Housing trajectories of EU migrants: between quick emigration and shared housing as temporary and long-term solutions

Abstract

Over the past two decades, many European countries have witnessed new immigration patterns related to the gradual expansion of the European Union (EU). While migration motives and labour market positions of EU migrants are well-understood, relatively little is known about their housing positions in the hosting countries. Using sequence analyses and logistic regression on longitudinal register data from Statistics Netherlands, this article examines housing trajectories of EU migrants from seven countries in the Netherlands, over an eight-year period (2012–2019). Our results show that, while housing trajectories vary substantially in terms of length of stay in the Netherlands and access to social housing, private renting and homeownership, sharing is at the centre for all migrant groups, both as a temporary and long-term solution. Moreover, we show that varying housing trajectories can partially be explained through contrasting demographic and socio-economic profiles. Yet, even after controlling for such factors as income, age, and household composition, some differences between country of origin persist.

Keywords:

• Housing trajectories
• shared housing
• EU migration
• sequence analysis
• housing careers
• the Netherlands

Introduction

Many European countries have witnessed a rise in new and more differentiated immigration, over the past two decades, through the gradual expansion of the European Union (EU) and the right to freedom of movement within the EU (Trenz & Triandafyllidou, 2017; Engbersen & Snel, 2013). The EU expansion towards Central and Eastern European countries since 2004, in particular, as well as the Euro Crisis at the turn of the last decade have sparked more complex and dynamic intra-European migration processes. Amongst other things, this has included work-related migration from Central-Eastern Europe to Western and Northern Europe. It also led to the renewal of South to North migration, particularly of younger people looking for better employment prospects, new opportunities in higher education, or simply to be able to establish residential independence from parents or experience new lifestyles (eg, Benson & O’Reilly, 2016; Maslova & King, 2020). The rise of intra-European migration has been staggering, in some cases. In 2005, for instance, about 25,000 immigrants from EU and European Free Trade Association countries moved to the Netherlands. By the turn of the last decade, this number had increased to about 70,000, and in 2019, it was close to 130,000 (Statistics Netherlands, 2020).

Most researchers studying these intra-European migration patterns focus on questions of spatial migration flows (where from, where to), immigrants’ labour market outcomes, migration and (return) migration motives, segregation, as well as immigrant integration processes more generally (eg, Bolt et al. 2008; Magnusson Turner & Hedman, 2014; Saar & Saar, 2020). Housing tenure positions and trajectories for these new immigrant groups, however, receive less attention in the literature.¹ This is noteworthy for at least two reasons: First, adequate housing has long been understood as an integral part of people’s willingness to invest in a new country or as a precondition for integration (eg, Nygaard, 2011; Zorlu et al., 2014). Second, social inclusion and economic success can hardly be achieved under poor housing conditions (Abramsson et al., 2002; Magnusson Turner & Hedman, 2014; OECD, 2018), while access to good quality housing may be stymied by failing integration processes. Indeed, numerous empirical studies have laid bare the often-precarious housing positions and trajectories of non-European immigrants and refugees, which include unhealthy housing conditions, residential segregation in disadvantaged neighbourhoods and restricted access to homeownership (eg, Borjas, 2002; Christophers & O’Sullivan, 2019; Finn & Mayock, 2022; Firang, 2019; Sinning, 2010; Usman et al. 2021; Uunk, 2017).

In parts, this lack of attention for intra-European migrants might be related to an implicit assumption about the more generous legal migration framework and, therefore, more equitable housing outcomes when moving between countries inside the European Union. However, cultural norms and economic standards vary substantially between EU Member States, meaning that the opportunities and outcomes on the housing markets in the hosting countries are likely to also vary significantly.

A few studies shed light on the role of shared housing amongst migrants as an option to cope with new emerging challenges, such as affordability, housing shortages, rising housing costs, uncertain labour market positions and more (eg, Baqai & Ward, 2020; Balampanidis, 2020; Lombard, 2021; Sawert, 2020; Smith, 2015). Shared housing, traditionally, has also been important amongst (international) students (Fang & Van Liempt, 2021) young adults (Arundel & Ronald, 2016; Heath et al., 2018), singles (Druta et al., 2021) or young professionals (Bobek et al., 2021), but also amongst low paid migrants (Lombard, 2021) or undocumented migrants (Balampanidis, 2020). It is often considered an ‘in-between phase’ and is associated with negative aspects, such as lack of privacy, insecurity or unhealthy living conditions. However, there are also positive effects, such as security, social stability in the face of loneliness, having meaningful relationships on a day-to-day basis or mitigating environmental impact (Druta et al., 2021). There appears to be much variation in shared housing within Europe. It is a relatively large sector in the United Kingdom and some Southern European countries, and a smaller sector in, for example, the Netherlands and Germany (Arundel & Ronald, 2016). The literature has hinted at sharing being particularly important for international migrants, but there is little knowledge on which types of migrants do so and whether shared housing is used as a ‘landing site’, a transitory tenure, or a long-term housing solution.

Our study addresses these gaps in the literature by exploring the variegated housing trajectories² of intra-EU migrants in the Netherlands. We particularly investigated (i) whether there is a variety in certain housing tenure outcomes amongst migrants from the same country of origin, and looked at (ii) the importance of shared housing in shaping post-migration housing trajectories. In this article, individuals are considered to live in shared housing when they live at a certain address with other non-related individuals. In contrast to more traditional approaches in which migrant housing tenures—mostly of a single migrant group—are compared to the native population, our study seeks to explicate and explain differences between and within migrant groups.

