been lifted in recent years.
Description of differences between the capitals concerning immigrant populations
As described in Andersson et al. (2010), there are some important differences in the nature of
immigration to the four Nordic countries as shown in Table 3. Finland has received much fewer
immigrants than the other countries, while the share of immigrants in Sweden is about 40 per cent
higher than in Denmark and Norway.
Table 3 Immigrants (foreign-born) in the four Nordic countries and in the capital regions 2008.
The whole countries
Denmark
Finland
Norway
Sweden
Proportion of population born outside the country, per cent
9,8
4,4
10,8
14,3
Per cent of population, who are immigrants coming from Eastern Europe
Per cent of population coming from Non-European countries
0,8
4,5
1,7
1,5
1,8
5,0
3,2
6,2
The capital regions
Population 1000 inhabitants
Copenhagen Helsinki
1,369
1,022
Oslo
1,079
Stockholm
1,849
8
Proportion of population born outside the country, per cent
Per cent of population, who are immigrants coming from Eastern Europe
Per cent of population coming from Non-European countries
Source: Andersson et. al. 2010.
11.7
8.8
1.9
6.8
3.5
3.6
14.3
??
21.3
3.5
9.3
11.1
The extent of immigration also differs across the four Nordic capital regions. In general the
proportion of immigrants is higher in the capitals compared to the average for the country,
especially in Helsinki and Stockholm.
Moreover, the ethnic composition of immigrants varies between the countries and the capitals. In
Finland the biggest Non-Western group comes from Eastern Europe, especially from Russia.
Finland has relatively few immigrants from countries outside of Europe compared with the three
other countries, in which in total five per cent of the whole population are immigrants coming from
outside Europe. Especially Stockholm has many immigrants from non-European countries.
Data and methods used
The data used in this paper comes from four databases, one in each of the countries, containing
register data on neighbourhoods in the four capitals on the population and their housing situation.
Data are available on the ethnic composition of neighbourhoods and on the tenure composition of
housing.
The choice of data has been a compromise between what is available in the different datasets. The
populations consist of all residents in the cities 20+ years. In some of the countries descendants
cannot be identified. Immigrants are thus defined as foreign-born residents, while descendants are
treated as natives.
The neighbourhood divisions are described in Table 4.
Table 4 Description of neighbourhood divisions in the capital regions 2008.
Copenhagen Helsinki
Oslo
Number of
716
257
Average size, no of residents 20+ years
1371
3000
Standard deviation
673
2730
1 Neighbourhoods with less than 70 residents excluded.
neighbourhoods1
Stockholm
249
814
3064
1831
1949
2205
The neighbourhood divisions differ somewhat as the size of them varies from about 1400 in
Copenhagen to about 3000 in Oslo. It is a well known fact that measures of segregation is
sensitive to the choice of geographical scale (fewer/larger units tend to make segregation less
pronounced) but we judge that the variation here is relatively modest.
Factors to be compared inside and across cities
The aim of the empirical analyses in the paper is to examine the connection between immigrants’
position in the housing market, the spatial structure of the housing market and ethnic residential
segregation. Figure 1 presents these factors and their anticipated relationships
9
Segregation between immigrants
and natives (index of dissimilarity)
Spatial distribution of
immigrants
Spatial distribution of
housing tenures
Level of
immigration
Ethnic segmentation of
the housing market
Figure 1. Factors explaining differences in ethnic segregation compared across cities.
The databases contain data on immigrants’ distribution on housing tenures in the cities. Based on
these data we calculate indices measuring to what extent immigrants are unevenly distributed
across housing tenures, called ‘ethnic segmentation of the housing market’. Two measures are
used:
1. Comparison of ethnic minorities’ overrepresentation in tenures:
This measure is defined as
=
(𝒑𝒆𝒓𝒄𝒆𝒏𝒕 𝒐𝒇 𝒊𝒎𝒎𝒊𝒈𝒓𝒂𝒏𝒕𝒔 𝒊𝒏 𝒕𝒆𝒏𝒖𝒓𝒆 𝒙 – 𝒑𝒆𝒓𝒄𝒆𝒏𝒕 𝒐𝒇 𝒕𝒐𝒕𝒂𝒍 𝒑𝒐𝒑𝒖𝒍𝒂𝒕𝒊𝒐𝒏 𝒊𝒏 𝒕𝒆𝒏𝒖𝒓𝒆 𝒙)
𝒑𝒆𝒓𝒄𝒆𝒏𝒕 𝒐𝒇 𝒕𝒐𝒕𝒂𝒍 𝒑𝒐𝒑𝒖𝒍𝒂𝒕𝒊𝒐𝒏 𝒊𝒏 𝒕𝒆𝒏𝒖𝒆 𝒙
× 𝟏𝟎𝟎
2. Index of ethnic tenure segmentation
= ½ * ∑𝑛
𝑥=1 𝐴𝑏𝑠𝑜𝑙𝑢𝑡𝑒 (𝑝𝑒𝑟𝑐𝑒𝑛𝑡 𝑜𝑓 𝑖𝑚𝑚𝑖𝑔𝑟𝑎𝑛𝑡𝑠 𝑖𝑛 𝑡𝑒𝑛𝑢𝑟𝑒 𝑥 − 𝑝𝑒𝑟 𝑐𝑒𝑛𝑡 𝑜𝑓 𝑤ℎ𝑜𝑙𝑒 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑖𝑛 𝑡𝑒𝑛𝑢𝑟𝑒 𝑥)
Moreover, data on the tenure composition in neighbourhoods is used to calculate indices on the
spatial distribution of tenures. For each tenure an index is calculated which indicates to what extent
the tenure is unevenly distributed across neighbourhoods. Because of limitations on the available
data it is in three of the countries calculated on the basis of the number of residents (20+ years)
living in the tenure in each neighbourhood, not on the number of dwellings. The index is calculated
as the simple index of segregation.
Index of uneven spatial distribution of tenure t:
𝑛𝑢𝑚𝑏𝑒𝑟 𝑙𝑖𝑣𝑖𝑛𝑔 𝑖𝑛 𝑡𝑒𝑛𝑢𝑟𝑒 𝑡 𝑖𝑛 𝑛𝑒𝑖𝑔ℎ𝑏𝑜𝑢𝑟ℎ𝑜𝑜𝑑 𝑥
𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑖𝑛 𝑛𝑒𝑖𝑔ℎ𝑏𝑜𝑢𝑟ℎ𝑜𝑜𝑑 𝑥
St = ½ ∗ ∑𝑛𝑥=1 𝐴𝑏𝑠𝑜𝑙𝑢𝑡𝑒 (
−
)
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑙𝑖𝑣𝑖𝑛𝑔 𝑖𝑛 𝑡𝑒𝑛𝑢𝑟𝑒 𝑡
𝑡𝑜𝑡𝑎𝑙 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛
10
To make a total index of how the composition of tenures varies between neighbourhoods an index
is constructed on how the composition varies
Total index of uneven tenure distribution = ½ ∗ ∑𝑇𝑡=1 𝑆𝑡 ∗
𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑖𝑛 𝑡𝑒𝑛𝑢𝑟𝑒 𝑡
𝑡𝑜𝑡𝑎𝑙 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛
In one city, Helsinki, the calculations are based on the number of dwellings instead of the number
of residents in each tenure. A similar calculation is made for Copenhagen to check for the
differences between resident-based and dwelling’s-based calculations.
