From collective centres to private accommodation: housing trajectories of asylum migrants in Switzerland

Abstract

This paper examines the housing trajectories of complete cohorts of asylum migrants in Switzerland. It emphasizes the logistics of housing allocation by local authorities and how it shapes individual opportunity structure. We use full-population register data and analyse one key transition: the transition out of collective centre assigned upon arrival to the private and subsidized housing sectors. Event history models show the association between individual and contextual factors and the speed of access to private housing. Despite the quasi-autonomous management of refugee housing by region, priority rules regarding access to private housing were found to apply across the country. When choosing between different profiles, women, older asylum migrants, married individuals, and members of larger national groups are favoured in obtaining access to private housing. Nevertheless, the time spent in collective centres largely depends on the region to which a claimant is assigned, pointing to the minimal agency asylum migrants have during their first years of residence.

Keywords:

• Residential trajectories
• housing policies
• asylum seekers
• refugees

1. Introduction

Quality housing is both a determining factor and a marker for the integration of asylum migrants (Ager & Strang, 2008). The literature consistently shows that long stays in poor housing negatively impacts asylum seekers’ and refugees’ socioeconomic integration, health and well-being, while proper housing shapes their sense of security and belonging (Bakker et al., 2014; Bakker et al., 2016; Phillips, 2006; Ziersch et al., 2017; Ziersch & Due, 2018). In Switzerland, the housing situation of asylum migrants regularly attracts media attention, given the shortage of appropriate accommodation and the increased use of underground bunkers as a temporary solution (Del Biaggio & Rey, 2017).

Following the rules of the Dublin Regulation, the country examining the asylum claim is responsible for housing the asylum seekers during the procedure. Upon arrival to Switzerland, asylum seekers typically live isolated from the local population in an assigned federal reception centre, after which they are transferred to a cantonal housing facility, usually collective (communal) housing at first and then private (subsidized) housing. Accommodation in collective centres is a common experience for asylum migrants. On the one hand, these centres provide support and assistance, enabling refugees to access material aid, health care and different integration programs (language courses and employment programs). On the other hand, life in these centres is characterized by a climate of control, uncertainty, and lack of privacy. Access to employment or school can be particularly difficult for residents of collective centres, who may experience multiple transfers between reception structures and live isolated from urban centres with limited access to public transportation (Mottet, 2022). In fact, the management of the asylum process considerably affects integration outcomes (see Hainmueller et al., 2016 for causal evidence on the economic integration of refugees in Switzerland).

Researchers in Europe, North America and Australia have documented the great difficulties asylum seekers and refugees face when moving from collective centres to private and subsidized sectors of the housing market (Berger, 2008; Murdie, 2008; Dwyer & Brown, 2008; Francis & Hiebert, 2014; Adam et al., 2019; Adam et al., 2020; Dotsey & Lumley-Sapanski, 2021). These difficulties are sometimes exacerbated by the discriminatory attitudes of some landlords who refuse to rent to refugees (Werner et al., 2018; Weidinger & Kordel, 2020). Given that refugees struggle to access this market through common paths (Adam et al., 2019), (local) authorities’ actions weigh heavily in this process, shaping opportunities and barriers to social inclusion (Bolzoni et al., 2015; Zetter & Pearl, 1999).

Using full-population register data (STATPOP 2012-2019), this study examines the transition from collective centres to private accommodations among asylum migrants in Switzerland. We contribute to the literature on refugees’ residential mobility and integration in three ways. First, the study focuses on the early stages of the residential trajectory of asylum migrants and emphasizes one key transition: the transition out of collective centres. Recent studies have drawn attention to refugees’ residential mobility and geographic choices (i.e. moves to urban vs. rural areas) after the restrictions of the dispersal policy have been lifted (see de Hoon et al., 2021 for the Netherlands and Vogiazides & Mondani, 2021 for Sweden). However, these studies take as their starting point the end of the asylum procedure and, more specifically, entry into the first private housing. Residential dynamics that precede refugee status and freedom of movement, although critical to integration, have not yet been studied by quantitative researchers. Thus, this study provides a complementary perspective on the housing experience of refugees by analysing residential moves from the deposition of an asylum claim and during the migrant’s first eight years in the country (regardless of whether the individual is granted refugee status).

Second, we take advantage of register data allowing for a comprehensive longitudinal mapping of the residential trajectories of all individuals who were registered as asylum seekers, provisionally admitted (similar to a subsidiary protection in European countries) or recognized as refugees in Switzerland between 2012 and 2019. Although qualitative research abounds on the subject of refugee housing (e.g. Adam et al., 2019, 2020; Aigner, 2019; Alberti, 2020; Darling, 2016; Dotsey & Lumley-Sapanski, 2021; Fozdar & Hartley, 2014; Gold, 2019; van der Horst, 2004; Kissoon, 2010; Lietaert et al., 2020; Mottet, 2022; Sirriyeh, 2010; Szczepanikova, 2013; Weidinger & Kordel, 2020; Werner et al., 2018; Ziersch et al., 2017), quantitative studies are scarce–let alone longitudinal ones. Event history analysis applied to interval-censored survival time data will provide a descriptive yet exhaustive overview of transitions out of collective centres. Third, we provide insights into the logic of housing attribution by local authorities since (as discussed below) the criteria for transfer to private accommodation are not publicly available. Kaplan–Meier estimates will first be used to emphasize the time spent in collective centres and the proportion of individuals still living in these accommodations over time. Cox proportional hazards models will then be used to further examine the determinants (both individual and contextual) of a transition out of collective centres.

