The author is grateful to the two anonymous reviewers for their constructive suggestions and insightful feedback.
1Administrative procedures, reception policies, and residence permits shape the integration processes of immigrants in destination countries, particularly among asylum-seeking migrants. In Belgium and in other destination countries, previous research has shown that the administrative pathways of immigrants are often complex, with a succession of precarious residence permits and recurrent periods of irregularity [Schoumaker et al., 2022 ; Schuster, 2005]. These elements impact the life course of immigrants in the host country, including their residential mobility, as they create insecurity and place individuals in a position of “permanent temporariness” [Bertrand, 2019 ; Schoumaker et al., 2022].
2Aligning with the life course perspective on internal migration [Bernard, 2022], we consider in this paper that the mobility of asylum migrants does not cease after they apply for protection in the host country. Forced migration is indeed part of a broader mobility sequence as it is often followed by internal or international moves [de Hoon, Vink and Schmeets, 2021 ; Stewart and Shaffer, 2015]. So far, existing research on the connection between geographical mobility and administrative pathways has mostly focused on the transnational mobility of asylum migrants [de Hoon, Vink and Schmeets, 2020 ; Schuster, 2005]. Few studies have taken an interest in the internal residential moves of refugees while also considering their administrative paths. Such research is relevant to acknowledge the impacts of asylum and reception policies on the settlement and integration of asylum migrants. In Belgium, as in other receiving countries, asylum procedures may last a few months up to several years. While waiting for the examination of their application, most asylum seekers are hosted in reception structures. Thanks to their status and rights, recognised refugees will have greater freedom to move than applicants whose procedures are still pending. This work thus focuses on how these elements interact with the internal migrations of recognised refugees. We begin by presenting the main background elements shaping our questions and hypothesis: (1) an overview of existing studies on the impacts of legal trajectories on migrants’ life course, (2) the legal context defining the residence permits of asylum migrants in Belgium, and (3) a brief explanation of the Belgian reception system. Next, we present our hypothesis concerning internal migrations of recognised refugees, and we develop our research design. We adopt a longitudinal and quantitative perspective using population register data. Finally, we present our findings relying on event history analysis methods, exploring the impact of administrative outcomes (i.e. procedure duration, protection obtained, type of reception structure) on internal migration and settlement in Belgium. Our contribution takes advantage of administrative register data, providing unique insights into asylum procedures and settlement paths in Belgium. Our main findings show that being hosted in a collective reception centre significantly increases the likelihood of internal migration after protection recognition, suggesting that individuals prefer to leave the municipalities in which they are hosted towards areas with better opportunities (housing, jobs, social networks etc.). Long asylum procedures decrease the likelihood of migrating internally after protection, whereas the consequence of the reception crisis in 2015 increased the internal mobility among recognised refugees.
3Immigrants often experience turbulent administrative trajectories in destination countries. Many studies report complex pathways with frequent changes in legal status [Bertrand, 2019 ; Merla and Smit, 2023 ; Schuster, 2005] and recurrent episodes of irregularity [Merla, Sarolea and Schoumaker, 2021 ; Schoumaker et al., 2022 ; Vickstrom, 2014]. Depending on their legal status, immigrants have different access to state-organised support and social rights [Bertrand, 2019 ; Jansson, 2024 ; Marlowe et al., 2024 ; Rea et al., 2014]. Less secure residence permits create “legal barriers” that hinder the integration process [Da Lomba, 2010]. Legal provisions are thus referred to as important determinants of settlement and integration in destination countries [Bakker, Dagevos and Engbersen, 2014 ; Farcy and Smit, 2020 ; Söhn, 2019]. Moreover, specifically among asylum seekers, administrative trajectories involve periods of uncertainty and long times waiting for a decision, depending on the duration of the asylum procedure [Bakker, Dagevos and Engbersen, 2014 ; Brekke, 2010 ; Farcy and Smit, 2020 ; Hainmueller, Hangartner and Lawrence, 2016 ; Jansson, 2024]. Existing research often demonstrates the negative outcomes of precarious legal trajectories on various aspects of the life course and integration in destination countries [Brekke, 2010 ; Farcy and Smit, 2020]. Several studies examine the impact of legal provision on the socio-economic outcomes among immigrants [Bakker, Dagevos and Engbersen, 2014 ; Bertrand, 2019 ; de Vroome and van Tubergen, 2010 ; Hainmueller, Hangartner and Lawrence, 2016 ; Söhn, 2019]. Through a quantitative study using population register data from Switzerland, Bertrand [2019] emphasizes the connection between residence permits and labour market integration: having stable residence permits enhances the likelihood of economic integration among asylum-seeking migrants. Also in Switzerland, Hainmueller and colleagues [2016] further add that the longer the administrative procedures, the lower the odds of successful employment integration.