To examine the connection between these two factors and segregation inside and across cities an
intermediate variable is used called ‘Spatial distribution of immigrants’. This index is expected to
have a more straight forward connection with the housing market variables within each city. It
shows to what extent immigrants are unevenly distributed or to what extent the proportion of
immigrants varies across neighbourhoods. The index is calculated in the same way as the socalled simple index of segregation.
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑖𝑚𝑚𝑖𝑔𝑟𝑎𝑛𝑡𝑠 𝑖𝑛 𝑛𝑒𝑖𝑔ℎ𝑏𝑜𝑢𝑟ℎ𝑜𝑜𝑑 𝑥
−
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑖𝑚𝑚𝑖𝑔𝑟𝑎𝑛𝑡𝑠
Index of uneven spatial distribution of immigrants = ½ ∗ ∑𝑛𝑥=1 𝐴𝑏𝑠𝑜𝑙𝑢𝑡𝑒 (
𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑖𝑛 𝑛𝑒𝑖𝑔ℎ𝑏𝑜𝑢𝑟ℎ𝑜𝑜𝑑 𝑥
)
𝑡𝑜𝑡𝑎𝑙 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛
Segregation is defined as spatial separation of immigrants from natives. The index used for
segregation is the index of dissimilarity, which has been used in many segregation studies. The
connection between the index of dissimilarity and the index of uneven distribution is to a certain
extent influenced by the level of immigration.
An important discussion is how the differences concerning level of immigration influence the
comparison of indices across the cities. Some have argued (Gorard and Taylor 2002) that the
index of dissimilarity is dependent on the proportion of immigrants in the cities, and it is also to be
expected that the probability for immigrants being separated from natives will increase with
increasing immigration. But it can be argued that the index is unchanged by increasing immigration
“If the newcomers follow the existing spatial distribution of those who were already settled in the
city” (Musterd 2012). More precisely it is to be assumed that every neighbourhood has grown in
size exactly with the number of immigrants living in it, or put in another way: it is assumed that
without immigration every neighbourhood would have been smaller corresponding to the number of
immigrants. This is, however, not so realistic in real cities where most neighbourhoods have been
relatively unchanged for a long period of time. Most of the growth of the cities takes place in a
limited number of neighbourhoods on the urban fringe. For these reasons it must be assumed that
segregation to some extent is dependent on the level of immigration and tend to be higher in cities
with higher immigration. This is also found when comparing the Nordic capitals as seen below.
However, the index made for uneven distribution of immigrants must also be expected to be
somewhat dependent on the level of immigration because cities and neighbourhoods change
concurrently with immigration. It must thus be expected that the index will tend to be relatively
lower in cities with higher immigration. By using both indices it is then possible to get a clearer
picture of to what extent segregation is caused by differences in housing markets.
11
Both indices we employ are sensitive to differences in neighbourhood size across cities. This
sensitiveness was tested on data from Copenhagen using two different neighbourhood divisions. In
the first division the average number of residents was about 500, in the last about 2000. The
measured indices on spatial distribution were 14 per cent higher when using the first division with
the smallest neighbourhoods.
Estimation of the connection between housing market segmentation and the spatial distribution
of immigrants
Statistical models (linear regression) of the connection between tenure composition in
neighbourhoods and the ethnic composition of them is used to estimate an ‘expected’ number of
immigrants in neighbourhoods based on the composition of tenures. If the expected number
matches the real value across all neighbourhoods, ethnic segregation would be fully explained by
the tenure composition.
Model:
The statistical analysis is based on data on each neighbourhood concerning number of immigrants
and number of residents 20+ years old living in different tenures (dwellings).
The dependent variable is the proportion of immigrants in the neighbourhood. To account for the
fact that the distribution of this variable across neighbourhoods is not quite normal, the logit value
of the variable is used. The independent variables are the proportion of residents living in different
tenures in the neighbourhood.
For each neighbourhood the model is used to calculate an expected proportion of immigrants
based on the tenure composition. Based on these expected number of immigrants, indices for the
expected effect of tenure segmentation on the spatial distribution of immigrants and on ethnic
segregation is calculated. These indices are compared to the actual measured indices. (A similar
method was used in Skifter Andersen, Andersen and Ærø 2000).
Ethnic segregation in the four capital regions
Figure 2 illustrates the distribution and concentration of immigrants in the four capitals. It is
produced by dividing the neighbourhoods into ten deciles. They are weighed by size (= number of
residents 20+ years) so the deciles have approximately the same number of residents 20+ years.
The neighbourhoods have been ordered after the proportion of immigrants (and Non-Western
immigrants, which are analyzed separately). The figure shows the average proportion of
immigrants 20+ years with increasing decile position.
12
All immigrants
Non-Western immigrants
70%
60%
Copenhagen
50%
40%
30%
70%
60%
Copenhagen
Helsinki
50%
Helsinki
Oslo
40%
Oslo
Stockholm
30%
20%
20%
10%
10%
0%
Stockholm
0%
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Neighbourhood deciles
Neighbourhood deciles
Figure 2. Proportion of immigrants (20+ years old) in neighbourhoods ordered in deciles after increasing
proportion.
For all four cities we find a skewed distribution with increasing gradients in the last deciles.
Stockholm reaches the highest concentration of immigrants in the last deciles followed by Oslo and
Copenhagen. This also applies to the Non-Western immigrants. The steep slopes at the end of the
curves indicates that self-perpetuating processes of ’white flight and avoidance’ have probably
taken place (Andersson 2013). But at the same time, Stockholm also has a considerable amount of
immigrants in the first deciles. Helsinki has the lowest proportion of immigrants and the slopes of
the curves seem a little more stable with increasing decile.
Table 5 shows indices of dissimilarity for the capitals, which measures the segregation between
immigrants and natives. The figures are compared with the average for the cities to illustrate how
they differ from each other.