Using longitudinal register data from Statistics Netherlands, we followed the housing trajectories of seven EU migrant groups (Polish, German, French, Italian, Spanish, Bulgarian, British), over the period from 2012 to 2019. We applied sequence analysis to map variations herein, differentiating between moves to and between shared residences or solo rented social housing, privately rented housing, homeownership and return migration.³ We use the term solo rent or homeownership for household who do not share an address with other non-related persons. This analysis leads to the definition of eight clusters of ‘housing trajectories’, such as, quick return migration, long-term shared residence and transitions to homeownership. In a second step, we applied a series of logistic regression analyses to explore how the migrants’ nationalities and their demographic and socio-economic profiles tie in with these eight clusters.

In the remainder of this article, our research is embedded in the international literature on migration and housing, further explaining our empirical approach, presenting the various empirical results, and this article ends with a discussion on the broader societal and political implications of these findings.

Prior research

Most studies were conducted in countries with long histories of massive inward migration: Australia, Canada, and the United States, often focussing on the homeownership gap between migrants and natives (eg, Boehm & Schlottmann, 2009; Firang, 2019; Hiebert, 2009). Similar literature exists on Europe (eg, Abramsson et al., 2002; Christophers & O’Sullivan, 2019; Constant et al., 2009; Nygaard, 2011; Sinning 2010; Uunk, 2017; Vono-de-Vilhena & Bayona-Carrasco, 2012). The results on progression in housing careers are not always clear, with some studies showing improving housing conditions on the side of the migrant population, while other studies report little progress and persisting gaps and inequalities between migrants and natives or between migrant groups. This inconsistency is probably partly due to variations in when, where and which migrants were studied. Uunk (2017) highlights the different explanations for this gap, varying from for instance financial constraints, housing market constraints and ethnic discrimination to varying ethnic homeownership preferences.

For the Netherlands, specifically, there is a large body of research that investigates the housing positions and careers of non-Western migrants who entered the country some decades ago (Bolt et al. 2008; Groot et al., 2013; Kullberg & Kulu-Glasgow, 2009; Özüekren & Van Kempen, 2003; Uunk, 2017; Zorlu et al., 2014). Similar to other country cases, most of these studies show a large and persistent gap in rent and homeownership. As such, housing trajectories of migrants in the Netherlands may prove to be a good example of the situation in other European countries.

In aforementioned research, a very common underlying assumption is that of homeownership being at the top of the linear housing ladder and renting ranking lower (thus, looking into the gap between renting and homeownership). This is because homeownership is associated with wealth accumulation, a higher level of social well-being of individuals or with more financial and tax benefits relative to renting and homeownership. Amongst migrants, it can also be a sign of further integration or of willingness to commit to the hosting country (eg, Zorlu et al., 2014). As such, the housing career is seen as a linear trajectory, going from one housing type to another, with progressively more quality and/or progressive steps from rent to ownership (eg, Lennartz & Helbrecht, 2018; Zorlu et al., 2014). A new research field is emerging showing that migrants have nonlinear housing trajectories with extended or multiple periods of sharing a home with non-related others (Balampanidis, 2020; Lombard, 2021; Smith, 2015). Although, in the Netherlands, the sector is not as large as in the United Kingdom (Heath et al., 2018), it is fast growing. This article adds new insights into this new field by studying shared housing amongst migrants.

Within Europe, there are large differences in housing prices, quality and the privately owned housing stock (Housing Europe, 2021). Relative to other European countries, the Dutch housing system used to have a highly accessible, high quality and affordable social housing sector. Also, it used to have a highly financialized mortgage market and strongly subsidised owner-occupied housing, which allowed for first-time homeownership at a relatively young age. Recent years have seen a transformation of the housing market towards a more expensive private rental sector, due to rapidly increasing house prices, more restricted access to mortgage credit and a move towards a more restricted social housing sector (Hochstenbach & Boterman, 2015). Accordingly, particularly younger households, starters on the housing market and housing market outsiders more generally have moved increasingly into the private rental sector (Arundel & Lennartz, 2020), with shared housing also becoming more common amongst those groups (Arundel & Ronald, 2016; Druta et al., 2021). In many ways, we expect recent migrants arriving in the Netherlands to face similar limitations to enter and move within the housing market and towards solo rent or homeownership.

Finally, immigrant groups not only differ in terms of their socio-economic profiles but also due to other factors, such as discrimination, degree of attachment to the home country or hosting country, ethnic-specific housing preferences (Zorlu et al., 2014). They also come from different housing cultures and might have different expectations and aspirations in terms of their post-migration living arrangements. While some housing systems, such as in the United Kingdom, Eastern and Southern European countries or, for that matter, the Netherlands itself, are strongly geared towards ‘the primacy of homeownership’, ‘renting for life’ is widely accepted across broader strata of the population in other countries, such as in Germany (Lennartz et al., 2016). We would expect these differences in housing cultures between different EU nationalities to shine through their respective housing trajectories in the Netherlands, as well.

