Spatial assimilation? The development in immigrants residential career with duration of stay in Denmark. Hans Skifter Andersen

Spatial assimilation? The development in immigrants residential career with duration of stay in Denmark Hans Skifter Andersen Journal of Housing and the Built Environment ISSN 1566-4910 J Hous and the Built Environ DOI 10.1007/s10901-015-9459-0 1 23

Your article is protected by copyright and all rights are held exclusively by Springer Science +Business Media Dordrecht. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your article, please use the accepted manuscript version for posting on your own website. You may further deposit the accepted manuscript version in any repository, provided it is only made publicly available 12 months after official publication or later and provided acknowledgement is given to the original source of publication and a link is inserted to the published article on Springer's website. The link must be accompanied by the following text: "The final publication is available at link.springer.com. 1 23

J Hous and the Built Environ DOI 10.1007/s10901-015-9459-0 ARTICLE Spatial assimilation? The development in immigrants residential career with duration of stay in Denmark Hans Skifter Andersen 1 Received: 22 October 2014 / Accepted: 8 July 2015 Springer Science+Business Media Dordrecht 2015 Abstract Many studies have shown that immigrants residential situation differs from that of natives and that factors other than housing needs and financial situation influence immigrants options and choices concerning housing and neighbourhood. Research has indicated that immigrants might have a stronger preference for renting due to insecurity about their future situation and that especially newly arrived immigrants tend to live in immigrant-dense, so-called multi-ethnic, neighbourhoods. However, the spatial assimilation theory claims that in the course of time immigrants will move to other kinds of housing and neighbourhoods. In this paper, the residential careers of immigrants in the first years after their arrival are examined and compared with Danes. The hypothesis tested is that over time the housing situation of immigrants gets closer to the comparable one for Danes. It is a longitudinal study based on data from 1985 to 2008 on non-western immigrants in Denmark. The results show that non-western immigrants steadily increased their presence in social housing and multi-ethnic neighbourhoods during their first 10 years of stay; then, their presence stagnates; and after 15 years, it declines. Part of the initial increase in the frequency of living in multi-ethnic neighbourhoods was not due to individual choices among immigrants but could be ascribed to the increasing number of multiethnic neighbourhoods over time. The study confirms spatial assimilation, also when controlling for degree of economic integration, but the change was not dramatic over the 24 years covered by the study. Keywords Immigrant settlement Spatial assimilation Segregation & Hans Skifter Andersen hsa@sbi.dk; hsa@sbi.aau.dk 1 Danish Building Research Institute, Aalborg University, Ålborg, Denmark

H. S. Andersen 1 Introduction Studies of immigrants settlement patterns in Europe and North America have shown that the residential patterns of immigrants differ much from that of the native population (see, e.g. Özüekren and van Kempen 2002; Musterd 2005; Johnston et al. 2002; Finney 2002; Fong and Chan 2010; Andersen et al. 2013). Immigrants tend to live in other kinds of housing and neighbourhoods than natives, and they often cluster in specific housing tenures and neighbourhoods. In this way, parts of the cities have obtained a large proportion of ethnic minorities. It has been discussed in the European literature, whether American theories on immigrants preferences for ethnic enclaves and on their spatial assimilation over time are valid in Europe, or whether immigrants settlement patterns can be fully explained by having more limited housing options due to their lower incomes and due to discrimination against them on the housing market (Peach 1998; Murdie 2002; van Kempen and Özuekren 1998; Andersen 2015b). Preferences for ethnic enclaves imply that new immigrants prefer to settle in neighbourhoods with many residents from their own country or culture. The spatial assimilation theory (Massey and Denton 1985; Massey 1985; South et al. 2005) postulates that the tendency of new immigrants to cluster decreases with the duration of their stay in their new country and that over time they will spread out more in the cities. This development is assumed to follow from a process of cultural, social and economic integration (in the American literature called assimilation ) into the host society where immigrants over time improve their financial situation and to a greater extent share fundamental norms and values of the native population (Gordon 1964; Alba and Nee 1997). The spatial concentration of immigrants in European cities is to some extent connected with their preferences for housing tenure. If ethnic minorities prefer to settle in certain tenures that are spatially separated from other tenures, there is a stronger risk of getting neighbourhoods with high concentrations of ethnic minorities (Arbaci 2007; Andersen et al. 2015). In most European countries, newly arrived immigrants have settled in rented housing, in Denmark, particularly in social housing. Newer research in the Nordic countries (Kauppinen et al. 2010) shows that immigrants much less often achieve homeownership than the native population, also when controlling for background variables such as income, education, age and family situation, which indicates stronger preferences for rented housing among immigrants. In the research literature, homeownership has been seen as an indicator of the integration/assimilation of immigrants (Alba and Logan 1992; Borjas 2002; Sinning 2010). Newly arrived and less integrated immigrants more often prefer to stay in rented housing. It is assumed that in the course of time more of them will move to homeownership, when they become more integrated in the host society and have improved their financial situation. The essence of this discussion is that it might be assumed that immigrants on their arrival more often settle in rented housing and immigrant-dense neighbourhoods and that they change their residence in the course of time to other kinds of housing and neighbourhoods. It is expected that such a development is mainly explained by improvements over time of immigrants employment and incomes. It might, however, also follow from a change in housing preferences due to a more general cultural and social integration where proximity to neighbours from the same ethnic group becomes less important over time. This article examines whether the hypotheses in connection with the spatial assimilation theory are valid in a European country, Denmark. In particular, it is examined whether their validity is independent of the development in immigrants individual employment and