Table 5. Ethnic spatial segregation of immigrants 20+ years (Index of dissimilarity)
The capital regions
Copenhagen
Helsinki
Oslo
Stockholm
Average for the cities
All immigrants
,25
,21
,27
,33
,27
From Eastern Europe
,28
,27
,22
,32
,27
From Non-western countries
,36
,27
,37
,48
,37
Index of dissimilarity
Relative deviation from average for all four cities %
All immigrants
-4
-22
2
25
From Eastern Europe
1
-1
-19
19
From Non-western countries
-2
-27
0
29
Note: The indices for Copenhagen and Stockholm can be expected to be 5-10 per cent too high because of smaller neighbourhood size.
The highest segregation between immigrants and natives is found in Stockholm, while the other
three cities are on the average of the four cities or below that. In Helsinki all immigrants but
Eastern Europeans have a lower segregation. In Oslo it is the opposite. Here, both‘all immigrants’
and Non-Western immigrants are at the same level as the Nordic capital average, while Eastern
European immigrants are less segregated. Comparing Table 5 with Table 3 a connection between
13
segregation and level of immigration seems obvious. For different reasons the cities with the
highest proportion of immigrants (especially Stockholm) also have the highest level of segregation.
Ethnic segmentation of housing markets in the cities
Ethnic segmentation on the housing market can be measured by comparing the distribution of
immigrants over tenures with the distribution of the whole population. Table 6 shows how
immigrants are distributed over tenures in the four Nordic capital cities and the table also reports
over- or under-representation in each tenure form in relation to the natives’ distribution. Moreover,
a calculated index of segmentation is shown for each city. The index is calculated as the sum of
the absolute value of the difference between the proportion of immigrants in each tenure and the
proportion of the whole population living in that tenure, weighed by the proportion of the whole
population living in the tenure.
Table 6: Immigrants (foreign born) distributed on housing tenure in the capitals compared to the total population
2008.
Copenhagen Helsinki Oslo Stockholm Copenhagen Helsinki Oslo Stockholm
Distribution (per cent)
Owner-occupied
Total population
42
57
57
34
Co-operatives
Private renting
16
16
3
16
22
15
29
17
Social/public housing
Other
25
1
22
2
3
2
17
4
Total
100
100
100 100
Distribution (per cent)
Owner-occupied
All immigrants
25
24
34
18
Co-operatives
Private renting
13
21
2
19
26
26
25
22
Social/public housing
Other
39
3
52
3
10
4
30
5
Total
100
100
100 100
Over-representation
Owner-occupied
-41
-58
-41
-45
Co-operatives
Private renting
-21
31
-35
21
18
74
-15
32
Social/public housing
58
137
206
76
Other
99
43
57
27
Index of tenure segmentation 21
34
24
20
Non-Western immigrants
19
8
18
-
27
30
14
20
14
58
16
62
25
13
24
41
1
4
4
100
100
99
100
-55
-48
-69
-65
-52
37
-59
-31
-11
131
4
182
66
306
41
141
18
33
23
42
70
30
57
32
0
Over representation= (per cent of immigrants in tenure x – per cent of whole population in tenure x)/ per cent of whole population in tenure x
Index of ethnic tenure segmentation = ½*sumx=1,n (Absolute (per cent of immigrants in tenure x – per cent of whole population in tenure x) *
proportion of population in tenure x).
Note: The data from Helsinki concerns foreign-language-speaking, not foreign-born, population.
In a recent article by Skifter Andersen, Magnusson Turner and Søholt (2013) is it shown that the
housing situation of immigrants in the four Nordic countries differs much. The authors also
conclude that the differences in immigrants’ position in the housing market cannot be explained by
the differences between the countries concerning immigration and general income inequalities
14
alone, but rather that differences in housing policies have special effects for the housing situation
of immigrants. This conclusion is arrived at by measuring ethnic segmentation of the housing
markets along with differences between immigrants and natives concerning the proportion living in
overcrowded housing.
Table 6 shows considerable differences between the capitals. Oslo has many immigrants living in
owner-occupied housing and few in public housing, which as pointed out above is a very small
sector. In Oslo immigrants have nearly the same distribution over tenures as the whole population,
except for a high overrepresentation in the small social/public housing sector. The distribution of
immigrants in Helsinki differs much from the natives and the index of tenure segmentation is high,
especially for Non-Western immigrants. This is especially due to a high overrepresentation of
immigrants in social/public housing. Also in Denmark and Sweden immigrants are overrepresented
in social/public housing, but at a lower level. The indices in these cities are still high, but smaller
than in Helsinki.
Housing and planning systems in the cities and the spatial distribution of tenures
The role of housing policy and housing market segmentation for affecting segregation depends on
the way different kinds of housing are distributed in urban space. This distribution is not a simple
result of segregating market forces but is also dependent on the politics, institutions and markets
that shape and change the urban structure.
In her comparison of ethnic segregation and its causes in European countries Arbaci (2007)
referred to earlier studies (Barlow and Duncan 1994) identifying different systems of housing
provision and land supply creating mechanisms of spatial differentiation. Three factors were
defined, called forms of land supply, forms of housing production and forms of housing promotion.
The forms of land supply are connected to who owns the land used for urban development and to
what extent it is owned by public actors. Moreover, it depends also on physical planning and other
kinds of urban policies, which regulates the use of land. To what extent is land use affected by
speculative market objectives and what is the role exercised by local democratic bodies?
Unfortunately, studies of residential segregation have rarely examined the role of land-use
regulations (Nelson et. al 2004).
The forms of housing promotion concerns who are the final owners and providers of the housing
produced. A dividing line is made between private and non-profit sectors.
Finally the forms of housing production are about the characteristics of developers and building
firms operating in the housing sector. What kind of developers that are dominating depends not
least on the form of land supply and ownership. For instance, if land is controlled by a municipality
it is more difficult for developers to make profits out of buying and using land for development and
instead they have to base the activity on profit-making in the building process alone. Moreover, it
has importance if the market is dominated by large builders and developers or by small-scale
builders. Large firms and developers tend to build larger and more uniform housing areas,
especially if they are operating in cities where the local authorities do not own the land and do not
regulate land use much.
It is therefore decisive how planning and housing systems are combined. High ethnic segregation
can thus either be a result of a combined liberal housing and planning system or of a strong
segmented housing system combined with a planning system that separates tenures in space. Low
15
segregation can be obtained if housing policies result in low ethnic segmentation of tenures or if a
strong planning system results in spatial dispersal of segmented tenures.