Data & empirical strategy

Data

To study intra-EU migrant housing trajectories, we used data from the System of Social Statistical Datasets (SSD) of Statistics Netherlands (Bakker, et al., 2014). The core of the SSD is the Personal Records Database, which is maintained by municipalities. This register contains the residential address and demographic characteristics (eg, gender, age, country of birth, family status) of every legal inhabitant of the Netherlands since 1995. All immigrants who intend to stay in the Netherlands for more than 4 months are legally obliged to register themselves in this register. In addition, from 2012 onwards, the Netherlands has a comprehensive register on buildings and dwellings. Information from this register was linked to the SSD using encrypted address keys.

Our study population was the migrant cohort with differing countries of birth (Polish, German, Italian, French, Bulgarian, Spanish and English) who moved to the Netherlands in 2011. Migrants who had left the Netherlands before 1 January 2012 were excluded. Furthermore, the analysis was restricted to immigrants aged 18 and older who were not living with their parents at the time of arrival in the Netherlands. This left us with a research population of 36,020 immigrants. At the time of the study, the necessary data were available up to 1 January 2019. We focussed on seven countries to provide as much variation as possible, while keeping the data analysis and interpretation of the results manageable. Furthermore, these specific countries were selected on the basis of the size of the immigrant population and a broad but unspecified reflection of different socio-economic and cultural standards across and within Southern, Western, and Central/Eastern European countries. Using register data for migrant studies has many advantages such as a large sample size or lack of selective panel attrition. Yet, there is no information on stated residential preferences as would be found in surveys or with qualitative interviews.

Sequence analysis

In our first analytical step, we deployed sequence analysis (SA) to study housing trajectories for migrant groups, as a whole. Due to memory restrictions, the SA was carried out on a 33% random sample of the selected research population (N = 11,941)—see Appendix 1 for a comparison with the full sample on various demographic characteristics). In SA, each life-course trajectory is represented by a string of characters that refer to a specific state. We distinguished between five possible states: (i) owner-occupied, (ii) social rent, (iii) private rent or (iv) shared accommodation with other non-related persons household(s), and (v) return migration. Each immigrant was followed for a period of eight years on an annual basis, from 2012 to 2019. Due to the large number of possible combinations of these eight years of observation and the five states, only a limited number of individuals experienced the exact same sequence of states, implying the presence of many different sequences in the data set.

First, we computed optimal matching (OM) distances between all pairs of sequences, using the TraMineR package in R (Gabadinho et al., 2011). The OM algorithm measures dissimilarities between two sequences by considering how much it ‘costs’ to transform one sequence into the other. There were three operations available: insertion, deletion, and substitution (Abbott and Tsay, 2000). A cost was attached to each operation. We followed the commonly applied solution of using unitary insertion/deletion costs and the inverse of the transition rates to define substitution costs (eg, Kleinepier et al., 2015). After the OM distances were calculated, we identified more-or-less homogeneous groups by applying Ward’s cluster analysis (Ward, 1963). This is an agglomerative hierarchical clustering method, meaning that each observation starts in its own cluster, and pairs of clusters are merged as one moves up the hierarchy (bottom-up approach). After comparing various cluster solutions (2–10 clusters) in terms of their theoretical meaningfulness, we decided on an 8-cluster typology exhibiting the variety in housing trajectories (see Table 1).

Table 1. Housing trajectories of migrants in 8 clusters, 2012–2019.

	Cluster	N	%
Private renting to return migration	1	1.317	11
Transitions into social housing	2	1.201	10
From sharing to private renting	3	1.258	10
Quick return migration	4	3.858	32
Transition into homeownership	5	879	7
Long-term sharing	6	1.279	11
From sharing to return migration	7	1.496	13
Long-term homeownership	8	653	6
Total		11.941	100

Logistic regression

In a second analytical step, we studied the determinants of cluster membership (ie, country of origin, socio-economic and demographic characteristics and deduced migration motives). We ran a series of separate binary logistic regression models, using each of the clusters as the outcome variable. We opted for this approach rather than for a set of multinomial regression models, to make it easier to interpret the differences between EU migrant groups. These regression analyses included time-constant independent variables only. The following independent variables were included in the models.

Country of origin refers to the country of birth and was measured with dummy variables. This study includes migrants from France, Germany, Italy, Spain, Bulgaria, Poland (reference group) and the United Kingdom.

Gender is a dummy variable, with female as the reference group.⁴

Age at immigration was measured in 2012 and operationalized as a four-way categorical, distinguishing between 18–25-year-olds (reference category); 26–30-year-olds; 31–40-year-olds; and those aged 41 and older.

Partnership status at immigration measured whether an individual moved to the Netherlands without a partner, with a partner, or moved in with a Dutch or a migrant partner (reference category). Having a Dutch partner may not only provide greater means, opportunities or information on the Dutch housing market, but will often also lead to a stronger urge to stay.

Standardised disposable household income was measured in quintiles with the lowest (first) as reference category. Due to the fact that, for a relatively large group, household income had not been recorded in 2012, we used a sixth income group in the model: ‘income unknown’. Due to the highly income-structured housing market, housing trajectories vary with income.