4Other aspects of the life course are also addressed in existing literature, mainly using qualitative approaches. In Belgium, the work of Merla and Smit [2023] explores “the frictions between institutional timescales and migrants’ everyday and biographical timescales” [Merla & Smit, 2023, p. 378]. Their analyses show that the complex administrative trajectories and the temporariness of residence permits constrain migrants’ family plans. Events such as having a child, getting married, or changing homes are put on hold because of legal statuses obtained by household members [Merla & Smit, 2023]. Long waiting periods as well as precarious and temporary legal situations also negatively affect asylum seekers’ mental health, as shown by Brekke in Sweden [2010] and as suggested by Bakker et al. [2014] in the Netherlands, for instance. Indeed, long procedures and uncertain legal situations create stressful situations [Hainmueller, Hangartner and Lawrence, 2016] and might reinforce pre-existing vulnerabilities [Sarolea, Raimondo and Crine, 2021a]. In Belgium, a report by the Superior Health Council also underscored the negative effects of long stays in collective reception centres on refugees’ mental health [Conseil Supérieur de la Santé, 2024].
5Other studies link the administrative pathways of immigrants with their geographical mobility [de Hoon, Vink and Schmeets, 2020 ; Liu and Kulu, 2023 ; Schuster, 2005]. Having stable residence permits is often considered as a way to improve “the mobility capital” [de Hoon et al., 2020]. In other words, having papers induces the ability to project in the future more easily and to become “geographically mobile again” [Schuster, 2005, p. 766]. De Hoon and colleagues use register data to study subsequent migrations among refugees who obtained citizenship in the Netherlands. Although acquiring citizenship comes with facilitated mobility across EU countries, they found that refugees with Dutch citizenship are less likely to leave the Netherlands than refugees without Dutch nationality [de Hoon et al., 2020]. In Italy, Schuster [2005] addressed the issue of rejected refugees who move back and forth between EU countries: their precarious legal position hinders their ability to plan and thus affects their mobility [see also Merla & Smit, 2023]. Research in Germany also highlights the impact of legal status on residential mobility and housing careers among refugees. Compared to other immigrants, refugees are found to be more mobile within the country due to legal barriers they encounter (e.g. housing and labour market restrictions), at least during the first 5 years of residence [Liu and Kulu, 2023]. However, this is a temporary outcome as Liu and Kulu [2023] observe that this residential instability disappears after legal restrictions are no longer in application. The effect of legal and administrative outcomes on mobility trajectories thus seems two-fold: on the one hand, the possibility of moving is enhanced by secure residence permits [Schuster, 2005]; on the other hand, asylum migrants would aspire to some sort of stability after long periods of wait and uncertainty [El Moussawi, 2023].
- 1 Directive 2013/32/EU of the European Parliament and of the Council of 26 June 2013 on common proced (...)
6Asylum procedures in the European Union are regulated by the EU procedure directive, which sets a maximum duration of 21 months1. However, this delay is indicative, and no sanction exists in case of violation of this maximum duration. In Belgium, asylum procedures vary from several months to more than a year before receiving a final decision on the application for international protection. The Superior Health Council mentions a mean duration of 392 days in 2019 [Conseil Supérieur de la Santé, 2024]. Among the cohort studied in this paper (see Data section below), the mean duration of the asylum procedure was 11.5 months and 66.1% of recognised refugees had a procedure lasting between 1 to 12 months (see more details in Appendix I). During this procedure, migrants are entitled to temporary residence permits, renewable until the end of the procedure [Schoumaker et al., 2022]. Initially, they are granted an “orange card”, the validity of which is three months. It can be extended up to three times, each extension lasting an additional three months. Subsequently, if the asylum procedure extends beyond a year, the orange card is renewed monthly until a decision (either positive or negative) is issued by the Aliens Office [ADDE, 2018]. The orange card attests that the asylum procedure is ongoing and that the migrant is registered in the waiting register [Aussems, 2012]. When facing long asylum procedures, individuals thus encounter long periods of uncertainty and difficulties in projecting themselves in the receiving country because of the short residence permits. During the asylum procedure, applicants become eligible for a work permit and may access the Belgian labour market after an employment ban of several months.