In her comparison of ethnic segregation and its causes in European countries Arbaci (2007)
divided the countries in four groups identified by welfare regimes. The data collected from different
sources on segregation in different cities showed systematic differences between these groups of
countries. Based on her data, she concludes that the Nordic social democratic welfare states (and
The Netherlands) have lower segmentation, but at the same time a relatively high degree of spatial
segregation (although somewhat lower than the liberal welfare states). She explains this
contradiction by the mode of building production, where the building activity in social democratic
welfare states is dominated by large scale builders and developers. She argues, that in spite of
strong planning systems and much public land ownership, the structure of the building and housing
sector with large builders, developers and owners has promoted the development of large urban
areas with homogeneous tenures and building types, which has promoted a segregation of tenures
in space. She argues that this especially has been the case in Sweden, followed by Finland,
Denmark and Norway. We share this view but would like to add that the neo-liberal planning trend
is nowadays prevalent in many cities across the region, not least in Sweden (Tasan-Kok and
Baeten 2011). The degree of municipal land ownership varies greatly across cities within each
country and it is apparent that local planning regimes tended to be more pro-active rather than reactive in relation to developers and construction firms twenty to thirty years ago.
An analysis has been made of the extent to which tenures are spatially separated in the capitals in
the countries. We can thus say something about the results of the combined planning, local policy
and building and housing sector organisation. Table 7 is calculated on data from the databases
described above. For Copenhagen, Oslo and Stockholm they are based on the number of
residents 20+ years old living in the different tenures in each neighbourhood. For each tenure we
calculate an index showing to what extent the distribution residents in the tenure form across
neighbourhoods deviates from all tenures. Moreover, a combined index is calculated as the sum of
the weighed indices per tenure.
Table 7. Measures of the spatial distribution of different housing tenures across neighbourhoods in the Nordic capitals
2008/2000.
Copenhagen Helsinki
Oslo
Stockholm
Average
Copenhagen Copenhagen
(resid. 20+)
(dwellings) (resid.20+) (resid. 20+)
for cities
(dwellings)
deviation %*)
Owner-occupied single
.56
.62
.32
.55
.51
.61
8.3
Owner-occupied flats
.47
.12
.39
.33
.45
-3.9
Co-operatives
.50
.57
.43
.36
.47
.48
-5.3
Private renting
.42
.27
.21
.45
.34
.39
-6.3
Social/public housing
.56
Combined index
.51
.33
.35
.52
.44
.54
-2.7
.23
.33
.46
.38
.50
-2.5
7
Relative deviation from average for the cities %
Owner-occupied single
10
21
-38
Owner-occupied flats
44
-63
19
Co-operatives
8
22
-8
-23
Private renting
24
-20
-38
34
Social/public housing
27
-25
-20
18
Combined index
33
-40
-14
21
*) The relative deviation between the indices measured respectively by using dwellings and using tenures for residents 20+ years
Note: The indices for Copenhagen and Stockholm can be expected to be a little too high because of smaller neighbourhood size
16
The data used for Table 7 are not fully comparable. For Helsinki they are based on the number of
dwellings in neighbourhoods while we use population data (residents aged 20+ years) for the other
three cities. A comparison of indices for Copenhagen using these two different modes of
calculation shows that the combined index is not affected, although the indices for owner-occupied
housing is a little higher when using dwellings, and somewhat lower for rented housing. The data
for Oslo is for the year 2000, while we use 2008 data for the other cities. Tests using Copenhagen
data for 2000 suggest very small differences compared to 2008. The last problem, which we have
already discussed, is that the neighbourhoods do not have the same size. In Oslo and Helsinki
they are on average twice as big population-wise compared with Copenhagen and Stockholm. Like
before, this means that Copenhagen and Stockholm values tend to be somewhat higher than
would they have been using larger data containers.
These complications cannot, however, explain away the marked differences between the cities.
Copenhagen and Stockholm seems to have a much more segregated housing market than Oslo
and Helsinki. In both cities owner-occupied single family houses and social/public housing have the
most uneven distribution across tenures. Special circumstances for Stockholm are that a large part
of social/public housing was transformed into co-operatives in the period 2000 to 2008. One of the
declared objectives for this transformation was to reduce segregation in neighbourhoods with
social/public housing (Öresjö 1997). It was, however, mostly social/public housing in the more
attractive central parts of the city that was transformed, resulting in a stronger concentration of
social/public housing in the suburbs (Andersson and Magnusson 2011). This resulted in a marked
increase in the segregation of tenures where the index raised from 0.39 to 0.52 from 2000 to 2009.
In Copenhagen the high segregation of private renting and co-operatives is mostly due to the fact
that these tenures are concentrated in the older housing stock located in central Copenhagen.
In Helsinki owner-occupied single family housing is much more unevenly distributed than the other
tenures and also more so than in Copenhagen and Stockholm. Co-operatives are a very small
sector without any importance for the combined index, but the rented tenures and owner-occupied
flats are quite evenly distributed across neighbourhoods. It is thus remarkable that the relatively
large social/public housing sector is more dispersed in Helsinki than in Stockholm and
Copenhagen. This can be directly attributed to policy: Helsinki has pursued a conscious policy that
was launched already in the end of 1970s to mix different tenures in housing areas (Lankinen
1997).
In Oslo, both owner-occupied housing and private renting are fairly evenly distributed across
neighbourhoods. The tenure most spatially concentrated in Oslo is the co-operative sector, which
comprises mainly large co-operative organisations.
Estimating the connection between tenure segmentation and spatial segregation
As described in the methodological section we make this analysis in two steps. First we build a
regression model to estimate the connection between the tenure composition and the proportion of
immigrants in the neighbourhood. We do this for both all immigrants and for the subcategory of
Non-Western immigrants. In the second step we use this model to estimate the expected number
of immigrants in the neighbourhoods and calculate the expected indices of dissimilarity and the
uneven spatial distribution for these expected numbers.
17
The regression model is described in the methods section. The dependent variable is the logit
value of the proportion of immigrants in the neighbourhood. The independent variables are the
proportion of residents 20+ years living in each of the different tenures in each neighbourhood. For
Helsinki we use the proportion of dwellings in each tenure form.
The models explain a considerable part of the variation in the proportion of immigrants in the
neighbourhoods. The first row of Table 8 shows the R2 values for the models for all immigrants. In
all models more than 40 per cent of the variation is explained. The model for Helsinki has an even
higher degree of explanation.