Spatial assimilation? The development in immigrants incomes. It is explored whether immigrants coming to Denmark have over time changed their residential situation in accordance with the spatial assimilation theory and with hypotheses on immigrants choice of tenure, independently of the development in their employment and financial situation. The article contains two main analyses: 1. To what extent did non-western immigrants coming to Denmark in the period 1985 2008 settle in rental housing (social housing) after their arrival and how did this change with the duration of their stay when controlling for background variables like income and employment? Did the probability of staying in social housing decline after some years of stay? 2. Did new immigrants often settle in neighbourhoods with a high concentration of ethnic minorities (immigrants and descendants) after their arrival and did this decline after some years of stay independently of changes in their financial situation? The analyses in the paper thus focus on the connection between the duration of stay and non-western immigrants (descendants are not included) settlement in, respectively, social housing and in immigrant-dense neighbourhoods. As such neighbourhoods in Europe seldom contain one single ethnic group, but a mixture of different groups, they are called multi-ethnic neighbourhoods. The analyses are based on an extensive database containing data on the settlement of non-western immigrants in Denmark during a 24-year period from 1985 to 2008, which made it possible to follow their residential careers over time. This unique database contains individual data on immigrants and their residences for every year in the period, thus making it possible to conduct longitudinal analyses. 2 Spatial assimilation as a determinant of immigrants residential career In earlier American research, the spatial location of immigrants was assumed to be closely connected with their social, cultural and economic integration in the host society. According to the American ecological theoretical tradition of the Chicago School, the spatial distribution of immigrant groups is a reflection of their human capital and the state of their social and economic integration, called assimilation (Alba and Nee 1997). The basic tenets of the ecological model are that residential mobility follows from cultural and social integration. A basic assumption is that for different reasons especially new immigrants prefer to settle in neighbourhoods dominated by their own ethnic group and with strong ethnic social networks, called ethnic enclaves (Fong and Chan 2010). Some authors (Peach 1998; Murdie 2002; van Kempen and Özuekren 1998) have argued that for new immigrants, moving to neighbourhoods with many countrymen called ethnic enclaves is part of a strategy for survival and integration in their new country. Some of the arguments for this strategy are that immigrants often have family or acquaintances in the enclaves, who can supply them with a social network, which can lessen their isolation, and who can support them in the face of disadvantage and discrimination. Finally, the feeling of security and safety in a well-known social and cultural environment might be important. Earlier, the spatial assimilation theory has been formulated (e.g. Massey and Denton 1985; Massey 1985; South et al. 2005). It states that immigrants often start their career in a new country by moving to ethnic enclaves, but that after some time most of them move out again to housing and neighbourhoods that are more in accordance with changes in their

H. S. Andersen preferences, resources and needs. Preferences among ethnic minorities for moving to and from neighbourhoods, where they find enclaves, are thus assumed to depend on the extent to which they are socially integrated in the new society and if they are stigmatised and discriminated against in general. New immigrants and less integrated ethnic minorities have a greater need of the support that they can get from networks in the enclave, which in turn influences their housing choice. On the other hand, residents in enclaves that during the course of time get a stronger position in the new country could change their preferences in favour of moving away from the enclave. The spatial assimilation theory is based on a chain of assumptions. Firstly, it is assumed that newly arrived and less integrated/assimilated immigrants prefer to settle in neighbourhoods with ethnic enclaves. Secondly, it is a premise that immigrants over time integrate/assimilate in the host country which improves their financial situation and to a greater extent make them share fundamental norms and values of the native population. Thirdly, it is believed that the changes in integration influence their preferences for living in enclaves. Finally, it is expected that these new preferences will result in residential mobility away from multi-ethnic neighbourhoods. There is much evidence that immigrants tend to cluster in specific neighbourhoods in European cities, but there is disagreement about to what extent this is caused by immigrants having special preferences for places where to live or for other reasons. Some researchers (Musterd et al. 1998; van Kempen 2003; Bolt et al. 2010a, b) have questioned whether preferences for enclaves are important for immigrants housing choice in Europe compared with other factors. Alternative explanations for ethnic segregation focus on limited options for immigrants on the housing market and are often called the place stratification model (Bolt and van Kempen 2010). Besides the disadvantage of lower incomes, immigrants might have special difficulties on the housing market. If the housing market is not easy to see through, it is likely to make it 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 2009). In parts of the housing market, good contacts to persons or institutions are decisive for access to dwellings. It is also important to have a good knowledge of the possibilities and rules on the housing market, which often also requires good language skills or good access to advisers. Moreover, immigrants might be exposed to discrimination on the housing market, which diminishes their options. Some studies (Aalbers 2002; Andersson 1998; Søholt and Astrup 2009) have concluded that discriminatory practices on the housing market are also found in Europe, where especially private landlords to some extent exclude ethnic minorities. It is, furthermore, documented that the creation of multi-ethnic neighbourhoods to a major extent is a result of the behaviour of the native population. When the presence of ethnic groups becomes very visible in a neighbourhood, segregation processes called White flight and White avoidance may begin to appear. In the USA, it has been observed that Whites flee when the share of Afro-American residents in their neighbourhood exceeds a certain proportion of the population (Wright et al. 2005). In recent years, it has been shown that the strongest process is White avoidance, meaning that natives tend to avoid moving to neighbourhoods with many immigrants or special ethnic groups (Clark 1992; Quillian 2002; Bråmå 2006; Bolt et al. 2008). As a consequence of these processes, it is easier for immigrants to get access to these neighbourhoods, which are often dominated by certain tenures, and prices/rents tend to be lower. The spatial clustering of immigrants might also be influenced by their preferences for housing tenure. Several studies have indicated that immigrants have a much higher preference for renting than natives, which limits their housing and neighbourhood options.