Migration motive was operationalised as a four-way categorical variable, distinguishing family, work (reference category), study, and other/unknown. Statistics Netherlands does not record stated migration motives of EU/EFTA migrants but deduces them from what migrants actually do after arrival (Statistics Netherlands, 2021). A family-related migration was deduced when an immigrant moves to the Netherlands within 120 days after their partner or parents have done so. If a migrant had a partner or parent who had been living in the Netherlands for more than 120 days, they were always defined as a family migrant. A work-related motive was deduced when a migrant’s main income was generated from work in the first 120 days after their arrival. If migrants had moved to the Netherlands as a couple, one of the adults was defined as a work migrant and the other as a family migrant, depending on who had first started working. An education-related migration motive was deduced when a migrant had started a higher education study within 365 days after arrival. Given that students have different housing suppliers than labour migrants and that family migrants more often move in with family members already residing in the Netherlands, housing trajectories may vary with migration motives.

Table 2 shows the distribution of these variables across the whole migrant population, but also reveals in what way immigrants from the seven EU countries differ from each other.

Table 2. Socio-economic and demographic profiles of EU migrants in the Netherlands, 2011, in percentages of total, per country of origin.

Country of origin		France	Germany	Spain	Italy	Poland	Bulgaria	UK	Total
N		660	2460	700	730	5160	1260	980	11,910
Migration motive (%)	Family	25	16	26	25	34	24	30	31
	Work	37	16	32	40	41	6	40	28
	Study	15	52	9	16	2	15	12	16
	Other	23	16	33	19	23	55	19	25
Sex (%)	Female	48	56	51	35	46	46	38	47
Age (%)	18–25	60	67	52	40	33	35	33	43
	26–30	18	12	21	26	28	18	16	22
	31–40	15	11	21	24	24	28	25	21
	>40	7	10	7	10	15	19	26	14
Partnership status upon arrival (%)	No partner	67	73	66	65	59	72	57	64
	Dutch partner	6	8	9	7	3	3	13	6
	Foreign partner	27	19	25	28	38	26	30	30
Mean standardized household income (€)		22.996	16.894	20.174	19.427	15.426	10.948	28.225	17.388

Results - Eight clusters of housing trajectories for EU immigrants in The Netherlands

Figure 1 shows the distribution of the various clusters of housing trajectories for migrants from the seven EU countries⁵ who arrived in the Netherlands in 2011, over the 2012–2019 period. It highlights the variegated ways of how migrants progress through the Dutch housing market and how common specific tenure changes and housing outcomes are.

Figure 1. Housing trajectory clusters of EU immigrants in the Netherlands.⁸

The first more general observation is that more than half of all EU migrants stay in the Netherlands only for a short while—a couple of years only (sum of clusters 1, 4 & 7). More precisely, the clustering procedure revealed that the most common trajectory is the ‘quick return migration’ cluster (4), representing 32% of the total sample. Cluster 7 also represents short stayers in shared housing, but for a slightly extended period (‘From sharing to return migration’—13%). Finally, 11% of all migrants remained in the Netherlands for only a couple of years, but did so as solo household in privately rented accommodation (Cluster 1—’From PRS (private renting sector) to return migration’). The large proportion of temporary migrants highlights that the legal migration framework in the European Union facilitates migrants to move across country borders temporarily, for example, to gain work experience abroad, study, or to just get to know a new culture and lifestyle, while other migrants move on to other countries or go back home, in search for jobs or opportunities elsewhere.

The second observation concerns the crucial role of shared housing in almost all housing trajectories. Looking at the housing position of migrants shortly after arrival and irrespective of trajectory cluster, (excluded in Figure 1) we noted that 62% of all migrants would start on the Dutch housing market in shared residence. Renting in the private sector as solo household was found to also be common amongst the migrant population, with 25% of all migrants doing so upon arrival. Only 5% of all migrants started as homeowners, and only 8% gained access to the social housing sector⁶ for their first type of housing.

It is often assumed that migrants more often share a home with non-related households when they initially arrive in the Netherlands and are still unfamiliar with the Dutch housing system. Figure 1, however, shows that sharing was not only common amongst short stayers but was also frequently used as long-term form of accommodation (see Cluster 6, where ‘long-term sharing’ amounts to 11% of all migrants). This challenges the notion of sharing being an ‘in-between phase’. Sharing was found to function equally often as a gateway to other forms of tenure. In Cluster 2 ‘Transitions into social housing’, Cluster 3 ‘From sharing to private renting (PRS)’, and Cluster 5 ‘Transitions into homeownership’, about half of all migrants were found to live in shared accommodation, during their first year. A final observation was the relatively low share of migrant households who moved to homeownership over the course of their stay in the Netherlands—see Cluster 5 ‘Transitions into homeownership’ (7%) and Cluster 8 ‘Long-term homeownership’ (6%). In part, this is due to the young age profile of the migrant group. However, it also reflects the gap between renting and homeownership amongst migrant groups vis-à-vis the native population more generally. Due to the transitory nature of their stay in the Netherlands, homeownership appeared to be too strong a commitment, and one that only a relatively small groups of migrants was willing and able to take on.

Results—exploring housing trajectories for various groups of EU migrants

How do various migrant groups from the seven EU countries progress through the Dutch housing market? Figure 2 illustrates the differences between country of origin by cluster affiliation. First of all, we saw that, for all nationalities, the most likely outcome was that they would leave the Netherlands after a relatively short period of time (Cluster 4 ‘Quick return migration’) — with particularly Spanish and French immigrants being classified as short stay migrants. The relatively large share of German immigrants who first move into shared accommodation and then leave the country within the eight-year observation period can most likely be explained by the large share of students amongst them. British migrants are equally likely to leave the country again, but they do so more often after having lived as a solo household in the private rental sector.