Table 8. Estimated indices of the spatial distribution of immigrants and segregation computed from
linear regression models, compared to the actual measured indices
Copenhagen Helsinki
Variation explained by regression
model R2 (all immigrants)
Oslo Stockholm
,44
,62
,43
,46
,22
,19
,22
,25
0
-15
1
14
Copenhagen
(dwellings)
,43
Index of uneven distribution
All immigrants
Actual index
Relative deviation from average for all cities %
Estimated tenure dependent index
,14
,15
,17
,18
Tenure effect on spatial distribution
64%
78%
77%
73%
Actual index
Relative deviation from average for all cities %
,33
,25
,36
,40
-1
-25
7
19
Estimated tenure dependent index
,23
,21
,28
,30
Tenure effect on spatial distribution
68%
81%
76%
74%
Actual index
Relative deviation from average for all cities %
,25
0,21
,26
0,33
-3
-21
-1
25
Estimated tenure dependent index
,17
,16
,20
66%
79%
78%
0%
Actual index
Relative deviation from average for all cities %
,36
0,3
,40
0,48
-4
-28
5
27
Estimated tenure dependent index
,25
,22
,31
0,14
Non-Western immigrants
0,22
Index of dissimilarity
All immigrants
Tenure effect on segregation
Non-Western immigrants
Tenure effect on segregation
69%
82%
78%
0%
Note: The indices for Copenhagen and Stockholm can be expected to be a little too high because of smaller neighbourhood size
table 8 shows the index of dissimilarity as well as the calculated index of uneven distribution. It can
be seen that the differences across the cities between the indices of uneven distribution are
smaller than the differences between the indices of dissimilarity, but the rank order of the cities is
not changed. This indicates that the differences in segregation across the cities cannot be
explained just by differences in the level of immigration. However, it also illuminates the fact that
forces of segregation tend to be stronger in cities with a higher level of immigration.
18
The calculated expected indices can be interpreted as the segregation or uneven spatial
distribution of immigrants that would have occurred if the housing market (tenure distribution) was
the only cause of segregation. As expected, all the calculated expected indices are smaller than
the actual indices. But the difference between the actual and the calculated indices is relatively
small. In Table 8, what we label the ‘Tenure effect on spatial distribution’ varies between 64 and 76
per cent. This is simple the proportion of the existing variance that can be explained by the
neighbourhoods’ tenure composition.
The percentage differs somewhat between the countries. The housing market in Helsinki, Oslo and
Stockholm seems to have a greater importance for the spatial distribution of immigrants than for
Copenhagen. It is also nearly the same differences when looking at Non-Western immigrants. For
Copenhagen it also does not matter if dwellings or residents 20+ years are used in the regression
model. This indicates that the figures from Helsinki are quite comparable to the other cities.
The estimated tenure dependent indices of dissimilarity are -–as expected– higher than the
expected indices of spatial distribution. The tenure effect on segregation is only a slightly higher
than the effect on spatial distribution and it seems like that this is the same in all four cities.
Comparative overview
In the research literature Nordic welfare states are often treated as a homogenous group. But as
we have shown in this article some important disparities exist between housing policies and
housing markets across these countries. These differences have important effects for the way
immigrants are distributed in space and to what extent they are spatially separated from natives.
Table 9. Main results of the comparison of ethnic segregation and the connection with housing
marketsegmentation.
Table 9. Main results of the comparison of ethnic segregation and the connection with housing markets.
The capital regions
Copenhagen
Helsinki
Oslo
Stockholm
-17
-37
2
52
Uneven spatial distribution of immigrants
-4
0
-22
-15
2
1
25
14
Income segmentation of housing markets
23
-23
40
-40
Ethnic segmentation do
-15
33
37
-40
-3
-14
-19
21
64%
79%
77%
74%
Relative deviation from average for all cities %
Proportion born outside the country
Ethnic segregation
Uneven spatial distribution of tenures
Tenure effect on spatial distribution
Note: The indices for Copenhagen and Stockholm can be expected to be a little too high because of smaller neighbourhood size
There are quite big differences between the levels of ethnic segregation in the four cities. In
Stockholm immigrants are most segregated from natives and in Helsinki the least. Oslo and
Copenhagen is at the same level somewhere in between. The differences can partly be explained
by a general higher proportion of immigrants in Stockholm and a smaller proportion in Helsinki.
Differences between the cities are thus reduced when looking at how immigrants are spatially
distributed across neighbourhoods. But Helsinki still has a more even spatial distribution than the
other cities, while Stockholm is closer to the level in Oslo and Copenhagen
19
Explanations for the differences between the cities are sought by comparing the housing policies
and housing market structure in the cities concerning tenure division and spatial distribution of
tenures.
It is shown that the ethnic segmentation of housing markets differ much. Helsinki has the most
ethnically segmented market while the other three cities are at about the same level. However, the
reasons for segmentation differ across the cities. The higher segmentation in Helsinki is much due
to a high concentration of immigrants in social/public housing. This is also found in the other cities,
but at a lower level. In Oslo the concentration in social/public housing is highest but at the same
time the sector is so small that it has a smaller impact on the measured general level of
segmentation. In Copenhagen segmentation is partly caused by low representation, especially of
Non-Western immigrants, in private renting and co-operatives, while immigrants are much better
represented in private renting in Oslo and Stockholm, and in co-operatives in Oslo. In all four cities
immigrants are much underrepresented in owner-occupied housing, but more so in Helsinki than in
the other cities.
The explanations for these differences in ethnic segmentation can to a great extent be found in the
differences in housing policies, which have different consequences for immigrants housing options.
As a result of these policies, tenures in the four countries are a different extent socially divided. In
the comparative literature on housing policies, unitary and dual housing systems are established
concepts. In the dual systems there are significant differences between the social composition of
homeownership and renting. Many immigrants belong to the lower income groups and under such
conditions a dual market setup will also lead to ethnic segmentation. Moreover, it is important if the
rental markets are unitary meaning that social/public supported housing is accessible for all
households and competing on even terms with private renting. In dual rental systems public
housing is a restricted sector reserved for low income groups.
Sweden and Finland have a less uneven social balance between renting and owing and thus have
more unitary housing systems. The rental market in Finland is on the other hand more dual
because there has been strong needs tests in the social/public housing sector. In Denmark and
especially in Norway there is a stronger separation between renting and owing and these countries
thus have more dual systems. In the Norwegian case it has to do with strong political objectives for
homeownership, which means that low-income groups to a greater extent are left behind and
isolated in the rental sector. The rental sector in Denmark is more unitary while it is more dual in
Norway because of a very small and strongly needs tested social/public housing sector. In Sweden
and Norway the large co-operative sectors, however, form a bridge between renting and owing,
which to some extent make the housing systems more unitary. Although we acknowledge the
existence of path dependency (Bengtsson et al 2006) it is important to realize that housing policy
sometimes undergo more radical changes. Recent changes in the Swedish system has led some
analysts to declare the present Swedish model as being “a monstrous hybrid” (Christophers 2013).
It is still too early to assess how these regulatory and institutional reforms may impact on ethnic
segregation processes. What they currently produce is without doubt increasing housing shortages
in the main urban areas and this will likely have negative effects on less well established
households such as the young, recently arrived immigrants, and socially vulnerable groups in
general.