2. Refugee housing in Switzerland

As in many other countries, the number of asylum applications lodged in Switzerland has varied over the years depending on the international context. The 1990s were marked in particular by the arrival of applicants from the former Yugoslavia (with over 40,000 applications in 1991, 1998 and 1999). In 2015, Switzerland also saw an increase in asylum requests due to the war in Syria (nearly 40,000 applications). On average, over 21,500 applications were filed each year in Switzerland between 1986 and 2021. The three main countries of origin of asylum seekers in Switzerland are Eritrea, Afghanistan and Syria, which together account for 38% of the asylum requests lodged in Switzerland in the last decade (2011–2021).

Following the rules of the Swiss Asylum Act, the authorities are responsible for housing the asylum seekers during the procedure and beyond (for those who are unable to do so on their own). Thus, upon arrival to Switzerland, asylum seekers “typically live isolated from the local population in an assigned [federal] reception centre” (Hainmueller et al., 2016:1), after which they are transferred to a cantonal housing facility–either in collective or private accommodation (see Figure 1). Such transitions can also happen after an asylum seeker is granted refugee status or subsidiary protection.

2.1. The federal reception and processing centres (CEP)

The reception of asylum seekers starts in one of the five federal reception and processing centres (CEPs) run by the Swiss Confederation, which can accommodate between 200 and 300 people each (ECRE, 2015).¹ Where there is a shortage of space, the Confederation may also choose to place asylum seekers–particularly those subject to removal under the Dublin Agreement–in ‘remote sites’, typically former military shelters (ECRE, 2015). The CEP registers asylum seekers (identity check, photograph and fingerprinting, questions about the route taken and first interview about the grounds for asylum). The migrants’ ability to leave the centre is limited to certain hours, and their comings and goings are highly controlled, with body searches at every entry and exit by security officers. In the case of unjustified absence or lateness, they may be punished by a ban on going out or financial penalties. As Gold (2019) notes, in these centres, “most of the day is spent waiting: for the results of the first interview, for papers to arrive from overseas, for news from relatives.” The overall duration of the stay in the CEP cannot exceed 3 months.

2.2. The cantonal collective centres

Asylum migrants are then allocated to the Swiss cantons at random according to a distribution key (SEM, 2019). This distribution key allocates a certain percentage of asylum migrants to each canton according to the size of its population. Only nuclear families cannot be separated². Each canton is then responsible for housing the asylum migrants, who “are bound to their canton of attribution and must reside within that canton” (ECRE, 2015, p.71). Cantonal reception systems generally include several types of accommodations (collective centres, individual housing, and specific housing for unaccompanied minors.). As each canton is free to manage the accommodation of asylum migrants as it sees fit, there are major differences between them, leading to significant variations in the living conditions of the persons concerned (Alberti, 2020) and consequently to a strong sense of injustice among them (Mottet, 2022).

Usually, the cantons organize accommodation in two phases: the first phase is in collective centres, and the second phase is in private accommodations (see Figure 1)³. Cantonal collective centres have two functions: to monitor asylum seekers during the procedure and beyond (that is, after they obtain refugee status or a provisional admission⁴) and to control their comings and goings to ensure “that they are located where they should be and are at the disposal of the authorities” (Alberti, 2020: 4)⁵. However, the centres also offer asylum migrants support and assistance to ease their integration in Switzerland. To this end, counselling and support in their daily lives are provided by social workers. Asylum migrants also receive material assistance and have access to health care, language courses and employment programs. In addition, they must attend training modules on various subjects (e.g. health, housing, professional integration, sex and gender, or the habits and customs of their host region). ‘Occupation programmes’ are also offered to keep them busy and prepare them for their future entry into the labour market. People in these centres may experience living in collective accommodation as an ordeal (lack of privacy, feeling of confinement, lack of links with the host society, sometimes spatial relegation away from urban centres) and look forward to their transfer to individual accommodation (Mottet, 2022). It should also be noted that transfers from one collective centre to another within the same canton are common. These successive rehousings are, like many ruptures, difficult for migrants to cope with as they seek stability (Mottet, 2022).

Overall, these collective centres can be considered a space “between home and prison. It is both a space of care and recomposition of the self, and a space of semi-confinement” (Alberti, 2020: 169). This dual function–control and assistance–is also found elsewhere in Europe (Szczepanikova, 2013). In the Swiss asylum system, collective centres are perceived as a transitional phase, a parenthesis whose aim is to ‘prepare’ people as well as possible for their transfer to private housing, which is synonymous with a more independent and autonomous life (Alberti, 2020).

2.3. From collective centres to private accommodation

In Switzerland, access to private housing for asylum migrants can either be the result of an individual initiative or decided (and financed) by the cantonal authorities, as private accommodations include subsidized flats; due to asylum seeker and refugees’ lack of financial resources, the latter is much more common. Therefore, when they are transferred into such private accommodations depends on the canton and its private housing capacity (OFS, 2016). In practice, when a place in private accommodation becomes available, the canton notifies the officers of the collective centres, who then propose the files of “persons suitable for transfer” (as the officers themselves put it–Alberti, 2020: 153). From this list, the cantonal authorities decide who will ultimately be transferred.