Spatial assimilation? The development in immigrants Kauppinen et al. (2010) have shown that immigrants in the Nordic countries far more seldom than natives move into homeownership and that this cannot be explained by differences in background variables such as income, education, employment, age, urbanisation and family situation. As discrimination of immigrants by banks in the Nordic countries must be expected to be low, 1 it was concluded that immigrants could be expected more often to have a preference for renting. A possible explanation was put forward that households preferences for owner-occupied housing are very much determined by their expectations of the future as described in the literature on housing demand (Artle and Varaiya 1978; Carliner 1974). If people have strong expectations of staying in a certain city and of a stable and perhaps increasing income, the preference for homeownership will be stronger. Immigrants might more often have uncertain expectations about their future employment and income and about whether they are going to stay in the country and would therefore be more reluctant to invest in homeownership. A limitation on most of the studies on causes of ethnic clustering is that they are based on data on how ethnic minorities actually settle and move, from which are hypothesised preferences and other causes (e.g. Johnston et al. 2002). A Canadian study, based on survey data (Fong and Chan 2010), on two recently arrived immigrant groups, Asian Indians and Chinese, concluded, however, that co-ethnic locational preferences were strong and to a high degree explained co-ethnic clustering. A Danish study among ethnic minorities, also based on survey data (Andersen 2006b, 2010, 2015b), found that some ethnic minorities in Denmark actually possess strong preferences for living close to an ethnic social network and in neighbourhoods dominated by ethnic minorities. A strong connection was found between these measured preferences and indicators of social integration, also when controlled for other variables like income. Ethnic minorities with strong preferences for enclaves tended less often to master the Danish language, to have work and to have good contacts to Danes. The study thus supported the hypothesis that some immigrants have preferences for ethnic enclaves and for multi-ethnic neighbourhoods and that this was influenced by their social integration and correlated with the duration of their stay in the country. The second assumption, which the spatial assimilation theory is based on, is that immigrants over time get culturally, socially and economically integrated in the host country. In classic American literature (Gordon 1964) on immigration and social integration, the theory was put forward that immigrants and their descendants over time are assimilated into the host society sharing fundamental norms and values of the native population (Park and Burgess 1969; Gordon 1964). The idea of the inevitability of assimilation has been criticised, however, for assuming that all ethnic content is imported by immigrants and for not recognising that it can be created in response to conditions and out of cultural materials in the host society. As stated by Alba and Nee (1997): Over time ethnic groups were expected to become assimilated into their host society: most did economically at least, but not always socially and assimilation involves the decline, though not always the disappearance of ethnic/racial distinctions. The idea of assimilation has been questioned in relation to the European experience, and it has also been questioned in connection with new waves of ethnic groups coming to the USA and Canada. Critics have argued that ethnicity may instead go through periods of recreation, if not renaissance (Glazer and Moynihan 1970; Conzen et al. 1992). Others maintain that the theory of assimilation still holds in the USA, but that it takes more time 1 A Danish survey among moving ethnic minorities by Andersen (2006b) showed that only seven per cent had trouble getting mortgage.