Figure 2. Housing trajectory clusters of migrants, in percentages per country of birth.⁹

Generally, we could see those immigrants from Western and Southern Europe, when compared to Polish and Bulgarian migrants, were less likely to settle for a longer period in the Netherlands. One explanation could be that their migration is motivated more often by lifestyle reasons or, in the case of Southern European immigrants, they were motivated to leave their home country, temporarily, due to the social and economic limitations in the post-financial and Euro-crisis period (see also Maslova & King, 2020; Lafleur & Stanek, 2017). In contrast, Polish and Bulgarian immigrants seemed more likely to move to the Netherlands in pursuit of a higher income and possibly higher living standards, compared to what they could achieve in their home country.

The position of Polish immigrants was found to be quite remarkable in this regard. They were much more likely than the other six immigrant groups to stay in the Netherlands more permanently, resulting in more diverse housing trajectories, overall. In part, Polish migrants more frequently remained living in long-term sharing—only being surpassed by Bulgarians—which most likely reflects on the relatively large share of lower-skilled migrants working in the agricultural and construction sector in the Netherlands (Gijsberts & Lubbers, 2015). However, a significant share of Polish migrants (Cluster 5 + Cluster 8 = 16%) eventually moves into homeownership; only the Italian and British migrant groups are more ‘successful’, in that regard, where the latter might be the result of an older age profile, greater economic resources, and the fact that they more often have a Dutch partner.

Most strikingly, however, Polish migrants are much more likely than any other group to enter the highly restricted social housing sector in the Netherlands—more than 17% of all Polish migrants are found in Cluster 2 ‘Transition into social housing’. This certainly reflects on their weaker income position, as compared to Western and Southern European migrants, but it may also be the result of a larger share of family migrants. The fact that they are more successful than Bulgarian households, who are even more often in precarious working conditions and have lower incomes (Gijsberts & Lubbers, 2015), might be an indication of a more established and well-connected Polish community, which is more versed in navigating the Dutch housing market, including the social housing sector (see also Kleinepier et al., 2015).

Results—the role of households, and socio-economic and migration-specific factors via logistic regression

So far, we discussed the differences between the seven migrant groups with regard to their varying socio-economic and demographic profiles and migration motives. But do the various housing outcomes uphold once we control for these characteristics? Or put differently, is there a disposition towards specific housing trajectories that can be linked to country of origin, cultural norms and, thus, the housing preferences of migrating households?

Figure 3 shows the outcomes of the logistic regression models for the eight clusters. The results are constrained to the chances of migrants following a specific housing trajectory in comparison to the reference group (ie, Polish migrants), controlling for income, gender, age, partnership status upon migration, and migration motive—see Appendix 2 for full regression results for all eight models.

Figure 3. Forest plots with odds of following a specific housing trajectory by country of origin (Polish migrants are reference group).¹⁰

In the first place, in general, even after controlling for individual characteristics, we found that most patterns hold across the eight clusters of housing trajectories. Except for Bulgarians, Polish migrants are the least likely to return to their country of origin or to re-migrate to a third country, supporting the notion that Central and Eastern European migrants have a stronger intention to settle for at least eight years in the Netherlands—amongst other things, for a higher standard of living and better job opportunities, compared to those in their home country.

Second, Polish migrants are still more likely to enter rented social housing than any other migrant group. Again, this supports the idea of better support networks within the Polish community, a larger Polish population with a longer migration history and, thus, a larger share of Polish households already residing in social housing, but probably also lower economic resources throughout the whole eight-year period and not only in the first year after arrival. Bulgarian migrants, on the other hand, are not able to translate their lower income into access to the social housing sector. More than for any other group, when living as a solo household, this results in a much greater likelihood of entering the private rental sector. Other explanations include that Bulgarian migrants have a shorter migration history and a much smaller share of other Bulgarians already living in the Netherlands, and that, in 2011, Bulgarian immigration was restricted by the Dutch Government.⁷

Third, looking at housing trajectories in homeownership, we found that, except for the Bulgarian groups, there were no differences between country of origin in Cluster 8 ‘long-term homeownership’—with the Spanish group being slightly less likely to start in the owner-occupied housing sector. Moreover, the results for Cluster 5 ‘Transitions into homeownership’ indicate a stronger urge amongst Italian migrants to move into homeownership, while German migrants are less likely to buy property in the Netherlands. Interestingly, the larger share of British migrants disappeared once we controlled for their greater economic resources. Generally, these patterns could indicate at least some disposition towards specific tenure outcomes; yet, we interpreted this as only a weak indication, since homeownership is the preferred tenure in Spain and the United Kingdom. On the other side of the spectrum, however, shared housing is especially common amongst Bulgarian and German migrants. This also remained true after taking into account that many German migrants are students or that many Bulgarian migrants have a relatively low income. A final observation was that country of origin certainly does not explain housing trajectories alone. While it was beyond the scope of this article to discuss all independent variables, in depth, it does highlight certain patterns. Different forms of return migration (Clusters 1, 4 and 7) are strongly influenced by income, partnership status at the time of immigration, and migration motive. Higher income households, individuals who came to the Netherlands as a single person household and those migrating for educational reasons are all strongly associated with return migration. Similarly, remaining in shared accommodation throughout the eight-year observation period (Cluster 6) was found to be strongly associated with moving to the Netherlands for higher educational purposes and doing so as a single person. Starting homeownership (Cluster 8) is correlated with having a higher income upon migration, being a family-migrant and moving in with a Dutch partner, while moving into homeownership (Cluster 5) is more strongly predicted by a higher income and moving at a younger age—this last only when the move to the Netherlands is not education-related.