At the national level, there are no official guidelines outlining the precise criteria for the profile of people who should be given priority for private housing, as this is left to the cantons. However, to the best of our knowledge, there are also no such guidelines at the cantonal level. Internal documents might exist within some of the organisations mandated to run the collective centres, but if so, these are not made public. Thus, the criteria for transferring to private housing are vague: while time spent in collective centres may play a large role, it is not entirely decisive, and other criteria may prevail. In particular, residence permits have an important impact (Alberti, 2020) since people with asylum-seeker status are much less likely to be allocated private housing than those with more stable residence permits. The situation of ‘vulnerable persons’ (families and persons with health problems) is more ambiguous: on the one hand, one can imagine that they are considered a ‘priority’ for a transfer; on the other hand, one can conceive that they require closer care, corresponding to a longer period in a collective centre⁶. Finally, Alberti (2020: 166) notes that the agents in charge of collective centres rely on participation in the various training modules to “assess whether a person is ready to be transferred to individual housing and to settle on his or her own”.

In all cases, the transfer is experienced as “a moment of joy for many residents who are looking forward to a personal living space that will allow them to live more independently. In this sense, the transfer is perceived by the residents as a success, a further step toward settlement in Switzerland.” (Alberti, 2020: 168). However, we have seen that individuals have little influence on when this transition occurs since it depends on structural and organisational aspects (available places), legal status hierarchy (priority given to the most stable residence permits), and the assessment of the officers in charge of managing the collective centres as to the supposed ability of individuals to get by ‘outside’ independently⁷.

2.4. Determinant of refugee access to housing

Alongside the growing media interest in the subject, researchers have studied the issue of the housing conditions of (asylum) migrants. However, to date, most research has focused on the impact of housing on different life outcomes, such as mental and physical health issues (Bakker et al., 2016; Ziersch et al., 2017; Ziersch & Due, 2018), socioeconomic integration (Bakker et al., 2014), family life (Lietaert et al., 2020), or integration in general (Adam et al., 2020; Francis & Hiebert, 2014). In contrast, the determinants of access to quality housing have received limited attention. This section focuses on the literature documenting the association between individual and contextual factors and access to housing for (asylum) migrants.

The literature documents variations in access to quality housing by socioeconomic characteristics. Women experience more stable housing trajectories than men, including a faster transition to the private rental sector of the housing market and access to homeownership (Berger, 2008; Shier et al., 2016). The reasons for migration explain part of this dynamic: women more often migrate as part of the family reunification program, meaning that they more often move in with an established partner who likely has further progressed on the housing ladder. Nevertheless, it was shown in the French context that after controlling for the reason for migration, women were less likely to live in precarious accommodations, such as hotels or other collective accommodation centres (Berger, 2008). In Switzerland, when there is a lack of space in ‘standard’ collective centres, single men are the first to be housed in emergency shelters (civil protection shelters) (ECRE, 2015). The climate of insecurity and tension that can prevail in collective reception facilities may be perceived as unsuitable for women and the children who often accompany them. As such, we expect women to make a faster transition to private housing.

Family configuration also plays a role in housing transfers. On the one hand, families are generally given priority in access to private housing (ECRE, 2015; Shier et al., 2016). On the other hand, it has been pointed out that the structure of the housing market (lack of small units) is not suited to accommodate the household structure of the asylum population (i.e. many single people and, to a lesser extent, large families) (Adam et al. (2020). This suggests that family size matters and that single people and large families may be at greater risk of prolonged stays in collective centres. However, many so-called private dwellings are in fact shared flats that house several unrelated single persons, which could counterbalance this ‘favouritism’ towards families.

The impact of legal status on housing type has also been highlighted in various countries (Baier & Siegert, 2018; Borevi & Bengtsson, 2015; Doherty, 2012; Murdie, 2008). From what we know from field observations in asylum centres in Switzerland (Alberti, 2020; ECRE, 2015), we expect the same logic to apply: “the closer a new arrival comes to full citizenship status, the less legitimate we would expect restrictions in their autonomy [in the housing market] to appear” (Borevi & Bengtsson, 2015: 2603). Specifically, recognized refugees (with a ‘B’ permit) would get an apartment more quickly than holders of provisional admission/subsidiary protection (the ‘F’ permit), and even more so than the asylum seekers (the ‘N’ permit) whose procedure is still pending. In short, we expect the chances to move out of collective centres to reflect the hierarchy of residence permits (N < F < B).

Studies also stress the role of social networks in the chances of accessing independent housing (Adam et al., 2019; Berger, 2008; Francis & Hiebert, 2014; Hanley et al., 2018; Murdie, 2008; Wyckaert et al., 2020). In addition to family and friendship networks, the size of ethnic communities plays an important role in the integration process (Adam et al., 2020). Belonging to the same ‘ethnic background’ or national community can provide access to information and support in the search for housing (Adam et al., 2020: 206). The logic of housing attribution for asylum migrants within subsidized shared flats may also accelerate the relocation of larger nationality groups: whenever possible, Swiss authorities group unrelated people of the same nationality together, thus following the logic of encouraging ‘subgroup identities’ (see Borevi & Bengtsson, 2015: 2604).