- 2 See Art. 49 and 49/2 of the law of 15 December, 1980.
- 3 Law of the 1st June, 2016 amending the law of the 15 December, 1980.
- 4 Note that obtaining the residence permit corresponding to the protection status granted is not inst (...)
7The conditions governing the right of residence of beneficiaries of international protection are set out in the law of 15 December 1980 on access to the territory, residence, settlement, and removal of foreigners2. If granted refugee protection, the migrant benefits from an unlimited right of stay, card B (see example 1 in Table 1). He/she is then registered in the foreign register [Aussems, 2012 ; Rea et al., 2014]. However, this provision changed in July 2016 (example 2 in Table 1): refugee statuses delivered after this date only give the right to a limited permit of 5 years (card A)3. The latter is changed to an unlimited right of residence (card B) after 5 years of stay in the country [CGRA, 2023b]. Recognised refugees benefit from the same social rights as Belgian citizens and are exempt from obtaining a work permit. If granted a subsidiary protection (example 3 in Table 1), the right of residence is, however, limited to one year (card A) [Aussems, 2012]. The card A is renewable twice for periods of two years each. After 5 years of residence in Belgium, beneficiaries of the subsidiary protection obtain an unlimited right of stay, similar to recognised refugees (card B) [CGRA, 2023]. If granted international protection, the residence permit obtained is thus relatively stable (one year or more)4; if the protection is denied (example 4 in Table 1), migrants who stay in Belgium often fall into irregularity as they receive an order to leave the territory. In this situation, migrants become undocumented and are often removed from the National Register even if they stay in Belgium. Otherwise, asylum applicants may also obtain a residence permit through other regularisation procedures (such as family reunification or humanitarian reasons).
Table 1 Illustration of plausible administrative paths
Note: This table gives hypothetical examples of plausible administrative paths among asylum-seeking migrants. ID 1 represents the trajectory of someone granted refugee protection before July 2016, 14 months after applying for international protection; ID 2 represents someone granted refugee protection after July 2016, 3 months after applying for international protection; ID 3 represents someone who receives subsidiary protection, 6 months after applying for asylum; ID 4 represents someone whose asylum request is rejected 15 months after applying.
8Administrative trajectories for asylum-seeking migrants are thus long and, complex, involving a succession of different residence permits. Plus, migration policies in Belgium have undergone many changes, leading toward more restrictive measures and precarious legal situations [Merla and Smit, 2023 ; Sarolea, 2021 ; Schoumaker et al., 2022]. This is the case, for instance, with the 2016 amendment to the 1980 law that replaced the unlimited right of residence for recognised refugees with a limited right. Similar evolutions of the law are observed elsewhere, such as in the UK where the indefinite right of residence has been changed into a temporary stay of 5 years [Da Lomba, 2010]. This reinforces the experiences of the temporariness of the legal statuses among asylum migrants as “the certainty that once came with refugee status is being eroded” [Da Lomba, 2010, p. 425].
- 5 Law of January 12, 2007 on the reception of asylum seekers and other categories of aliens, M.B. May (...)
9Understanding the reception system’s functioning is also crucial to studying the life course of asylum seekers in Belgium, and more specifically, the link between their administrative trajectories and their internal migrations. During the entire asylum procedure, as provided by the 2007 Belgian law5, applicants for international protection are entitled to material assistance, including accommodation.
10Although reception is not stringent (migrants choose to benefit from the material assistance or not), on average 70% of applicants entered the reception network upon their arrival in Belgium among those who arrived between 2012 and 2020 [Zimmer, 2023]. Among the cohort studied in this paper (see Table 2), this share goes up to 80% of asylum migrants hosted in a reception facility at the moment of their protection recognition. The rest is not hosted in the reception system (this situation is called “no show”), mainly because they do not have a right to material assistance, or there are no availabilities in the network. Those who decide not to reside in the allocated reception place lose their right to material assistance and only benefit from urgent medical aid. For those who do enter the network, the Federal agency for the reception of asylum seekers (Fedasil), assigns them to a reception place based on several criteria such as the specific needs of the individuals, their age, gender, family situation, etc. [EMN, 2014]. Reception facilities are either collective or individual. The former, called “open centres” are more common while individual structures, called “local reception initiatives” (LRI) represent a smaller share of the total reception capacity [Zimmer, 2023]. The allocation also largely depends on the current capacity of the reception network [AIDA, 2016]. In 2015, as in many other European receiving countries, Belgium faced an important increase in the number of asylum applications, resulting in high pressure on the reception network. Within a few months, the reception capacity was doubled with the opening of many emergency centres [EMN, 2016].