H. S. Andersen for the newer waves of immigrants, sometimes several generations (Alba and Nee 1997). It is argued (Alba and Nee 1997) that assimilation theory made a mistake in presupposing integration into the values of the White middle class. Some (Valdez 2006; Portes and Rumbaut 2001) has formulated a theory of segmented assimilation, where immigrants and subsequent generations are integrated into different segments of American culture and some into permanent poverty and assimilation into the underclass, what has been coined downward assimilation (Model 1991; Portes and Zhou 1993). Results from more recent studies of the development in employment and education for immigrants and their descendants in the USA and Canada show signs of assimilation over time, but there are big differences between different groups (Valdez 2006; Greenman and Xie 2008; Slack and Jensen 2007). Greenman and Xie s study points to the important role of social context: Both processes and consequences of assimilation should depend on the local social context in which immigrants are embedded. But they find little empirical evidence supporting hypotheses derived from segmented assimilation theory. In Europe, immigration in most countries has a much shorter history than in the USA. In Denmark, guest workers from Turkey and Pakistan first appeared in the 1960s and the first really big waves of refugees arrived around 1990. The first generation of immigrants is still heavily dominating. Some European studies, mostly based on data on education, employment and incomes, support the classical assimilation theory, while others do not. In a German study, Kalter and Granato (2001) conclude that in spite of the fact that the educational gap (between immigrants and natives) has clearly widened over the years under observation, it will turn out that in respect of the other aspects of life the general trend appears to be towards assimilation, especially for the second generation of the classical labour migrants. Luthra (2009) concludes from his study that second generation guest workers in Germany, in particular Turks, Iberians and Greeks, show a significant immigrant advantage compared with the first generation. But there are considerable differences between different ethnic groups. It was found that different ethnic groups establish deviant subcultures, often in specific urban locations. Safi (2008) made a French study, which could not support the existence of a uniform convergence process for different immigrant groups and concluded that Other, more complex, segmented models seem to characterize the various communities represented in the survey. Luthra (2009) discusses the opposite hypothesis of the assimilation theory; some immigrant groups tend to develop a subculture or to end up in permanent deprivation which tend to isolate them from society. His conclusion is: Thus, the predictions of divergence of compounded disadvantage (or, poor outcomes for poorly received groups) as well as compounded advantage (accelerated progress for positively received groups) does not appear to hold in the German case. I therefore conclude that ethnic origins do matter, but not in the ways consistent with assimilation theories as they are applied in the United States. In another German study (Schaeffer and Bukenya 2010), it was concluded that language is one of the most critical factors for determining integration and assimilation at the workplace and in society. Another important factor found in the study was the attitudes by Germans towards immigrants. In a study comparing Germany, France and the Netherlands, Ersanilli and Koopmans (2010) found that host culture adoption is best measured by host country identification, host country language proficiency and use, and interethnic social contacts. The last two presumptions for the spatial assimilation theory are that increased integration/assimilation in the host country of immigrants will weaken their preferences for living in enclaves and that these changes will result in moves away from multi-ethnic neighbourhoods. There has been some discussion about this in the literature. Some argue that the access to ethnic network and resources, combined with mobility possibilities inside

Spatial assimilation? The development in immigrants enclaves, will motivate ethnic minority households to pursue a housing career inside the same neighbourhood (Murdie and Ghosh 2010). Competing with the spatial assimilation theory is the Ethnic resources theory (Portes and Bach 1985), the Cultural Preference theory or the Ethnic enclave model (Bolt and van Kempen 2010). They all assume that the advantages for ethnic minorities to live in neighbourhoods with a strong ethnic network are so important that they exceed the benefits by spatial assimilation. Based on a review of American literature, Wright et al. (2005) state that The successive waves of arrivals who enter the immigrant neighbourhood spin intricate webs of social networks that provide information on housing and labour markets and other institutions. As the number of foreign-born arrivals increases, so too does the density of networks and social bonds. These links operate to unite first-generation immigrants (and their progeny) together in a mutually supportive system of reciprocal relations centred on the immigrant neighbourhood. Fong and Chan (2010) concluded from their study of Asian, Indian and Chinese immigrants in Toronto that levels of co-ethnic clustering are not related to the economic resources of immigrants, which points to a persistent clustering among the more economically integrated members of these groups. In Europe, immigrant neighbourhoods are more mixed and have many different minorities than in the USA due to more regulated housing markets, especially because of the strong role of public housing in many countries where immigrants tend to cluster. It is possible that ethnic networks and bonds are less intensive in these neighbourhoods and advantages for immigrants to remain there are weaker. A Danish study (Andersen 2010) has thus shown that a considerable proportion of immigrants in multi-ethnic neighbourhoods move out every year and that outmovers are more integrated than both permanent residents and inmovers measured by indicators such as employment, income and Danish citizenship. A Dutch study of immigrants moving pattern (Schaake et al. 2014) also showed signs of spatial assimilation, but this study was only based on data on income and education. A study of immigrants moves into homeownership in the Nordic countries (Kauppinen et al. 2010) showed that this migration, as expected, was influenced by indicators of economic integration, such as employment and income, but that these factors were weaker for immigrants than for natives in the countries. One of the reasons was found in the fact that living in a multi-ethnic neighbourhood reduced the probability of moving into homeownership. To sum up, it can be concluded that the spatial assimilation theory is open to discussion, especially in a European context. It is uncertain to what extent the clustering of immigrants in multi-ethnic neighbourhoods is a result of preferences for enclaves or whether it is due to other causes. It is questionable whether all immigrants get culturally, socially and economically integrated/assimilated over time. And it is arguable whether this integration results in changed residential preferences and moves out of multi-ethnic neighbourhoods. 3 Data and methods In Denmark, researchers have the possibility of creating their own databases inside the national statistical organisation, Statistics Denmark, where all public registers can be used as sources. Moreover, Denmark has a special housing and building register, which makes it possible at any point in time to connect data on the population with data on their housing conditions and tenure.