The temporal and geographical context is not to be overlooked. The so-called ‘refugee crisis’ of 2015–2016 impacted reception structures in many European countries (Werner et al., 2018), including Switzerland (Alberti, 2020). Hence, it is expected that 2015 might be a turning point in the likelihood of moving into private accommodation due to a ‘saturation effect’ of available (subsidized) flats. In addition, as in other countries, such as Germany (Adam et al., 2020; Werner et al., 2018), Austria (Rosenberger & Konig, 2012) and Belgium (Wyckaert et al., 2020), there are significant subnational differences in the reception and support conditions for asylum migrants in Switzerland (Alberti, 2020). Asylum migrants are randomly assigned to a canton with no option to change it, and the canton of attribution strongly influences their housing trajectories (ECRE, 2015). In 2015, for example, some cantons (Aargau, Glarus, Zug, Appenzell Innerrhoden) housed almost all asylum social assistance beneficiaries in collective centres, while others (Appenzell Ausserrhoden, Ticino) made very little use of them (OFS, 2016). These intercantonal discrepancies can be explained in particular by financial (according to the respective costs of the different types of housing in each canton (Alberti, 2020)) and political (as collective centres are generally not well accepted by the population) considerations. Finally, access to independent housing is more difficult in urban areas (Berger, 2008), and asylum migrants can face difficulties in finding a flat in small and densely populated cantons (such as Geneva and Basel-Stadt) with tight housing markets (Alberti, 2020), especially when there is a shortage of subsidized flats (Adam et al., 2020).

3. Analytical strategy

3.1. Data

We used linked administrative data from the Swiss population register (STATPOP) and income register (CCO) for the period 2012–2019. The population register contains information on different demographic dimensions for all persons living legally in the country on December 31 each year. The variables include gender, age, marital status, country of birth, residence permit, municipality of residence, and dates of immigration and emigration. The register also document the type of housing—private or collective—in which people live. The income register provides information on salary from (self-)employment, excluding any social benefits.

To be identified as an asylum migrant, one must hold an “N” or “F” permit on December 31 the year of their arrival to Switzerland (i.e. when they are first observed in the data). The N permit is the asylum seekers’ permit and is for people whose asylum request is being examined by the authorities (Art. 42 AsylA). The F permit is a provisional admission that allows temporary stay in the territory for rejected asylum seekers that Switzerland cannot deport (either because sending them back would put their lives at risk or merely for practical reasons, Art. 83 FNA) (Bertrand, 2019). The residence permit that allows the identification of asylum migrants is only available on a yearly basis. This means that a few housing trajectories are not observable from these data, such as those relating to individuals for whom the asylum claim is promptly rejected (e.g. Dublin procedure) and those who are quickly recognized as refugees. Specifically, individuals who applied for asylum and left Switzerland before December 31 during their first year of arrival are not recorded in the data. In addition, individuals who obtained refugee status during this time window cannot be identified as asylum migrants. This is because when people are recognized as refugees, they receive a B permit. As the B permit is also issued to foreigners who have obtained a residence permit in Switzerland for professional or family reasons, recognized refugees who obtained this permit the same year as their arrival in the country cannot be distinguished from other migrant populations. Minors are excluded from the analysis because housing allocation procedures and reception facilities differ considerably for this population. The study population comprises 50,871 asylum seekers and refugees aged 18 or older.

3.2. Methods

We used event history models to analyse the transitions out of collective centres. Individuals are at risk from arrival to Switzerland (at the earliest on January 1, 2012) until a transition to private housing, emigration, death, or the end of observation (December 31, 2019). People can enter the risk population each year between 2012 and 2018. The data document the exact date of arrival in Switzerland, emigration, and death. Housing transitions are available on a yearly basis. All trajectories start in a federal collective centre. When they are observed (i.e. on December 31 each year), asylum migrants are classified as living in a federal or cantonal collective centre (event = 0) or private accommodation (event = 1). The observation period ranges from 1 to 2922 days (8 years).

Because events can only occur within certain intervals (i.e. between December 31 of year t and December 31 of year t + 1), we use event history models applied to interval-censored survival time data using the icenReg package in R. In contrast to continuous survival time data, we define a time interval within which the event is known to have occurred. The lower and upper intervals correspond to the last time the person was observed in a collective centre and the first time the person was in private accommodation, respectively. First, Kaplan–Meier survival curves are used to estimate the (median) time spent in collective centres. These curves describe the proportion of individuals who did not experience a transition to private housing over time. Second, using a Cox proportional hazard model, we examine the association between individual and contextual factors and the speed of access to private housing. These models are well suited to deal with right-censored trajectories—a common experience among asylum migrants for whom rejection of an asylum application can lead to forced departure.

Missing values are an important limitation of these analyses: approximately one-third of the trajectories have at least one missing value on housing type (Table A3 for details on missing values). In most cases, we could impute plausible values based on the following assumptions. First, individuals for whom the first housing type was missing and who arrived at the end of the year (i.e. between September and December) were assigned to the collective centre category. This situation applied to half of the missing values (14% of the study population), likely due to registration delays. Second, individuals for whom the first type of housing was missing and the second type of housing was collective were categorized as living in a collective centre the first year (2.6%). Additional analysis confirmed that moves to collective centres after being lodged in private housing are very uncommon. Third, individuals for whom the first type of housing was missing and the second type of housing was private were assumed to have spent half the duration of the first observation year in collective housing and the other half in private housing (5.1%). The sensitivity analysis in Table A4 shows the results for the subset of trajectories without missing values; the results are consistent with those that include trajectories with imputed values. Finally, trajectories that had two consecutive missing values were dropped from the analyses (7% of the study population).

3.3. Variables

The dependent variable is the type of housing, which is available on a yearly basis. “Collective accommodation” refers mainly to “ordinary” cantonal centres but may also include other types of collective accommodations, e.g. detention centres for the purpose of removal or underground shelters that have sometimes been opened in an emergency to alleviate the lack of accommodation (ECRE, 2015). In turn, “private housing” can cover a variety of situations, including subsidized and nonsubsidized housing, shared apartments between unrelated refugees, and (rare) cases of people living with “host families”.⁸ However, the data do not provide any further details on the type or quality of private housing.