11With this particular reception system in place, asylum seekers frequently do not choose where they initially live. The number of collective reception centres and their spatial dispersal vary significantly from one year to another [Zimmer, 2023]. Given that the creation of centres generally responds to emergencies [El Moussawi and Schuermans, 2020 ; Wyckaert, Leinfelder and De Decker, 2020], their geographical location is not specifically designed to be the most appropriate for the integration of asylum applicants. Regarding the overall capacity of collective accommodations that were in operation between 2012 and 2020, 35% to 45% of the available beds in open centres are located in rural municipalities [Zimmer 2023]. Qualitative studies highlight the difficult living conditions in the reception centres: limited living spaces, lack of privacy, feelings of insecurity, lack of autonomy (notably, fixed meal schedules, regulated exits from the centres) [Beeckmans and Geldof, 2024 ; Fournier et al., 2023 ; Sarolea, Raimondo and Crine, 2021b], preventing individuals from developing a sense of “feeling a home” [Beeckmans and Geldof, 2024]. At the end of the asylum procedure, if international protection is granted, individuals must leave the reception centre within two months. Unfortunately, the transition from reception accommodation to private housing is not supported by any level of governance in Belgium [Hantson, Westerveen and Adam, 2022]. Nevertheless, the shift in legal status from asylum seeker to recognised refugee implies a change in rights, and individuals have to rapidly find accommodation on their own [Weidinger and Kordel, 2023].
12In this work, we explore the impact of the reception system and the length of administrative procedures on the settlement and the internal mobility of asylum seeking-migrants after protection recognition. Based on the case of Belgium, this research addresses the following question: how do administrative procedures and the functioning of the reception system influence internal migration after international protection recognition? We formulate four main hypotheses using a longitudinal approach to internal migration, linking pre-recognition experiences to residential mobility after protection recognition.
13First, we expect high mobility rates soon after protection recognition, especially among those hosted in the reception system (H1), for two reasons. First, recognised refugees accommodated in collective centres must leave these reception centres within 2 months post-protection, contrasting with the situation of those who are not accommodated at the moment of protection who do not face such constraints. Second, the rural and remote location of many centres constrains migrants from changing municipalities if they wish to benefit from better job or housing opportunities. For instance in Switzerland, the allocated canton of residence affects housing trajectories of asylum migrants as reception conditions differ across the country’s regions [Lacroix and Bertrand, 2024]. Related to this, we expect the place of residence at the time of recognition to influence internal mobility after protection (H2). If people are already residing in urban areas, internal migrations would be less likely compared to those residing in rural areas (H2a). De Hoon and colleagues similarly demonstrated in the Netherlands that refugees allocated to rural accommodation most often change residence over time [de Hoon, Vink and Schmeets, 2021]. Also, people living in Brussels or Flanders would be less likely to move than those living in Wallonia (H2b). Brussels and Flanders usually attract more immigrants than Wallonia, as observed in other studies [Agentschap Binnenlands Bestuur, 2019 ; El Moussawi and Schuermans, 2020 ; Zimmer and Schoumaker, 2025]. Moreover, divergence exists between the regions regarding integration policies [Rea et al., 2014 ; Sarolea, Raimondo and Crine, 2021a], and economic conditions. For example, data from the Belgian Statistical Office (Statbel), derived from the Labour Force Survey, indicates consistently higher employment rates in Flanders than in Wallonia [Statbel, 2024a].
14Third, focusing more specifically on administrative aspects, we expect longer asylum procedures to lower the odds of internal migrations after protection recognition (H3). Asylum migrants would indeed start their integration process during this lengthy waiting period (time to learn the language, to adapt to local context, to develop social networks, etc.). Also, after long periods of uncertainty, migrants would aspire to stabilize their situation in the host country. This would result in them staying in the municipality in which they are residing at the time of protection.
15Finally, we hypothesize that internal migration patterns might vary according to periods of arrival (H4). Those who arrived in Belgium in 2015-2016 would show higher odds of internal mobility after status recognition, compared to those who arrived earlier. On the one hand, Belgium faced a reception crisis in 2015 and 2016. Asylum seekers were no longer hosted according to their specific needs but rather according to the availability in the reception network, as in other countries across Europe [AIDA, 2016]. On the other hand, those who were granted refugee protection after July 2016 were entitled to a limited residence permit of 5 years (contrary to those recognised before July 2016 whose residence permit is unlimited). We argue that the shorter the permit, the greater the uncertainty and the difficulty of projecting oneself in the host country, and the greater the residential instability.