H. S. Andersen For this study, a database was constructed that contained information on ethnic minorities aged 15? and on a 7 % sample of Danes, and their housing, every year from 1985 to 2008. The database contained information on 7.6 million cases (person-years) with Danes and 3.4 million cases with non-western ethnic minorities, corresponding to in average about 141,000 non-western immigrants and descendants, 15? years old, for every year. The data contain one case for every year for every person. Each case is treated as an individual observation with information on age, income, length of stay, housing, neighbourhood, etc. for that particular year. A person can thus be represented in several cases, but most of his data are different in different cases. This made it possible to construct data on immigrants situation on arrival in Denmark and the development over time in their residential situation. Moreover, it was possible to construct data on the neighbourhoods where people had been living during the period. In the study, Denmark was divided into 9000 neighbourhoods with in average of about 600 inhabitants. This division was constructed in an earlier project on segregation (Damm et al. 2006) based on the following principles: cities were first divided in accordance with physical boundaries such as railways, big roads, space used for industrial purposes and empty spaces. These bigger areas were subdivided in such a way that different housing tenures were separated as much as possible. This meant that large social housing estates were separated from neighbourhoods with other tenures. Aggregate data on the population and housing in these neighbourhoods were constructed for the years 1985, 1990, 1995, 2000, 2005 and 2009. In particular the proportion of residents belonging to different ethnic minorities (immigrants and their descendants) was calculated. The analyses in the paper concentrates on immigrants aged 15?, who did not live with parents and who came to Denmark in the period 1985 to 2008 from so-called non-western countries. These countries are defined as being outside Western Europe (the old EU, Switzerland, etc.), North and South America, Australia and New Zealand. Immigrants from the new EU member states from Eastern and Central Europe are included in non-western immigrants, like all Asia and Africa. In the statistical analyses, they are divided into four groups: (1) Eastern Europe, (2) Middle East and North Africa (all the Muslim countries from Afghanistan to Morocco), (3) Other Africa and (4) Other Asia. The analyses in this paper focus on the connection between duration of stay and staying in, respectively, social housing or multi-ethnic neighbourhoods with more than 20 % non- Western ethnic minorities (immigrants? descendants in all age groups) among the residents. 4 Immigration to Denmark Denmark is a relatively new immigration country. The number of immigrants and descendants from non-western countries increased from 150,000 in 1980 to 490,000 in 2009. Some of these immigrants became permanent settlers, while others stayed only temporarily in the country. Immigrants from the Middle East (North Africa and Western Asia) more often stayed in Denmark, which was the reason why their number increased steadily over the years from 20,000 in 1980 to 160,000 in 2009. The number of immigrants from other African countries rose from 2200 to 33,000 in the same period. The number of immigrants from Eastern and Central Europe and from Eastern Asia also increased, but in recent years these

Spatial assimilation? The development in immigrants immigrants have to a greater extent been seeking work or education. Many of them can be expected to leave the country again and cannot be considered permanent settlers in the country. This is even more pronounced for immigrants coming from the Nordic countries, from Western Europe and from North America. As a result, the proportion of immigrants in Denmark rose to about 10 % in 2010. A little less than half are immigrants from non-european countries and about 1 % is from Eastern and Central Europe (Table 1). 5 Family situation at the time of immigration Different immigrants are in very different positions when they arrive. Besides their education and ability to get work, it is also important for their housing options how they came to the country and their family situation. Especially immigrants who were granted family reunification found themselves in another situation because they moved in with their family who had settled earlier in the country and had already taken the first steps in a housing career. Figure 1 shows to what extent non-western immigrants (including Central and Eastern Europe) aged 15? came by themselves as a single person or a family, or were granted family reunification with someone already living in Denmark. The last group is divided into family reunification with parents, with other immigrants or with Danes. Significant changes took place over the years. Except for 1995, when a large number of families came from Bosnia, single immigrants constituted the largest group, especially after 2000 when labour migration and students were dominating. Family reunification with other immigrants was the second largest group, but not after 2002. Quite a lot of non- Western immigrants were unified with Danes. Family reunification with parents increased steadily over the years and was the second largest group at the end of the period. 6 Immigrants position on the housing market in Denmark Denmark has a relatively large social housing sector, which it is relatively easy to get access to, even if some waiting time occurs in large cities. In general, it may be said that immigrants in Denmark, compared with many other countries (Andersen et al. 2013; Nielsen et al. 2014) have had very easy access to social housing. Table 2 lists how non-western immigrants and descendants aged 15? are distributed on housing tenures. In Denmark, the dominating majority of ethnic minorities was located in social housing, while relatively few had obtained homeownership. The proportion of all non-western immigrants and descendants living in social housing rose from about 38 % in 1985 to 60 % in 2001. In the following years, the proportion declined somewhat to 54 % in 2008. Table 1 The proportion of immigrants in Denmark 2010 Proportion of population born outside the country (%) 9.8 Immigrants from Eastern and Central Europe (%) 0.8 Immigrants from non-european countries (%) 4.5 Source: Statistics Denmark