Most explanatory variables are measured on December 31 of the first year following arrival (i.e. when people are first observed in the data). Individual characteristics include age at arrival, gender, marital status at arrival, citizenship, residence permit, and annual salary from (self)employment. Family composition at arrival and whether individuals are accompanied by their spouses or child(ren) is unknown. Marital status is used as an indicator of family configuration, although we expect women to be more often accompanied by their families than men. Because we expect group size to matter in the housing allocation process, we group nationalities into five categories according to their respective share in the asylum migrant population: Eritreans, Syrians, Afghans, other large groups (Sri Lanka, Iraq, Somalia, China/Tibet, Turkey, Iran, Nigeria, and Ethiopia), and other small groups.

Information on the residence permit (and therefore the time at which refugee status is obtained) is only available on a yearly basis. As a result, the order of events (i.e. permit changes and housing transitions) is not easily determined. More importantly, asylum migrants who obtained a B permit during the first year were excluded from the analysis. Including the residence permit as a time-varying covariate would, therefore, result in biased estimates, i.e. the effect of the N permit would be overestimated and that of the B permit underestimated. Instead, we considered the legal pathway (which permit the person has obtained at the end of the trajectory). Thus, we assessed the reciprocal influence of (or association between) permits and housing trajectories, although, theoretically, legal status is more likely to influence the housing trajectory than the other way around. We considered three distinct legal pathways: asylum seekers (N permit), provisionally admitted (F permit), and recognized refugees (B permit⁹). Similarly, we accounted for the association between an individual salary and the speed of access to private housing. This variable indicates whether the person was (self)employed at any point during the observed trajectory and their corresponding salary level. We distinguished individuals who have not received any salary from employment during the observation window, from those who participated to the labour market at some point but received a precarious salary¹⁰ (below 20,000 CHF during a given year), and those who received a better salary (above 20,000 CHF during a given year). This variable serves as an indicator of labour market integration and financial independence. However, due to the annual granularity of the available information the measure is one of association.

Contextual variables included the canton to which the asylum migrant was first assigned and the cohort (year) of arrival. In addition, we have tested whether the size of the arrival cohorts plays a role in the speed of access to private housing, distinguishing small cohorts from medium and large cohorts. We found no association between cohort size and the housing trajectories (not shown).

4. Results

Table 1 describes the number of person-days (time at risk) and transitions to private housing (event) by individual characteristics. Men account for the largest share of person-days (73%), and the transition rate to private housing is lower for men than for women. The age structure is considerably young, with one-fifth of the individuals aged 35 years or older. The differences in transition rates are relatively small across age groups; in general, the older the individuals are, the higher the transition rate will be. Most asylum migrants are single on arrival (73%), a quarter are married, and only 3% are divorced or widowed. Married individuals have the highest transition rate, followed by divorced/widowed individuals and singles. Eritreans are the most represented group (30%), followed by Afghans (16%) and Syrians (13%). Other large nationality groups account for a quarter of the time at risk, and other less represented nationalities account for 16%. Syrians have the highest transition rate, and other small groups have the lowest.

The majority of trajectories are made up of asylum migrants who have never integrated into the labour market (i.e. no income category) and they have the lowest transition rate to private housing. Less than 10% of individuals have earned a salary level above 20,000 CHF for at least one year during their trajectory. As with many countries, Switzerland received an increased number of asylum applications in 2015. This cohort accounted for 37% of the time at risk.

Kaplan–Meier estimates (Figure 2) show the proportion of asylum migrants who live in a collective centre over time (from arrival and during the first eight years at most). Individuals are observed for a minimum of one day and a maximum of 2922 days. After one year in Switzerland, 42% had not yet experienced a transition to private housing; this proportion was 23% after 2 years and 7% after 3 years. The median time spent in a collective centre is 318 days. Overall, 72% of asylum seekers and refugees who were registered in Switzerland between 2012 and 2019 experienced a transition to private housing (Table A1). Right-censoring covers a large share of the trajectories: 21% left the country during the observation period—most often due to referral to the end of the procedure—and 7% were still sheltered in a collective accommodation at the end of the observation period.

Table 2 presents the results of the Cox proportional hazard models and shows variations in the time to transition to private housing as a function of individual and contextual characteristics. Model 1 presents results for individual characteristics only; Model 2 adds contextual variables. Women move into private housing more quickly than men, with a 40% higher hazard ratio. This aligns with the expectation that women are favoured when a place in private housing becomes available. The older a person is, the greater their chances will be of obtaining private housing.

Compared to singles, married individuals experience a faster transition to private housing. Separated and widowed individuals, in contrast, wait longer. However, the association between marital status and the speed of access to private housing differs by gender (Table A5): single women access private housing 1.5 times faster than single men. In fact, the speed of access to private housing for single women is comparable to that of married men. Marital status is a proxy for family configuration, and whether the person is accompanied by their spouse and chil(dren) is unknown. It is more common for men (single or married) to undertake asylum migration on their own and then to bring their families from abroad. In contrast, married women are generally accompanied by their families. Thus, marital status is a better reflection of family composition for women than for men. Group size also matters in housing allocation processes: Syrians make a faster transition to private housing, with a 30% higher hazard ratio; all other large groups have similar transition rates. Only asylum migrants that belong to less represented nationality groups wait longer to access private housing (although the differences are small).