16To examine internal migrations among recognised refugees (post-recognition), we rely on data from the Belgian National Register. This centralised administrative tool collects various information on every legal resident in Belgium. It contains detailed and precisely dated data on the asylum application procedure as well as residential changes. Such data sources are highly useful for studying the migration trajectories of cohorts within a life course framework [Bernard, 2022]. It is exhaustive though only considering formal situations of legal residence and right of residence. People remaining on the territory despite the final rejection of their asylum application are only monitored until the end of their asylum procedure, as they are removed from the administrative registers afterward.
17For this work, we rely on a cohort of asylum applicants who arrived in the country between 2012 and 2016, which we monitor until the end of 2020. More than 75,600 individuals had applied for international protection during this period and 51.1% of them were granted international protection by the end of 2020 (either refugee status before July 2016 (45%), refugee status after July 2016 (30.8%) or subsidiary protection (24.2%)). Table 2 presents the main descriptive statistics of the cohort studied. Individuals aged under 18 at arrival were not considered (N=4,533) nor were individuals born in EU countries (N=87). At the moment of protection recognition, 79.8% of the cohort was hosted in a reception facility, either individual (Local Reception Initiatives) or collective (open centres).
Table 2 Descriptive statistics for the cohort of arrival 2012-2016
|
|
N=33,973
|
%
|
|
Sex
|
|
|
|
Male
|
23,029
|
67.8
|
|
Female
|
10,944
|
32.2
|
|
Place of birth
|
|
|
|
Europe non EU
|
1,876
|
5.5
|
|
Western Asia
|
23,313
|
68.6
|
|
Eastern Asia
|
547
|
1.6
|
|
Northern Africa
|
694
|
2.0
|
|
Subsaharan Africa
|
7,210
|
21.2
|
|
Latin America
|
206
|
0.6
|
|
Unknown
|
127
|
0.4
|
|
Age group at arrival
|
|
|
|
18-19
|
2,570
|
7.6
|
|
20-29
|
16,294
|
48.0
|
|
30-39
|
8,861
|
26.1
|
|
40-49
|
3,759
|
11.1
|
|
50+
|
2,489
|
7.3
|
|
Year of arrival in Belgium
|
|
|
|
2012
|
3,886
|
11.4
|
|
2013
|
3,116
|
9.2
|
|
2014
|
5,363
|
15.8
|
|
2015
|
16,921
|
49.8
|
|
2016
|
4,687
|
13.8
|
|
Type of protection obtained
|
|
|
|
Refugee (before July 2016)
|
16,069
|
47.3
|
|
Refugee (after July 2016)
|
10,260
|
30.2
|
|
Subsidiary Protection
|
7,644
|
22.5
|
|
Type of reception at protection recognition
|
|
|
Collective reception (centre)
|
19,113
|
56.3
|
|
Individual reception (LRI)
|
8,001
|
23.6
|
|
No reception ("no show")
|
6,856
|
20.2
|
|
Length of the asylum procedure
|
|
|
|
0-6 months
|
12,900
|
38.0
|
|
6-12 months
|
9,548
|
28.1
|
|
12 + months
|
8,524
|
34.0
|
|
Country of birth (10 most frequent, representing 80% of the cohort)
|
|
Syria
|
10,433
|
30.7
|
|
Afghanistan
|
5,185
|
15.3
|
|
Iraq
|
4,941
|
14.5
|
|
Somalia
|
1,834
|
5.4
|
|
Guinea
|
990
|
2.9
|
|
Iran
|
910
|
2.7
|
|
Ethiopia
|
893
|
2.6
|
|
Former USSR
|
784
|
2.3
|
|
Palestine
|
724
|
2.1
|
|
Turkey
|
543
|
1.6
|
Source: own calculations based on the Belgian National Register
- 6 The municipality level is the lowest administrative level in Belgium. The country counts 581 differ (...)
- 7 Hazard rates represent the likelihood of an event occurring in a period of time, knowing that the i (...)