20
As shown in Skifter Andersen et. al. (2013) it has special importance for the housing options of
immigrants if housing policy contributes making the housing market transparent and whether
conditions for ethnic discrimination are limited, counteracted or enhanced. The size of the
social/public housing sector and the rules for access to this sector are of special importance. In
Oslo, research has indicated that the small rental sector has resulted in strong competition
between house hunters looking for rented accommodation, which has increased discrimination.
This could also partly have been the case in Helsinki, but more important here is the easier access
to social/public housing for immigrants with lower incomes because of strong needs tests. In
Copenhagen rent and price control in private renting and co-operatives has created a surplus
demand, which to some extent has pushed immigrants out from these tenures creating a strong
concentration in social/public housing.
But as discussed in the theoretical section other factors have importance, among others that
immigrants might have special housing preferences which influence their choice of housing and
neighbourhood location. It can be seen from Table 9 that there is no straight forward connection
between income segmentation and ethnic segmentation across the countries.
The effects of housing market segmentation on immigrants distribution in space depends on to
what an extent tenures are spatially divided or mixed. This segregation of tenures is influenced by
planning systems and by the kind of developers and builders which dominates the market. In
Helsinki, Stockholm and Copenhagen social/public housing estates have mainly been built by large
housing associations or municipal companies. In particular during the 1960s and 1970s, they
tended to build large estates with a concentration of social/public housing. In Oslo large cooperative companies have been behind the creation of large co-operative estates. In Stockholm
cooperative housing traditionally was mixed in with rental housing and many cooperatives were
and still are small (around 7,600 independent cooperatives in 2012 in Stockholm county, many
owing less than 20 dwellings). Moreover, the historic development of housing markets and cities
has importance. Especially in Copenhagen private renting and co-operatives are found in the older
housing stock, which are concentrated in the central parts of the city.
Analyses of the spatial distribution of tenures across the cities reveals that Copenhagen and
Stockholm seem to have the most segregated housing markets; especially social/public housing is
spatially concentrated. What has also been found in Stockholm is that there is a pronounced ethnic
segregation within each tenure form, partly indicating local housing careers where relatively
successful immigrant households tend to start in a public housing large estate and later move into
cooperative or home ownership housing in surrounding areas (Andersson, Hogdal, Johansson
2007). Despite having a dominance of market forms (cooperatives or home ownership) some
neighbourhoods adjacent to the large rental dominated housing estates have relatively high levels
of ethnic minorities. And at the same time, public housing in inner city neighbourhoods has very
low levels of minority presence.
In Helsinki, owner-occupied single family houses are very segregated, but all the other tenures
seem to be more mixed in space. The final result is that Helsinki in general has the lowest spatial
segregation of tenures. It is thus remarkable that the relatively large social/public housing sector
has been more spread out in the city than in Stockholm and Copenhagen. This relates directly to
the social mix policy pursued in Helsinki from the 1970s onwards (Lankinen 1997). In Oslo, owneroccupied housing and private renting are quite evenly distributed across neighbourhoods. The most spatially
concentrated tenure is the co-operatives, which have predominantly been produced by large co-operative
organisations, much of it in suburbs with a high concentration of immigrants.
21
The connection between tenure segmentation and the spatial distribution of immigrants is
examined by employing a regression model built on the correlation between tenure composition of
neighbourhoods and the proportion of immigrants. Based on this statistical model estimates for the
tenure dependent number of immigrants in the neighbourhoods were used to calculate indices for
the expected spatial distribution of immigrants if solely dependant on the segregation of tenures.
The analyses showed high correlation between tenure segregation and the distribution of
immigrants. Variation in tenure composition of neighbourhoods explains between 40 and 70 per
cent of the variation in the proportion of immigrants. The calculated index for the expected spatial
distribution of immigrants, determined by tenure segregation, accounted for 60 to 80 per cent of the
actual values. This indicates that the housing market has a very profound impact on the spatial
distribution and segregation of immigrants in the four cities.
The impact of respectively tenure segmentation and spatial segregation of tenures can be
illuminated by comparing the cities.
Helsinki is the most interesting case, because the city has the highest ethnic segmentation of the
housing market while it has the lowest segregation and most even spatial distribution of
immigrants. The explanation is that tenures are much more mixed in space in Helsinki than in the
other cities. This indicates that urban policies and planning to a large extent can counteract the
potential segregation coming from a much segmented housing market.
But an explanation for the low segregation in Helsinki could also be that self-perpetuating
processes of ‘white flight and avoidance’ are weaker or at least concentrated to much fewer
neighbourhoods. In Copenhagen and Stockholm segregation mainly stems from high
concentrations of immigrants in large social/public housing estates where these processes are at
work (Bråmå 2006). In Copenhagen this is partly because immigrants to some extent are excluded
from private renting and co-operatives. Co-operatives are not as spatially segregated in Stockholm
as could expected but the trend is that both recent new constructions and the conversion from
public and private rental into cooperative housing will increase the level of spatial concentration
(Andersson and Magnusson Turner 2011).
Given the low ethnic segmentation of the housing market in Oslo and the relatively low segregation
of tenures one would have expected that segregation in Oslo would be lower than in Copenhagen
and Stockholm. The socially divided and ownership dominated housing market in Oslo has
resulted in the same segregation and uneven spatial distribution of immigrants as in Copenhagen
and nearly as pronounced as in Stockholm. According to Skifter Andersen et. al. (2013) this is
partly due to the fact that many immigrants have been excluded from the rental sector and are
more or less forced to share owner-occupied housing with other families. Moreover, owneroccupied housing is more spatially dispersed than in the other cities. The most important tenure
factor behind segregation seems to be the co-operatives built in larger estates in the suburbs, in
particular in the North-East of Oslo.
Conclusions
In this article segregation and housing systems in four social democratic welfare states has been
compared for the purpose of examining how segregation is influenced by housing policies and
housing markets. The hypothesis to be examined is that a stronger social and ethnic segmentation
22
of the housing market in a country combined with higher spatial segregation of housing tenures will
lead to increased ethnic sorting and segregation.
Based on a comparison of cities in eight European countries Arbaci (2007) concluded that
differences in housing and planning systems play a very important role in explaining differences in
ethnic segregation across countries divided into four ‘welfare and housing regimes’. She found that
countries in ‘The social democratic welfare cluster’, in spite of more uniform and less socially
segmented housing markets, had a relatively high level of ethnic segregation, which she ascribed
to the forms of housing production prevailing in these countries. Production was dominated by
large scale builders and developers creating larger more uniform housing areas, resulting in spatial
segregation of housing tenures. More specifically, ethnic segregation in the countries could be
ascribed to the existence of large spatial concentrations of social/public housing with many
immigrants.
It seems to be assumed by Arbaci that there is a straight forward connection between social
segmentation and ethnic segmentation of housing markets. The comparison of the Nordic
countries shows that this is not always the case. Housing policies have special effects for ethnic
segmentation besides the effects on social segmentation.