The association between legal status and the speed of access to private housing reflects the permit hierarchy: the hazard ratios are the highest among recognized refugees (B permit), followed by those who obtained a provisional admission (F permit) and those for whom the decision was still pending at the end of observation (N permit). We reiterate that the permit is measured a posteriori and must be interpreted with caution. Integration into the labour market and the level of salary are also associated with the speed of access to private housing. Individuals who entered the labour market, even at a low wage level, experienced a faster transition than those who had no formal paid employment during this period. This association is even more pronounced for people with a higher salary. Overall, hazard ratios for individual characteristics are almost identical between Model 1 and Model 2, suggesting that, despite the quasi-autonomous management of refugee housing by the cantons, priority rules covering access to private housing apply across the country.

Contextual effects are shown in Model 2 and Figure 3. As expected, the increase in arrivals in 2015 seems to have had an effect on access to housing for asylum migrants already in the centers. Asylum migrants who sought protection in Switzerland in 2014 make a faster transition to private housing. The results also show important variations in access to private housing depending on the canton to which applicants have been assigned (Figure 3). In the cantons of Appenzell Ausserrhoden, Neuchâtel, Ticino, Lucerne, Valais, St. Gallen, and Solothurn, asylum migrants access private housing at least twice as quickly as those in the canton of Zurich (the reference canton and the largest Swiss canton). In contrast, asylum seekers assigned to the cantons of Appenzell Innerrhoden, Geneva and Nidwalden wait at least twice as long. Considering that refugees do not have a choice of where to live, local policies strongly influence the living conditions of this population, and their experiences may be random and uneven.

5. Conclusion

This study is the first to produce quantitative insights into transitions between collective and private housing for complete cohorts of asylum migrants. Using the population register for Switzerland, we tracked the type of housing people live in over a period of up to eight years, starting from the deposition of an asylum claim. Previous studies have documented how the management of asylum procedures (e.g. dispersal policies, employment bans, length of procedures) affects different integration outcomes (Auer, 2018; Hainmueller et al., 2016; Marbach et al., 2018). This study takes a complementary view by focusing on the logistics of housing attribution, the different rationales governing these procedures, and how they shape individual opportunity structure. We found that despite a quasi-autonomous management of refugee housing by regions (cantons), priority rules covering access to private housing apply across the country. When choosing between different profiles, women, older asylum migrants, married individuals, and members of larger national groups are favoured in access to private housing. The results are consistent with the expectation that collective centres are perceived as unfit environments for women and families, prompting more rapid placement of these groups in private housing. In addition, the belief that people from the same cultural background are better candidates to share a home may be one of the reasons why the largest national groups access private housing more quickly. Another reason may lie in the anticipation of recognition of refugee status (or subsidiary protection) for the most represented groups and, therefore, a favourable recommendation for transfer to private housing by the local authorities. The positive association between an individual’s salary (i.e. whether a person had participated in the labour market at any time during the observation period) and the speed of access to private housing reflects, to some extent, the ability to navigate the housing market without the involvement of local authorities. One can also expect local authorities to favour the transfer of those who have successfully integrated the labour market; the two processes are undistinguishable in the current framework. The 2014 cohort seems to have benefited from faster access to private housing. This situation likely results from a saturation effect accelerating the transition out of collective centers for the previous cohort. Nevertheless, the time spent in collective centres largely depends on the region to which a claimant is assigned, indicating the very little agency asylum migrants have during their first years of residence. There are several limitations that should be mentioned to properly assess the significance of the results. Information on residence permits allowing the identification of asylum migrants is only available annually, meaning that not all trajectories can be included. Those who promptly obtain refugee status are indistinguishable from other international migrants and are excluded from the analyses. This likely results in an overestimation of the time spent in collective centres–these cases are theoretically more likely to make a fast transition out of such centres. As a consequence, transition rates for individuals with a legal trajectory leading to a B permit (recognized refugee) are likely underestimated. Moreover, the population register contains a limited number of variables. Individual characteristics linked to integration outcomes, such as human capital, are not included, nor are other personal situations (e.g. (mental) health issues) that could affect the time spent in collective centres. Another limitation concerns the type of moves considered. A move to private housing is considered an absorbing state, although a few trajectories (5%) include at least one move from private housing back to collective accommodation. This may be the case for people who have been transferred to a private home and are subsequently denied refugee status (or granted subsidiary protection). In addition, people may experience repeated transfers to different collective centres. This situation may have a significant impact on integration outcomes and that we did not cover in this single-event framework. Although documenting the specificities of more complex trajectories is beyond the scope of this study, it represents an important avenue for future research.

One must also consider that although a transfer to private accommodation is usually eagerly awaited, private housing reflects a variety of situations and may not be a desirable outcome for all. For instance, individuals placed in shared flats with strangers (although more intimate than large collective centres with dormitories) may not completely feel at home. Some private homes may also be located in remote areas where access to public transport or contact with the community network is limited. As long as asylum migrants cannot arrange their housing on their own, they will remain at risk of being placed in accommodations that are not suited to their needs. This is a significant issue, given that the search for housing can prove very challenging for asylum migrants, whether due to a reluctance of landlords to rent to this population (especially those with subsidiary protection) or because of a difficult integration into the labour market. Asylum migrants who succeed in finding employment often end up in the lower-paid sector of the labour market. As a result, when seeking an apartment, asylum migrants often find themselves at the bottom of the list of applicants. In tight, competitive housing markets, such as those in most Swiss cities, the chances of finding an apartment may be substantially reduced.