18To capture internal migrations after international protection, we measure residential changes at the municipality level6. The timing of these events is measured in continuous time to fully exploit the longitudinal monitoring provided by the register data. We first rely on descriptive analysis using Kaplan-Meier survival curves, smoothed hazard rates7 and cumulative incidence functions to describe the timing of internal migrations among refugees after protection recognition. It contextualises when the mobility takes place in the life course of recognised refugees. We then use Cox proportional hazard models to explain the likelihood of a first internal migration following protection recognition. This research strategy allows us to consider the timing and occurrence of the event of interest, key elements when adopting a life course perspective [Finney and Catney, 2012]. After running graphical checks of the proportionality of hazard curves, we stratified our models according to time periods. The non-proportionality of the hazard through time suggests that different determinants might be at stake at different periods, hence the choice to conduct two separate analyses. We first consider a model for the first 2.5 months after protection recognition (the period during which the hazard rates are the highest, see Results section), and a second analysis from 2.5 months onward. The limit of 2.5 months post-protection is further justified in our study as individuals must leave the Belgian reception facilities within two months.
19Several independent variables are added to the models to test our hypothesis. The year of arrival and the type of protection obtained (either refugee before or after July 2016, or subsidiary protection) are included to verify the effect of the period of arrival and the type of residence permit (1 year, 5 years, or unlimited). Using detailed register information about the asylum procedure, we were able to compute the duration of the asylum procedure, which we categorized in 3 intervals ([0-6 months, [6-12, [12 months and more]. The type of reception at the time of protection recognition is included to measure differences between those accommodated in collective or individual structures and those who are not accommodated in the reception network (“no show”) [see Zimmer, 2023]. Also, two characteristics of the municipality of residence at the time of recognition are included: the region of residence on the one hand and the level of urbanisation on the other hand. Finally, we control for gender, region of birth, and age group at arrival.
20Figure 1 represents the Kaplan-Meier survival curve (Fig 1a) and the smoothed hazard rates (Fig 1b) for the first internal migration after recognition. Note that changes of residence within the same municipality are not considered in this study. Also, it is important to acknowledge that the changes of municipality of residence we measure could reflect short-distance moves. The risks of migrating internally are highest in the first 3 months after being granted protection, and about 75% of the cohort had moved internally within the first semester. In other words, there is a high probability of leaving the municipality of residence when an international protection is recognised, which supports our first hypothesis (H1). This is, however, less pronounced among those not accommodated in a reception facility, of whom only 25% changed municipality of residence within the first quarter after recognition.
Figure 1 First internal migration after protection recognition
Note: N=33,973 individuals who arrived between 2012 and 2016 in Belgium and who obtained international protection before the end of 2020.
Source: own calculations based on Belgian National Register
- 8 Cumulative Incidence Functions (CIF) show the cumulative proportion of individuals encountering a c (...)
21Figure 2 breaks down these descriptive findings by region of residence at protection and by region of destination of the first internal move. The Kaplan-Meier survival curves by region of residence (Fig 2d.) present the share of individuals who did not migrate internally through time. From Flanders and Brussels, only 17% of individuals stayed in the municipality of residence during the observation period, but this share is even smaller among those living in Wallonia (6.8%). Then, we compare the directions of the moves according to the starting point (the municipality of residence at protection recognition). Figures 2a to 2c show cumulative incidence functions (CIF)8 according to the region of residence at the time of recognition. From Wallonia (Fig 2b), internal migrations are mainly directed to Flanders and Wallonia, whereas from Flanders (Fig 2a), internal moves are almost exclusively directed to Flemish municipalities. At around 4 months after protection, 50% of those who were living in Flanders had changed municipalities within Flanders which highlights how rapidly moves are occurring after recognition. Concerning the level of urbanisation of the municipality of residence (Appendix II), internal migrations to urban municipalities are most common regardless of the place of residence at recognition. Although only 20% of those living in urban municipalities stayed in their initial municipalities, this share is even smaller among those living in rural areas: only 6.8% did not leave their rural municipality.
Figure 2 First internal migration post-protection by region of residence
Note: N=33,973 individuals who arrived between 2012 and 2016 in Belgium and who obtained international protection before the end of 2020.
Source: own calculations based on Belgian National Register
- 9 Due to the hazard proportionality issues, we tested the robustness of our results using binary logi (...)
22To examine what determines the 1st internal migration, we focus on the first 2.5 months and conduct Cox regression models9. Each model evaluates the effect of different independent variables according to each of our hypothesis. We do so as we expect some variables to mediate other variables’ effect if included in a same model (e.g. the place of residence at recognition is expected to mediate the effect of the type of reception at recognition, hence Model 1 does not adjust for the characteristics of residence). Yet, results are found to be relatively consistent across models, advocating for the robustness of the effects observed.