Conclusions and discussion

This article started from the premise that, compared to more traditional migrant groups, there is little understanding about and empirical evidence of the housing positions of intra-EU migrants in the hosting countries and how these positions may evolve over time. We applied sequence analysis and logistic regression to empirically demonstrate the divers’ nature of migrants’ housing trajectories in the Netherlands. We particularly aimed to show in what way socio-economic and demographic profiles, but also country of origin matter for these variegated housing patterns. Also, we specifically brought forward the role of shared housing in the newly emerging housing trajectories of migrants.

Our results showed a multitude of potential trajectories amongst EU migrants who came to the Netherlands at the turn of the previous decade. Most strikingly, the majority of migrants under study did not stay long in the Netherlands but had returned to their home country or re-migrated to another country between 2012 and 2019. In line with what is described in other studies, intra-EU migration, thus, has a strongly transitory nature (Engbersen & Snel, 2013) and appears to be as much about educational enrolment and lifestyle choices as improved economic opportunities. However, our study also revealed important differences between migrant groups, with migrants from Central and Eastern Europe seemingly more susceptible to economic incentives than those from Western and Southern Europe. Not only does this translate into larger shares of long-stayers in the Netherlands, but it also induces specific patterns of housing consumption amongst Central and Eastern European communities, such as higher sharing raters amongst Bulgarian migrants and higher social housing rates amongst those from Poland. Herein, migrants seem to be primarily influenced by the current housing market context in the Netherlands; experiences made in the housing system of the home country seem to play an inferior role. For instance, migrants from the UK, a homeowner friendly country, do not enter homeownership in the Netherlands more often than other migrant groups, after controlling for economic resources. Here, in line with other studies (eg, Zorlu et al., 2014), unspecified cross-country differentials remain controlling for relevant socio-economic and migration-specific factors

Another key finding concerns the crucial position of shared accommodation in defining housing trajectories. Relative to the literature on homeownership among migrants (eg, Christophers & O’Sullivan 2019; Sinning, 2010; Vono-de-Vilhena & Bayona-Carrasco, 2012; Zorlu et al., 2014), shared housing is under-studied, but the body of literature is increasing (eg, Balampanidis, 2020; Lombard, 2021; Sawert, 2020; Smith, 2015). On the one hand, we find it was most often used by new migrants as their gateway to the Dutch housing market. Given the reduced costs and the fact that employers, recruitment agencies and higher education institutions are often directly involved in organising shared accommodation for newcomers, this is a rather expected finding. On the other hand, our analysis also showed that a sizeable group of EU migrants, tended to remain in shared housing on a more long-term basis. Some migrants were using sharing housing not only as a landing site or a transitory tenure, but also as along-stay housing solution. This is most likely related to their more precarious labour market positions. However, it could also be a reflection of more recent systemic shifts in the Dutch housing market with owner-occupation becoming highly unaffordable, access to the social housing sector becoming more restricted due to regulatory changes and insufficient new supply, and private renting, particularly in shared accommodation, having become the default tenure amongst younger people and housing starters more generally (Hochstenbach & Boterman, 2015). Hence, our results challenge the notion that shared housing is an ‘in-between phase’ and can very well become a long-term housing solution, even if an undesired one, particularly for the migrant population.

We would argue that these findings have important implications for European migrants themselves, but also for society and policymakers in the Netherlands. First of all, the more transitory nature of intra-EU migration means that some neighbourhoods will experience more rapid population turnover and more overcrowding. This may pose new challenges to the liveability and social cohesion in neighbourhoods with a high rate of immigration from the EU Member States. Second, a larger degree of shared housing amongst those who intend to stay long may lead to more stymied integration processes. In line with the literature, above, we argued that integration and good and stable housing are strongly intertwined. If EU migrants are increasingly trapped in unstable and often unhealthy shared accommodation for extended periods of time, they would thus largely be at risk of failed integration processes. This, in turn, might lead to increased resentment amongst the native population towards intra-European migration and rights to free-movement. Here, our study shows that particularly the Bulgarian migrant community, as well as parts of the Polish community, might fall victim to these developments. However, such conclusions are definitely in need of more research on shared housing to increase the understanding about whether sharing is indeed of an undesired nature and migrants get stuck in such circumstances involuntarily, or whether they feel more positive about their sharing experiences. Also, there is a need for more insight into other relevant factors in post-migration housing trajectories, such as housing cultures in hosting or sending countries, people’s level of attachment to their home or hosting country, return intentions, ethnic-specific housing preferences and the role of remittance. These factors may hinder housing investment in the hosting countries (Skovgaard Nielsen, 2017). Third, and connected to this, the Dutch labour market is highly reliant on working migrants, both in lower- and higher-skilled professions. In this study, we did not measure whether the housing conditions of different migrant cohorts improved and/or deteriorated, over time. However, if a growing group of international migrants is subject to poor and increasingly unaffordable housing conditions, the Netherlands would become an increasingly unattractive migration destination within the European Union, with important negative consequences for the Dutch labour market and economy, as a whole. Here, the need to provide attractive and stable housing for the sake of migrants’ well-being as well as their improved social and economic development poses a challenge to policymakers in the Netherlands, but is a topic that needs to be addressed, nonetheless. As such, the search for migrant workers to overcome labour market shortages is an issue that many EU Member States are confronted with, and housing quality is one of the factors to may help to attract such migrants. This study showed that, while housing trajectories vary substantially in terms of length of stay in the Netherlands and access to social housing, private renting and homeownership, sharing is at the centre for all migrant groups, both as a temporary and long-stay solution. But many other issues were not addressed in this study. First, register data do not allow us to analyse the role of individual’s residential preferences or experiences with shared housing. Nor does it allow us to investigate other relevant factors such as discrimination, remittance or the degree of attachment to home or hosting country. Third, future studies could apply life course analyses to shared housing, instead of to rent or homeownership more extensively.