More generally, difficulties in accessing decent housing are another facet of asylum seekers’ uncertainty regarding their right to stay in the territory (Bertrand, 2019, 2020; De Coulon, 2019; Gold, 2019). A precarious, ‘in-between’ status is reflected in the near invisibility of asylum centres since they are located in anonymous buildings in the city (Alberti, 2020), in isolated locations (Adam et al., 2020), or buried underground (Del Biaggio & Rey, 2017). Where people live shapes their opportunities for economic and social inclusion or imposes barriers to these goals. Nevertheless, the residential instability and insecurity imposed on asylum migrants is often paradoxically at odds with the integration objectives expected of this population. The provision of appropriate housing adapted to the needs and structure of this population could substantially alleviate the inclusion challenges they often experience.

Acknowledgements

We would like to thank Adrien Remund, Didier Ruedin, and the Laboratory for the Study of Social Processes (LAPS) for helpful comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Swiss National Science Foundation. Postdoc Mobility fellowship: P400PS_199269, and the Laboratory for the Study of Social Processes at the University of Neuchâtel.

Notes on contributors

Julie Lacroix

Julie Lacroix is a research fellow at the University of Geneva, and associated research fellow at the University of St Andrews. She obtained her PhD in demography at the University of Geneva and her MA at the University of Montreal. She uses population registers and event history models to analyse immigrant residential mobility, and population redistribution more generally. She also makes use of experiments to examine discrimination in the housing market.

Anne-Laure Bertrand

Anne-Laure Bertrand holds a PhD in demography from the University of Geneva. She is a lecturer at the University of Neuchâtel, where she conducts research on asylum and migration, and teaches quantitative methods to social science students.

Notes

1 See ECRE (2015) for the procedure at airports. It should also be noted that we describe here the procedure as it was in force at the time of the data collection. This procedure has changed somewhat since the 2019 asylum reform (see https://www.sem.admin.ch/sem/fr/home/asyl/asylverfahren/asylregionen-baz.html [15.07.2022]).

2 The failure to consider other criteria, such as closeness to extended family members or language skills, is a recurrent criticism (ECRE 2015).

3 This organisation of housing for asylum seekers is similar to that used elsewhere in Europe, as Werner et al. (2018) shows with the example of the German city of Leipzig.

4 Provisional admission is a status roughly equivalent to subsidiary protection in EU law.

5 Life in the cantonal centres is less strictly regulated than in the CEP, particularly with regard to the monitoring of entry and exit (ECRE 2015).

6 Alberti (2020) notes that the notion of ‘care’ dominates in collective housing, while it is the notion of ‘integration’ that guides the follow-up of individuals after their transfer to a flat.

7 Supervision does not end at the door of the centre but is later conducted on a more ad hoc basis, usually with regular appointments at the office responsible for providing social benefits.

8 This last case mainly concerns unaccompanied children who are not taken into account in our analyses.

9 Permit B trajectories do not exclusively include people recognized as refugees. Some people who were initially granted a subsidiary protection may later receive a B permit if they can demonstrate that they are sufficiently integrated.

10 According to the OFS, the poverty line for a single person corresponds to an annual income of CHF 27,500 (see https://www.bfs.admin.ch/bfs/fr/home/statistiques/situation-economique-sociale-population/bien-etre-pauvrete/pauvrete-privations/pauvrete.html [01.02.2024]).

Appendices

Figure 1. Housing for asylum seekers and refugees in Switzerland.

Source: Schematic representation inspired by the one proposed by the State Secretariat for Migration (undated, year 2010)

Figure 2. Kaplan‒Meier survival estimate for the transition to private housing among asylum migrants over 18 years of age in Switzerland from 2012 to 2019.

Note: Time (in days) since arrival in Switzerland. Source: Swiss population register.

Figure 3. Cantonal differences in access to private housing among asylum migrants over 18 years of age in Switzerland from 2012 to 2019.

Note: Canton fixed effect (hazard ratio) from Model 2. (Ref: Zurich). Source: Swiss population register.

Table A1. Number of events and censored trajectories.

	Freq.	Percent
Transition to private housing	36,684	72%
Right-censored - emigration/death	10,672	21%
Right-censored - end of observation (2019)	3,515	7%
Total	50,871	100%

Table A2. Number of events and censored trajectories by year of arrival.

	2012	2013	2014	2015	2016	2017	2018
Transition to private housing	64.2	69.1	84.3	77.1	68.9	64.3	45.9
Right censored - emigration/death	35.2	30.2	13.6	16.9	19.7	17.9	19.3
Right censored - end of observation (2019)	0.7	0.8	2.2	6.0	11.4	17.8	34.8
Total	7,893	5,097	8,814	17,496	5,355	3,430	2,786

Table A3. Trajectories with missing values.

	Freq.	Percent
First value is missing: arrived between September and December	7,614	13.9
First value is missing; second value is collective centre	1,400	2.6
First value is missing; second value is private housing	2,796	5.1
Trajectories with two consecutive missing values	3,859	7.1
Trajectories without missing values	39,061	71,4
Total	54,730	100,0
Total main analysis*	50,871	92.9

* For the main analyses, only trajectories with two consecutive values were removed (N = 54,730-3,859 = 50,871). Source: Swiss population register.

Table A4. Cox model with interval-censored data for the transition to private housing among asylum migrants over 18 years of age in Switzerland from 2012-2019, without missing values.