23Around 52% of our sample migrated internally within the first 2.5 months post-protection (Fig 1a). We present in Table 3 the hazard ratios of each regression model (95% confidence intervals are found in Appendix III). In model 1, the type of reception at recognition shows a clear pattern: those who were not accommodated in the reception network present 64% lower hazards of changing municipality compared to those accommodated in collective reception facilities; those hosted in individual reception facilities present 48% less hazard of moving. This is in line with H1 as recognised refugees tend to quickly change municipality of residence after protection recognition, especially among those living in collective reception centres. We observe that 67.3% of the latter had migrated within the first 2.5 months.
24As predicted by our descriptive findings, Models 2.1 and 2.2 show statistically significant differences according to the place of residence at recognition. Those living in Flanders and Brussels at the time of recognition have lower hazards of changing municipality than those living in Wallonia (13% lower in Flanders, and 11% in Brussels compared to Wallonia). We also find differences according to the level of urbanisation: those living in urban municipalities are 12% less likely to migrate internally during the first 2.5 months compared to those living in rural regions. Although estimated differences remain limited, these findings are in line with H2. Internal migration trajectories after being recognised as refugees differ according to the place of residence at the time of protection recognition. In other words, the place where asylum migrants are hosted plays a significant role in shaping subsequent residential moves in the country.
- 10 Models 3, which included the asylum procedure’s duration, was also tested by including the variable (...)
25Model 3 shows that the duration of the asylum procedure also affects the likelihood of migrating internally after obtaining protection in the first 2.5 months post-protection. Those whose procedures lasted 6 to 12 months present lower hazards (-0.9%) of moving internally compared to those with very short procedures (less than 6 months). On the other hand, those whose procedures lasted more than a year encountered a slightly higher hazard of first internal migration (+3%) compared to procedures between 0 and 6 months. This last result is, however, not statistically significant. Our hypothesis H3, according to which longer procedures induce lower mobility, finds partial confirmation in the first 2.5 months post-protection recognition10.
26Differences according to the year of arrival are as expected (H4). Model 4 shows that those who arrived more recently have higher hazards of migrating internally. Hazards are 20% lower among those who arrived in Belgium between 2012 and 2014, in comparison to 2015. This might be a consequence of the reception crisis occurring in 2015 and 2016, during which the reception network was saturated, and the accommodations were first and foremost assigned based on availabilities rather than on individuals’ needs [Zimmer, 2023]. This would explain why individuals who arrived during the crisis are more at risk of leaving the municipality of reception when obtaining international protection. Regarding the type of protection obtained (see Model 2.2), those recognised as refugees following the change of law in July 2016 present higher hazards (1.23) of internal migration compared to refugees recognised before this date. This might be a result of the greater precariousness of the residence permit introduced by the new law, with residence permits of a 5-year validity (instead of an unlimited permit). With limited residence permits, individuals would remain in an uncertain position and, therefore find it difficult to settle. That said, this result could also indicate an interaction with the year of arrival, as those who arrived later were more likely to receive their status more recently and, consequently, after the law was amended.
Table 3 Cox regression models on the first internal migration within the first 2.5 months post-protection
|
|
Hazard ratio
|
|
|
Model 1
|
Model 2.1
|
Model 2.2
|
Model 3
|
Model 4
|
|
Sex
|
|
Male
|
ref.
|
ref.
|
ref.
|
ref.
|
ref.
|
|
Female
|
0.93***
|
0.93**
|
0.93***
|
0.93***
|
0.86***
|
|
Region of birth
|
|
Europe non-EU
|
0.79***
|
0.79***
|
0.79***
|
0.79***
|
0.69***
|
|
West Asia
|
ref.
|
ref.
|
ref.
|
ref.
|
ref.
|
|
East Asia
|
1.24***
|
1.26***
|
1.27***
|
1.27***
|
1.21***
|
|
North Africa
|
0.98
|
0.97
|
0.97
|
0.98
|
0.94
|
|
Sub Saharan Africa
|
0.97
|
0.97*
|
0.96
|
0.96**
|
1.00
|
|
Latin America
|
0.64***
|
0.64**
|
0.64**
|
0.64***
|
0.46***
|
|
Age group at arrival
|
|
18-19
|
1.05
|
1.06
|
1.05
|
1.06*
|
0.99
|
|
20-29
|
ref.
|
ref.
|
ref.
|
ref.
|
ref.