H. S. Andersen Number of persons 14,000 12,000 10,000 8,000 6,000 4,000 Mode of immigra on and familiy situa on on arrival Coming alone Coming together with family Unified with parents Unified w immigrant Unified w Dane 2,000 0 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Fig. 1 The development 1985 2008 in the mode of immigration and family situation on arrival for non- Western immigrants aged 15? (Source: Database on the population in Denmark 1985 2008) Table 2 Danes and ethnic minorities from non-western countries aged 15? distributed on housing tenures in 2008 (%) Owneroccupied detached Owneroccupied flats Cooperatives Social housing Private renting Not a selfcontained flat All Danes 60 4 6 15 14 1 100 Ethnic minorities 20 5 5 54 14 2 100 Overrepresentation by ethnic minorities -67 25-17 260 0 100 Ethnic minorities include descendants aged 15?, but this is a relatively small group Source: Database on the population in Denmark 1985 2008 7 Immigrants settlement in social housing over time after arrival The tendency to settle in social housing was influenced by differences in background factors across the different periods and arrival times. A statistical model (logistic regression) was constructed comparing immigrants with Danes concerning their presence in social housing. Moreover, models were constructed to compare different immigrant groups. In the first models, the probability of living in social housing for immigrants was compared with the probability for Danes. As immigrants who arrived by family reunification did not start a housing career from scratch, we chose only to include those in these analyses, who did not arrive by family reunification. Immigrants were divided into five groups in accordance with length of stay in the year in question: 0 1, 2 5, 6 10, 11 15 and 16 24 years.

Spatial assimilation? The development in immigrants In the model, the following control variables were used: Age: 17 29, 30 44, 45 59, 60 Sex Family situation: single without children, single with children, couple without children, couple with children, complex households Size of family: number of persons Employment situation: on pension, on welfare, on unemployment benefit, student, employed Household income: grouping in national income quintiles for the current year Urbanisation: capital region, three largest provincial cities, other cities larger than 10,000 inhabitants, other. Time period: 1985 1987, 1988 1992, 1993 1997, 1998 2002, 2003 2006, 2007 2008 The analysis was performed for three different groups: (1) All non-western immigrants (excluding those who arrived by family reunification), (2) immigrants coming from refugee countries, because since 1998 refugees cannot choose their residence themselves in the first 3 years after arrival, but are placed by local authorities, and (3) immigrants arriving alone without family because singles have other conditions on the Danish housing market. We did not have data in the database identifying refugees, so we chose to incorporate all immigrants from certain countries with many refugees in Denmark. 2 The dependent variable is living in social housing or not. Immigrants are divided into groups according to their duration of stay, and Danes are a reference group in the regression. The control variables are not interesting as they mostly depict the pattern of Danes. Only the results for the length of stay variable are shown in Table 3 because the other variables depict the whole group of Danes plus immigrants. Just after arrival, the probability of living in social housing was not much higher than for Danes compared with later in their stay. The probability then increased with their length of stay until after 10-year stay when it stagnated. For immigrants having stayed more than 16 years in the country, the probability was then getting lower, even when controlling for variables measuring economic integration such as income and employment. This indicates that some immigrants are leaving social housing after 11 15 years of stay in Denmark even if they do not experience a progress in their financial situation. Refugees constitute a more vulnerable group. Since 1998, they have been allocated housing by local authorities and that might be the explanation why they show a higher probability already in the first years, but also for this group it declined after 10 years of stay. Immigrants, who arrived alone, less often stayed in social housing during the first years but after 10 years that proportion increased to nearly the same level as the average. Some of these immigrants probably later received relatives granted family reunification. 7.1 Comparison of immigrant groups To gain more insight into differences between different immigrant groups, two models only containing non-western immigrants were constructed including variables on their background. Besides the control variables described, new variables are as follows: 2 Among refugee countries are chosen: Iraq, Lebanon (Palestinians), Iran, Somalia, Vietnam, Sri Lanka, Afghanistan.