It is shown in this article that ethnic segmentation of housing markets has a marked influence on
immigrants’ spatial location and thus segregation because there is a strong connection between
tenure compositions of neighbourhoods and the proportion of immigrants. But the comparison of
cities indicates that the effect is very much dependent on the way housing tenures are distributed
across neighbourhoods. The effects of high segmentation can be eliminated by urban policies that
mix housing tenures in space. In the case of Finland a high concentration of immigrants in
social/public housing has not resulted in a high ethnic segregation, because there has been a
conscious policy of mixing social/public housing with other tenures. We should however be aware
of the fact that the Helsinki analyses are based on relatively large neighbourhoods (average 3000
residents), which might conceal a more fine-graded segregation pattern existing within particular
housing estates (ref Vilkama?). It is nevertheless a fact that housing mix policy in Helsinki aims at
mixing also within particular neighbourhoods and blocks and we judge that ethnic clusters in
general are smaller than in the other three cities and that this not only has to do with the lower
proportion of immigrants but rather with housing policy.
It can also be interpreted from the comparison that differences in segregation across the cities
cannot be explained exclusively by differences in housing markets. The differences in segregation
between Stockholm, Copenhagen and Oslo cannot be explained by just comparing ethnic
segmentation and spatial distribution of tenures. Moreover, the ‘social housing hypothesis’ cannot
be used in one of the cities, Oslo, where this sector is very small while ethnic segregation is
nevertheless pronounced.
References
Aalbers, M. B. (2002) The Neglected Evidence of Housing Market Discrimination in the
Netherlands, Radical Statistics Journal, 79/80, pp. 161–169.
Aarland K. and Nordvik V. (2008) Boligeie blant husholdninger med lave inntekter. NOVA Rapport
15/2008. Oslo: Norsk institutt for forskning om oppvekst, velferd og aldring.
23
Alba, R. and Nee, V. (1997) Rethinking Assimilation Theory for a New Era of Immigration.
International Migration Review, vol. 31, No. 4, pp. 826-874.
Allen J. (2004) Welfare Systems in Southern Europe, in J. Allen et. al. (Eds.) Housing and Welfare
in Southern Europe. Oxford: Blackwell.
Andersson, R. (1998) Socio-spatial dynamics: ethnic divisions of mobility and housing in postPalme Sweden, Urban Studies, 35(3), pp. 397–428.
Andersson, R. (2013) Reproducing and reshaping ethnic residential segregation in Stockholm: the
role of selective migration moves. Forthcoming in Geografiska Annaler B.
Andersson, R. et. al. (2010) Immigration, Housing and Segregation in the Nordic Welfare States.
Department of Geosciences and Geography, University of Helsinki.
Andersson, R., Bråmå, Å., Hogdal, J., (2009) Fattiga och rika - segregerad stad. Flyttningar och
segregationens dynamik i Göteborg 1990-2006. Stadskansliet i Göteborg.
Andersson, R., Hogdal, J. & Johansson, S., (2007) Planering för minskad bostadssegregation.
Rapport 1:2007 Regionplane- och trafikkontoret. Stockholm: Stockholms läns landsting.
Andersson, R. and Magnusson Turner, L. (2011) Segregation, gentrification, and residualisation:
from public housing to market driven housing allocation in inner city Stockholm. Paper presented at
the Symposium on Public Housing Futures, City University of Hong Kong and Fudan Centre for
Housing Studies, Fudan University, 22nd – 23th August 2011.
Arbaci S. (2007) Ethnic Segregation, Housing systems and Welfare Regimes in Europe.
International Journal of Housing Policy, 7:4, 401-433.
Atkinson, R. and Kintrea, K. (2000) Owner-Occupation, Social Mix and Neighborhood Impacts,
Policy and Politics 28, pp. 93-108.
Balchin P. (1996) Housing Policy in Europe. London: Routledge.
Barlow j. and Duncan S. (1994) Success and Failure in Housing Provision. European Systems
Compared. Oxford: Elsevier Science.
Bengtsson, B. et. al. (2006). Varför så olika? Nordisk bostadspolitik i jämförande historisk ljus.
Malmö: Egalitè.
Bleiklie I. (1997) Service Regimes in Public Welfare Administration. Case Studies of Street-level
Bureaucrats and Professionals as Decision Makers. Oslo: Tano Aschehougs Fonteneserie.
Bråmå Å. (2006) 'White Flight'? The Production and Reproduction of Immigrant Concentration
Areas in Swedish Cities, 1990-2000. Urban Studies 2006 43: 1127-1146.
Bråmå, Å. and Andersson, R. (2010) Who leaves rental housing? Examining possible explanations
for ethnic housing segmentation in Uppsala, Sweden, Journal of Housing and the Built
Environment, vol. 25, no. 3, pp. 331-352.
Briggs, X., ed. (2005) The Geography of Opportunity. Washington, DC: Brookings Institution Press.
24
Christophers, B. (2013) A monstrous hybrid: the political economy of housing in early-twenty-firstcentury Sweden. New Political Economy Vol. 18. DOI:10.1080/13563467.2012.753521.
Clark, W. A. V. (1992) Residential preferences and residential choices in a multiethnic context,
Demography, 29, pp. 451–466.
Ellen, I. (2000): Sharing America’s Neighborhoods: The prospects for stable racial integration,
Cambridge: Harvard University Press.
Finney N. (2002) Ethnic Group Population Change and Interaction: A Demographic Perspective on
Ethnic Geographies. Ethnicity and Integration. Understanding Population Trends and Processes,
volume 3, 27-45.
Fong E. and Chan E. (2010) The Effect of Economic Standing, Individual Preferences, and Coethnic Resources on Immigrant Residential Clustering. International Migration Review. Volume 44,
Issue 1, pages 111–141.
Fonseca M. L. , J. McGarrigle and A. Esteves (2010) Possibilities and limitations of comparative
quantitative research on immigrant’s housing conditions. Working Paper No. 06. PROMINSTAT.
Galster, G. (2007) Neighbourhood social mix as a goal of housing policy: A theoretical analysis.
European Journal of Housing Policy 7(1), pp. 19-43.
Hamnett C. and T. Butler T. (2010) The Changing Ethnic Structure of Housing Tenures in London,
1991–2001. Urban Studies 47(1) pp. 55–74.
Johnston R., Forrest J. and Poulsen M. (2002). Are there Ethnic Enclaves in English Cities? Urban
Studies 39:591-618.
Kemeny J. (1995) From Public Housing to Social Renting: Rental Policy Strategy in Comparative
Perspective. London: Routledge.
Kempen, R. van & Murie, A. (2009). The new divided city: changing patterns in European cities.