	Model 2
	H.R.	p
Gender
Men (Ref.)	1,00
Women	1,47	***
Age
18–24 (Ref.)	1,00
25–34	1,10	***
35–44	1,14	***
45–54	1,14	***
55+	1,27	***
Marital status
Single (Ref.)	1,00
Married	1,40	***
Separated/Divorced	0,91	*
Citizenship
Eritrea (Ref.)	1,00
Syria	1,40	***
Afghanistan	1,11	***
Other large groups	1,08	***
Other small groups	0,99
Residence permit
Asylum seekers (N permit) (Ref.)	1,00
Provisionally admitted (F permit)	1,52	***
Recognized refugees (B permit)	1,90	***
Salary
No income	1,00
Very low income	1,16	***
Low to high income	1,28	***
Cohort (year of arrival)
2012 (Ref.)	1,00
2013	1,13	***
2014	1,22	***
2015	1,16	***
2016	1,03
2017	1,04
2018	0,92	*
Canton	X
N Subject	39,061

Note: Model 2 without trajectories with missing values. H.R. = Hazard Ratio.

* p < 0.05.

** p < 0.01.

*** p < 0.001; Source: Swiss population register.

Table A5. Cox model with interval censored data for the transition to private housing among asylum migrants over 18 years of age in Switzerland from 2012-2019.

		Model 2
		H.R.	p
Men	Single (Ref.)	1.00
	Married	1.43	***
	Separated/Divorced	0.97
Women	Single	1.50	***
	Married	1.83	***
	Separated/Divorced	1.28	***

Note: Model 2 with an interaction between marital status and gender. Note: H.R. = Hazard Ratio; Std. Err. = Standard error.

* p < 0.05.

** p < 0.01.

*** p < 0.001.

Source: Swiss population register.

Table 1. Number of person-days and housing transitions by individual and contextual characteristics, asylum migrants aged 18 or older in Switzerland (2012–2019).

	Person-days	%	Event	Rate
Gender
Men	18619766	72,6	23055	0,0012
Women	7033251	27,4	13629	0,0019
Age group
18–24	11477490	44,7	14571	0,0013
25–34	8920990	34,8	13320	0,0015
35–44	3525957	13,7	5746	0,0016
45–54	1112911	4,3	1887	0,0017
55+	615669	2,4	1160	0,0019
Marital status
single	18783548	73,2	23692	0,0013
maried	5991755	23,4	11757	0,0020
separated	877712	3,4	1235	0,0014
Citizenship
Eritrea	7585952	29,6	11400	0,0015
Syria	3205669	12,5	6975	0,0022
Afghanistan	4163021	16,2	5807	0,0014
Other large groups	6605597	25,7	8857	0,0013
Other small groups	4092777	16,0	3645	0,0009
Residence permit
Asylum seekers (N permit)	8569418	33,4	6711	0,0008
Provisionally admitted (F permit)	9597464	37,4	15580	0,0016
Recognized refugees (B permit)	7486134	29,2	14393	0,0019
Salary
No income	15808466	61,6	19021	0,0012
Very low income	7446351	29,0	13322	0,0018
Low to high income	2398199	9,3	4341	0,0018
Cohort (year of arrival)
2012	3982933	15,5	5066	0,0013
2013	2297852	9,0	3521	0,0015
2014	4304745	16,8	7429	0,0017
2015	9463911	36,9	13494	0,0014
2016	2948692	11,5	3689	0,0013
2017	1619842	6,3	2206	0,0014
2018	1035042	4,0	1279	0,0012

Note: Authors’ own calculation. The number of person-days is the exact number of days for individuals who emigrate, die or are right-censored on December 31, 2019; for those experiencing a transition, the event is assumed to have occurred in the middle of the interval. The variable salary is a time-fixed variable that indicates whether the person was (self-)employed at any point during the observed trajectory. The highest salary received in a given year was assigned to the entire trajectory and time at risk. The residence permit’s value was also retroactively attributed to the entire trajectory. Source: Swiss population register.

Table 2. Cox model with interval-censored data for the transition to private housing among asylum migrants over 18 years of age in Switzerland from 2012 to 2019.

	Model 1		Model 2
	H.R.	p	H.R.	p
Gender
Men (Ref.)	1,00		1,00
Women	1,47	***	1,48	***
Age
18-24 (Ref.)	1,00		1,00
25-34	1,07	***	1,10	***
35-44	1,13	***	1,17	***
45-54	1,15	***	1,18	***
55+	1,27	***	1,32	***
Marital status
Single (Ref.)	1,00		1,00
Married	1,36	***	1,36	***
Separated/Divorced	0,90	**	0,88	***
Citizenship
Eritrea (Ref.)	1,00		1,00
Syria	1,26	***	1,32	***
Afghanistan	1,00		1,01
Other large groups	1,01		1,04	**
Other small groups	0,89	***	0,94	**
Residence permit
Asylum seekers (N permit) (Ref.)	1,00		1,00
Provisionally admitted (F permit)	1,47	***	1,49	***
Recognized refugees (B permit)	1,75	***	1,83	***
Salary
No income	1,00		1,00
Very low income	1,18	***	1,18	***
Low to high income	1,25	***	1,31	***
Cohort (year of arrival)
2012 (Ref.)			1,00
2013			1,09	***
2014			1,21	***
2015			1,10	***
2016			1,02
2017			1,08	*
2018			1,03
Canton			X
N Subject	50,871		50,871

Note: H.R. = Hazard Ratio.

* p < 0.05.

** p < 0.01.

*** p < 0.001; Source: Swiss population register.