H. S. Andersen Table 3 Results of logistic regression (odds ratios) comparing the probability of living in social housing for non-western immigrants aged 15?, who did not arrive by means of granted family reunification, with the probability for Danes Years of stay Odds ratios All Refugees Arrived single 0 1 year 1.9 3.4 1.2 2 5 years 5.0 8.4 3.1 6 10 years 9.4 13.4 7.1 11 15 years 9.5 12.1 8.7 16 24 years 7.1 7.9 7.2 Nagelkerke R 2.307.316.246 Only immigrants not living with parents and who did not immigrate by means of granted family reunification are included. The dependent variable is living in social housing or not, and Danes are a reference group. All odds ratios are significant below 0.001 Country group: Eastern Europe, Middle East and North Africa, Other Africa and Other Asia From refugee country? Situation on arrival: arrived alone, arrived as a single parent, arrived as a couple without children, arrived as couple with children, arrived by family reunification with parents, family reunification with another immigrant, family reunification with a Dane, arrived together with parents. The two models have two different selections of immigrants: (1) All non-western immigrants who arrived after 1984. (2) All non-western immigrants arriving after 1984, who had not been granted family reunification when they came. Finally, control variables for Danes are shown. The duration of stay groups were here compared with the frequency of living in social housing for immigrants just arrived (stay 0 1 year; Table 4). In both analyses, the probability of living in social housing increased for those who had been longer in the country like in Table 3. The stagnation after 10 years was more pronounced. The pattern was thus the same for all non-western immigrants as for immigrants without family reunification. The way of immigration had an importance for settlement in social housing. It can be seen from the first column, third section, that immigrants with family reunification in general had a higher probability of living in social housing, probably because their partners or parents in the country were often living in social housing. The odds ratio was lower, only if immigrants were granted family reunification with a Dane. For immigrants without family reunification, those who arrived as singles had the lowest probability, while couples with children had the highest. There are some differences between country groups. Among immigrants arrived without family reunification, those coming from Eastern European or Other Asia less often lived in social housing, while immigrants from Other Africa most often did. The composition of immigrants who settled in social housing differed from that of Danes. It can be seen from column 3 that among Danes residents in social housing mostly are younger and singles. Among all immigrants, and especially those without family reunification, older people more often lived in social housing than the youngest and

Spatial assimilation? The development in immigrants Table 4 Results of three logistic regressions (odds ratios) on factors determining when non-western immigrants and Danes aged 15?, not staying with parents, lived in social housing 1985 2008 All arrived after 1984 Arrived without family reunification Danes Duration of stay (ref.: 0 1 year) 2 5 years 1.58 2.05 6 10 years 2.06 2.94 11 15 years 2.04 2.75 16 years and above 1.79 2.10 Country group (ref.: other Asia) Eastern Europe 1.02.96 Middle East and North Africa 1.56 1.20 Other Africa 1.49 1.44 Situation at arrival (ref.: arrived alone) Arrived as single with children 1.21 1.10 Arrived as couple without children 1.20 1.13 Arrived as couple with children 1.33 1.14 Family reunification with parents 1.48 Family reunification with immigrant 1.89 Family reunification with Dane.69 Arrived with parents 1.34 Coming from refugee country 1.90 2.22 Period (ref.: 1985 1987) 1988 1992.86.75 1993 1997.84 1.04 1998 2002 1.11.96 1.05 2003 2006.91 1.07 2007 2008.85.75 1.02 Women?.80.91.82 Age group (ref.: 15 29) 30 44.92 1.07.95 45 59 1.13 1.33.88 60 1.56 1.87.54 Family type (ref.: single without children) Single with children 2.32 2.10 2.88 Couple without children 1.01 1.00.51 Couple with children 1.21 1.36.44 Complex household.61.59.32 Size of family: number of persons 1.17 1.15 1.02 Employment situation (ref.: employed) On pension 2.08 1.91 2.20 On welfare, etc. 1.19 1.19 1.43 On unemployment benefit 1.46 1.42 1.41 Student 1.02.99.71 Household income quintiles (ref.: Quint 5) Quint 1 2.34 2.82 4.96

H. S. Andersen Table 4 continued All arrived after 1984 Arrived without family reunification Danes Quint 2 2.84 3.51 4.32 Quint 3 2.45 2.98 3.49 Quint 4 1.88 2.12 2.43 Urbanisation (ref.: towns <10,000 inhabitants) Capital region 1.91 1.74 4.11 Three biggest provincial cities 5.16 4.89 4.20 Other cities and towns [10,000 inhabitants 3.79 3.49 3.30 Nagelkerke R 2.294.318.195 All shown odds ratios are significant on the 0.05 level families with children more often settled there than those without children. This is the opposite of the pattern found among Danes where families with children mostly had homeownership and very seldom lived in social housing. It might be assumed that the most vulnerable families with the lowest income would be the most overrepresented in social housing. This was very apparent for Danes, who showed major differences between different income groups. But it was not so clear-cut for immigrants. Looking at the differences between the income groups (quintiles), it is seen that quintile 2 had the highest probability and the differences between groups were smaller than for Danes. This indicates that also more well-to-do immigrants settled in social housing. The results thus depict a different pattern of settlement in social housing for immigrants than for Danes. While social housing for Danes is mostly used by the more marginalised households or as a start in their housing career, it is much more common among immigrants to perform their housing career inside the social housing sector and stay there when they establish a family and get children. 8 Settlement in multi-ethnic neighbourhoods As far back as the 1970s, immigrants in Denmark began to concentrate in certain parts of the cities. In the first years, they settled in the older parts with cheap deteriorated private rented housing. As the immigrant population grew and many of these neighbourhoods underwent urban renewal, immigrants settled more and more in neighbourhoods in suburbs with social housing. Figure 2 shows the development in the number of neighbourhoods with more than 20 % residents from non-western countries. Partly due to the increasing number of immigrants and their descendants, the proportion of ethnic minorities increased much in some neighbourhoods. The number of neighbourhoods with more than 20 % ethnic minorities from non-western countries increased from 60 in 1985 to nearly 900 in 2008 (out of 9000 neighbourhoods). It can be seen that the fastest growth took place in the 1990s. After 2000, it slowed down, especially after 2005. In 1985, only half of the dwellings in these multi-ethnic neighbourhoods were social housing. In 2008, it was 70 %. However, it must be expected that the actual concentration in social housing estates was higher because some of the neighbourhoods could contain other tenures with fewer ethnic minorities. The proportion of non-western ethnic