Tijdschrift voor Economische en Sociale Geografie, 100(4), 377-398.
Lankinen M. (1997) Bostadspolitik mot segregation: läget i Finland, in Nordisk Ministerråd:
Boligpolitik mod segregation. TemaNord Bygog Bolig.
Lindberg G. and A-L. Lindèn (1989) Social segmentation på det svenske bostadsmarknaden
(Social Segmentation on the Swedish Housing Market), Sociologiske Institutionen, Lunds
Universitet.
Lujanen M. (ed.) (2004) Housing and Housing Policy in the Nordic Countries. Nord2004:7.
Copenhagen:Nordic Council of Ministers.
MacDonald, J. S. and MacDonald, L. D. (1964) Chain Migration Ethnic Neighborhood Formation
and Social Networks, The Milbank Memorial Fund Quarterly, 42.
Magnusson Turner, L. (2010) Study on housing and exclusion. Country report Sweden. European
Commission
25
Massey, D. S. (1985) Ethnic residential segregation: a theoretical synthesis and empirical review,
Sociology and Social Research, 69, pp. 315–350.
Molina, I. (2010) Nedslag i diskriminering på bostadsmarknaden. Unpublished report to the Evenity
Ombudsman (DO).
Musterd S. (2012) Ethnic Residential Segregation: Reflections on Concepts, Levels and Effects.
Sage Handbook of Housing Studies. London: SAGE Publications
Musterd S. (2005) Social and Ethnic Segregation in Europe: Levels, Causes and Effects. Journal
of Urban Affairs, Volume 27, Number 3, pages 331–348.
Musterd, S. & Andersson, R. (2005) Social Mix and Social Opportunities. Urban Affairs Review
40(6), pp. 761-790.
Myles J. and Hou F. (2003) Neighbourhood attainment and residential segregation among
Toronto’s visible minorities. Business and Labour Market Analysis Division, Statistics Canada
Nelson A. C., Dawkins C. J, and Sanchez T. W. (2004) Urban Containment and Residential
Segregation: a Preliminary Investigation. Urban Studies 41 vol. 2, 423-439.
Park, R. E. (1925) The city: Suggestions for the investigation of human behaviour in the urban
environment, in: R. E. Park & E. W. Burgess (Eds) The City, pp. 1–46. Chicago, IL: University of
Chicago Press.
Quillian L. (2002) Why is Black–White residential segregation so persistent? Evidence on three
theories from migration data, Social Science Research, 31, pp. 197–202.
Rothenberg, J. et al. (1991) The Maze of Urban Housing Markets: Theory, Evidence and Policy,
University Press of Chicago.
Røed Larsen, E. and Sommervoll D. E. (2011) ”Prisdannelsen i det norske leiemarkedet: en
teoretisk og empirisk analyse av hovedmekanismer generelt og utsatte grupper spesielt” in NOU
2011:15 Rom for alle. En sosial boligpolitikk for framtiden. Ministry of Local Government and
Regional Development, Oslo.
Schaffer B and Huang W (1975) “Distribution and the theory of access” in Development and
Change Vol. 6, No. 2, pp. 13-36. London: Sage Publications.
Skifter Andersen H., Magnusson Turner L. and Søholt S. (2013). The Special Importance of
Housing Policy for the Housing Situation of Ethnic Minorities. Evidence from a Comparison of Four
Nordic Countries. International Journal of Housing Policy vol 13, Issue 1.
Skifter Andersen H. (2012). The importance of housing policies and housing markets for the
housing of immigrants in the Nordic countries. Report for the NODES project. Horsholm: Danish
Building Research Institute at Aalborg University.
Skifter Andersen H. (2010) Spatial Assimilation in Denmark. Why do Immigrants move to and from
Multi-ethnic Neighbourhoods? Housing Studies, Vol. 25, No. 3,pp 281–300.
26
Skifter Andersen H. (2008) Privat boligudlejning. Motiver, strategier og økonomi. Hørsholm:
Statens Byggeforskningsinstitut.
Skifter Andersen H., Andersen H. T. and Ærø T. (2000) Social polarisation in a segmented
housing market. Segregation in greater Copenhagen. Geografisk Tidsskrift nr. 100.
Søholt S (2007) Gjennom nåløyet – en sammenligning av tilpasninger til boligmarkedet blant
hushold av pakistansk, tamilsk og somalisk bakgrunn, Oslo 1970 – 2003 (Through the eye of a
needle – a comparison of adaptions to the housing market among households with Pakistani,
Tamil and Somali background, Oslo 1970 – 2003). Thesis. Oslo: Institute for Political Science,
Faculty of Social Sciences, University of Oslo.
Søholt S and Astrup K (2009) Etniske minoriteter og forskjellsbehandling i leiemarkedet (Ethnic
minorities and uneven treatment in the rental market). NIBR-report: 2. Oslo: Norwegian Institute for
Urban and Regional Research.
Solid, D., Andersson, R. & Molina, I., 2007, Ungdomarna och allmännyttan – en kartläggning av
kommunernas ägardirektiv och bostadsföretagens regler för tilldelning och uthyrning av lägenheter.
In: Andersson, R. (red.) Måste man ha tur? Studier av yngre på bostadsmarknaden i svenska
städer. SOU 2007:14, s. 7-79, rapport nr 2 från Boutredningen. Stockholm: Fritzes.
Swedish Equality Ombudsman (2008) Diskriminering på den svenska bostadsmarknaden – En
rapport från DO:s särskilda arbete under åren 2006–2008 kring diskriminering på
bostadsmarknaden.
(http://www.do.se/Documents/material-gamla-ombudsman/DO_Bostadsrapport_2008_ny.pdf
Accessed March 11, 2013)
Tasan-Kok, T. and Baeten, G. (eds.) The contradictions of neoliberal planning: cities, policies and
politics. Berlin: Springer.
Taylor C., Gorard S. and Fitz J. (2007) A re-examination of segregation indices in terms of
compositional invariance. Social Research Update, University of Surrey.
van Kempen R. (2003) Segregation and Housing Conditions of Immigrants in Western European
Cities, Eurex Lecture 7, 13 March. Available at
http://www.shiva.uniurb.it/eurex/syllabus/lecture7/Lecture7-VanKempen.pdf.
Turner, B., and Whitehead C. (2002) "Reducing Housing Subsidy: Swedish Housing Policy in an
International Context ", Urban Studies, Vol. 39 No. 2, pp. 201 – 217.
Wright R., Ellis M. and Parks V. (2005) Re-placing Whiteness in Spatial Assimilation Research.
City and Community, Volume 5, Issue 2, pp. 111-135.
Öresjö E. (1997) Sverige, in Nordisk Ministerråd: Boligpolitik mod segregation. TemaNord Bygog
Bolig
27
Download