Acknowledgements

We would like to thank the editor and anonymous referees for their valuable comments and constructive suggestions.

Disclosure statement

No conflict of interest has been declared by the authors.

Notes

1 Some notable exceptions are: Balampanidis, 2020; Finn & Mayock, 2022; Lombard, 2021; Smith, 2015).

2 We use the term ‘housing trajectories’ to describe the different stages which households progress in 2011-2019. We relate this to tenure position (social or private renting vs. owning) but also whether a household lives without another household at one address alone, and whether the migrant stays in or leaves the country. Similar to the contemporary conceptualization of the term ‘housing career’ (Arundel & Lennartz, 2020) and the alternative concept of a ‘housing pathway’ (Clapham, 2002) there is no predefined path towards a final stage, which in practice would often be assumed to be homeownership. Instead, these notions will be used interchangeably; however, for the sake of clarity we will use the notion of a ‘trajectory’ throughout the text.

3 This implies both return migration to the country of origin as well as migration to a third one.

4 So far, Statistics Netherlands only registers gender as a binary variable. We are thus bound to this classification.

5 During the period of observation, the United Kingdom was still an EU Member State, which is why we refer to it as an EU country rather than a former EU Member State.

6 Since access to social housing in the Netherlands is restricted by waiting lists and income limits, migrants who moved into rented social housing straight away must have moved in with a partner who was already living in social housing.

7 Up to 2015, Bulgarian immigration to the Netherlands was restricted to the self-employed and to employees with a work permit, despite the country’s EU membership. Some Bulgarian migrants worked here via informal networks or via Bulgarian recruitment agencies, which might explain the large number of unknown migration motives.

8 PRS = private rental sector

9 PRS = private rental sector; HO = homeownership

10 PRS = Private rental sector

Appendix 1.

Percentual distribution of variables in total and 33% sample.

Table

Country of origin	Total	33% sample
France	5,6%	5,5%
Germany	20,9%	20,6%
Spain	6,3%	5,9%
Italy	6,0%	6,1%
Poland	42,5%	43,2%
Bulgaria	10,5%	10,5%
United Kingdom	8,2%	8,2%
Total	100,0%	100,0%
Gender
Female	47,6%	47,1%
Male	52,4%	52,9%
Total	100,0%	100,0%
Age at immigration
18–25	43,6%	43,2%
26–30	21,5%	21,7%
31–40	20,6%	20,9%
41+	14,3%	14,1%
Migration motive
Family	27,4%	27,7%
Work	31,4%	31,3%
Study	16,3%	16,3%
Other	24,8%	24,7%
Total in %	100%	100%
Total N	36.020	11.941

Appendix 2.

Full regression results.