Spatial assimilation? The development in immigrants 1000 900 800 700 600 500 400 300 200 100 0 Number of multi-ethnic neighbourhoods 1985 1990 1995 2000 2005 2008 Prop. of n-w minorities (%) > 20 > 30 > 40 > 50 > 60 Fig. 2 The development in the number of multi-ethnic neighbourhoods with more than 20 % residents being non-western ethnic minorities (immigrants? descendants). Source: Database on the population in Denmark 1985 2008 Proportion (%) of newly arrived settled in neighbourhoods with more than 20 % ethnic minorities 70 Alone 60 Single with children 50 Couple without children 40 30 20 10 0-1994 1995-96 1997-2004 2005- Couple with children Reunification with parents Reunification with immigrant Reunification with Dane All Fig. 3 Proportion of newly arrived immigrants, divided after mode of arrival, who settled directly in multiethnic neighbourhoods with more than 20 % ethnic minorities from non-western countries in four time periods minorities living in multi-ethnic neighbourhoods grew very rapidly in the 1990s but stagnated after 2000. Actually, there was a small decrease from 2005 to 2008. 8.1 Settlement of newly arrived immigrants Figure 3 shows the proportion of immigrants in different time periods, who settled directly in multi-ethnic neighbourhood just after their arrival to Denmark. For all immigrants, there was an increase in the proportion settled in multi-ethnic neighbourhoods with more than 20 % ethnic minorities from 15 % in the first period until 1994 to 22 % in the period 1997 2004. After 2004, it declined somewhat. Immigrants with family reunification much more often moved directly into multi-ethnic neighbourhoods, especially those who were reunified with parents. Among all new

H. S. Andersen immigrants without family reunification, only a small proportion settled directly in multiethnic neighbourhoods and it was a rather stable proportion over the years, which is remarkable as the number of these neighbourhoods increased much. This development might be ascribed to the change in the composition of immigrants over the years from asylum seekers and family reunification in the early 1990s to labour immigration after 2000. But also urban policies spreading asylum seekers and regulating access to multiethnic neighbourhoods might have been important. As documented in the next section, however, immigrants often settled in multi-ethnic neighbourhoods in the years after arrival. 8.2 Connection with duration of stay in Denmark As more and more neighbourhoods over time exceeded the chosen limit of 20 %, it might be expected that the proportion of immigrants living in such neighbourhoods would be steadily increasing with the duration of their stay in Denmark. To control for differences between periods of time, and for other background factors, some logistic regressions were made in the same way as those made above for living in social housing. As for social housing, statistical models were constructed in order to examine what background factors had been important for whether non-western immigrants had settled in multi-ethnic neighbourhoods with more the 20 % non-western ethnic minorities. Here it is especially important that the models control for differences between time periods, where the number of multi-ethnic neighbourhoods differs. In the model, the following control variables were used like in the analyses of settlement in social housing: Age: 17 29, 30 44, 45 59, 60 Sex Family situation: single without children, single with children, couple without children, couple with children, complex households Size of family: number of persons Employment situation: on pension, on welfare, on unemployment benefit, student, employed Household income: grouping in national income quintiles for the current year Urbanisation: the capital region, three largest provincial cities, other cities larger than 10,000 inhabitants, other. Time period: 1985 1987, 1988 1992, 1993 1997, 1998 2002, 2003 2006, 2007 2008 Country group: Eastern Europe, Middle East and North Africa, Other Africa and Other Asia From refugee country? Situation on arrival: Arrived alone, arrived as single parent, arrived as couple without children, arrived as couple with children, granted family reunification with parents, family reunification with another immigrant, family reunification with a Dane, arrived with parents. The increase in immigrants presence in multi-ethnic neighbourhoods during the first 10 15 years might be a result of two different developments. It might be because immigrants moved to multi-ethnic neighbourhoods during those years, or it might be because they stayed in a neighbourhood where the proportion of immigrants was increasing. We examined to what extent the development was a result of moves by examining only those immigrants who had moved within the year before. We therefore constructed five models.