Table

	Cluster 1 - PRS to emigration			Cluster 2 - Sharing to SH			Cluster 3 - Sharing to PRS			Cluster 4 - Quick emigration
	B	S.E.	Sig.	B	S.E.	Sig.	B	S.E.	Sig.	B	S.E.	Sig.
Country of origin (Poland = ref)
Bulgaria	0,323	0,130	0,013	−0,424	0,128	0,001	0,627	0,103	0,000	−0,017	0,079	0,829
Germany	0,906	0,096	0,000	−0,971	0,125	0,000	−0,201	0,105	0,055	0,323	0,072	0,000
France	0,869	0,128	0,000	−1,796	0,276	0,000	−0,676	0,181	0,000	0,883	0,098	0,000
Spain	0,726	0,131	0,000	−1,254	0,204	0,000	−0,069	0,141	0,621	0,820	0,096	0,000
Italy	0,836	0,121	0,000	−1,407	0,207	0,000	−0,045	0,132	0,734	0,463	0,095	0,000
UK	0,924	0,108	0,000	−1,361	0,176	0,000	−0,257	0,128	0,045	0,683	0,087	0,000
Migration motive (work = ref)
Family	−0,150	0,089	0,093	0,128	0,092	0,162	0,073	0,091	0,426	−0,425	0,074	0,000
Education	0,329	0,116	0,005	−0,081	0,164	0,620	0,141	0,128	0,267	−1,377	0,083	0,000
Other	−0,160	0,102	0,115	−0,029	0,096	0,766	0,037	0,091	0,683	−0,252	0,063	0,000
Male (female = ref.)	0,066	0,062	0,291	−0,291	0,066	0,000	−0,082	0,062	0,188	0,218	0,045	0,000
Age at immigration (18–25 = ref)
26–30	0,122	0,089	0,171	0,145	0,088	0,098	0,129	0,085	0,127	−0,208	0,062	0,001
31–40	0,226	0,091	0,013	0,214	0,091	0,019	0,193	0,086	0,026	−0,381	0,065	0,000
>40	0,155	0,105	0,140	0,519	0,098	0,000	0,014	0,102	0,889	−0,214	0,072	0,003
Partnership status (foreign partner = ref)
Single	−0,696	0,080	0,000	−0,541	0,088	0,000	−0,174	0,084	0,037	0,548	0,065	0,000
Dutch partner	−0,767	0,134	0,000	0,089	0,141	0,526	−0,535	0,154	0,000	−0,069	0,125	0,583
Household income (1st quintile = ref)
2nd quintile	0,216	0,104	0,037	0,145	0,081	0,072	0,509	0,086	0,000	−1,232	0,079	0,000
3rd quintile	0,657	0,105	0,000	−0,098	0,102	0,340	0,291	0,106	0,006	−1,406	0,101	0,000
4th quintile	0,669	0,120	0,000	−0,426	0,144	0,003	0,403	0,126	0,001	−1,331	0,119	0,000
5th quintile	1,065	0,115	0,000	−1,490	0,252	0,000	0,358	0,139	0,010	−1,435	0,123	0,000
HH income unknown	−0,130	0,093	0,161	−1,246	0,126	0,000	−0,510	0,096	0,000	0,824	0,059	0,000
Constant	−2,476	0,116	0,000	−1,238	0,107	0,000	−2,124	0,111	0,000	−0,833	0,082	0,000
	Cluster 5 - Transition HO			Cluster 6 - Long-term sharing			Cluster 7 - Sharing to emigration			Cluster 8 - Long-term HO
	B	S.E.	Sig.	B	S.E.	Sig.	B	S.E.	Sig.	B	S.E.	Sig.
Country of origin (Poland = ref)
Bulgaria	−0,180	0,174	0,301	0,175	0,102	0,085	0,180	0,107	0,093	−1,656	0,352	0,000
Germany	−0,437	0,138	0,002	−0,534	0,104	0,000	0,341	0,092	0,000	0,175	0,133	0,190
France	0,067	0,160	0,676	−0,651	0,164	0,000	−0,139	0,146	0,343	−0,233	0,213	0,274
Spain	−0,155	0,164	0,344	−1,008	0,185	0,000	−0,017	0,140	0,905	−0,409	0,208	0,049
Italy	0,380	0,135	0,005	−0,806	0,160	0,000	−0,023	0,133	0,864	0,081	0,181	0,653
UK	−0,248	0,146	0,089	−0,772	0,145	0,000	0,071	0,123	0,566	0,062	0,146	0,674
Migration motive (work = ref)
Family	0,069	0,100	0,489	−0,069	0,108	0,527	−0,222	0,103	0,031	0,735	0,126	0,000
Education	−0,989	0,247	0,000	0,938	0,114	0,000	0,818	0,100	0,000	−1,658	0,526	0,002
Other	0,009	0,115	0,941	0,201	0,086	0,019	−0,126	0,088	0,153	0,634	0,160	0,000
Male (female = ref.)	−0,269	0,075	0,000	0,230	0,063	0,000	0,088	0,059	0,132	−0,489	0,092	0,000
Age at immigration (18–25 = ref)
26–30	0,316	0,094	0,001	−0,225	0,092	0,014	−0,051	0,084	0,538	0,467	0,133	0,000
31–40	−0,011	0,102	0,911	0,091	0,090	0,314	−0,099	0,091	0,275	0,653	0,132	0,000
>40	−0,692	0,142	0,000	0,220	0,098	0,024	−0,253	0,106	0,018	0,768	0,143	0,000
Partnership status (foreign partner = ref)
Single	−0,382	0,099	0,000	0,595	0,095	0,000	0,425	0,087	0,000	−0,836	0,137	0,000
Dutch partner	−0,047	0,141	0,738	−0,298	0,211	0,158	0,143	0,161	0,375	0,978	0,119	0,000
Household income (1st quintile = ref)
2nd quintile	0,903	0,103	0,000	0,274	0,091	0,003	−0,048	0,093	0,606	0,539	0,140	0,000
3rd quintile	1,218	0,111	0,000	−0,102	0,129	0,426	−0,390	0,126	0,002	1,083	0,138	0,000
4th quintile	0,977	0,136	0,000	0,308	0,143	0,031	−1,244	0,216	0,000	1,514	0,147	0,000
5th quintile	0,997	0,149	0,000	−0,232	0,192	0,226	−0,716	0,181	0,000	1,393	0,154	0,000
HH income unknown	−0,956	0,172	0,000	−0,218	0,084	0,009	−0,221	0,075	0,003	−0,739	0,231	0,001
Constant	−2,466	0,130	0,000	−2,614	0,122	0,000	−2,267	0,113	0,000	−3,655	0,178	0,000