|
|
30-39
|
0.96***
|
0.96*
|
0.96**
|
0.96**
|
0.93***
|
|
40-49
|
0.93***
|
0.93**
|
0.93***
|
0.92***
|
0.89***
|
|
50+
|
0.88***
|
0.88***
|
0.88***
|
0.88***
|
0.81***
|
|
Year of arrival
|
|
2012
|
0.86***
|
0.89***
|
0.86***
|
0.85***
|
0.79***
|
|
2013
|
0.90***
|
0.92***
|
0.92***
|
0.91***
|
0.79***
|
|
2014
|
0.97
|
0.97
|
0.97
|
0.96*
|
0.82***
|
|
2015
|
ref.
|
ref.
|
ref.
|
ref.
|
ref.
|
|
2016
|
1.11***
|
1.12***
|
1.12***
|
1.12***
|
1.13***
|
|
Type of protection
|
|
Refugee (before July 2016)
|
ref.
|
ref.
|
ref.
|
ref.
|
|
|
Refugee (after July 2016)
|
1.23***
|
1.23***
|
1.23***
|
1.20***
|
|
|
Subsidiary protection
|
1.09***
|
1.09***
|
1.09***
|
1.08***
|
|
|
Type of reception at protection recognition
|
|
Collective reception
|
ref.
|
ref.
|
ref.
|
ref.
|
|
|
Individual reception
|
0.52***
|
0.54***
|
0.53***
|
0.53***
|
|
|
No reception ("no show")
|
0.36***
|
0.37***
|
0.38***
|
0.38***
|
|
|
Characteristic of residence at protection recognition
|
|
Region
|
|
|
|
|
|
|
Brussels
|
|
0.89***
|
0.97
|
0.96
|
|
|
Flanders
|
|
0.87***
|
0.89***
|
0.89***
|
|
|
Wallonia
|
|
ref.
|
ref.
|
ref.
|
|
|
Level of urbanisation
|
|
Urban
|
|
|
0.88***
|
0.88***
|
|
|
Suburban
|
|
|
0.98
|
0.98
|
|
|
Rural
|
|
|
ref.
|
ref.
|
|
|
Duration of asylum procedure
|
|
0-6 months
|
|
|
|
ref.
|
|
|
6-12 months
|
|
|
|
0.91***
|
|
|
12 + months
|
|
|
|
1.03
|
|
Note: N=33,973 individuals who arrived between 2012 and 2016 in Belgium and who obtained international protection before the end of 2020. 50.7% of the cohort had migrated internally within the first 2.5 months post-protection. Each model tests a different hypothesis with a different main explanatory variable highlighted in bold.
*** p<0.01; ** p<0.05; * p<0.1
Source: own calculations based on Belgian National Register
27As the hazards are not proportional through time (see Method section), we conducted a separate analysis of the occurrence of a first internal migration post-protection by looking at the period from 2.5 months post-protection and onward. Again, we apply Cox proportional hazard models using the same covariates as in the previous section. These models are only applied to recognised refugees who did not migrate internally within the first 2.5 months of observation (49.3% of the cohort). Table 4 presents the hazard ratios (95% confidence intervals can be found in Appendix V).
28The type of reception (either collective, individual or no reception) remains a significant predictor of the first internal migration (Model 1). Those who are not accommodated in a reception structure are much less likely to migrate internally (65% lesser hazard) compared to those hosted in collective reception facilities. The effects of the place of residence are even stronger than for the first 2.5 months post-protection recognition. Living in Flanders reduces the hazard of migrating internally by 33% compared to living in Wallonia (Model 2.1). Living in an urban area reduces the hazard of migrating internally by 43% compared to living in rural areas (Model 2.2). In other words, living in Wallonia and/or in rural areas raises the hazard of internal migration. Concerning the asylum procedure’s duration (H3, see Model 3), results show a clearer pattern. Those with the longest asylum procedures encounter lower hazards (0.88) of internal migration compared to those with very short procedures. In other words, there is a gradient according to the duration of the procedure, as longer procedures present lower hazards of first internal migration.
29From 2.5 months onward, the effect of the year of arrival loses its significance (Model 4). Individuals who arrived in Belgium in 2016 now have lower chances of changing municipality compared to those who arrived in 2015. Similarly, the effect of obtaining refugee status after July 2016 is now reversed: the hazard is lower than for refugees recognised earlier. Therefore, the fact of arriving later in Belgium and being recognised later seems to encourage internal migration mainly in the period soon after protection recognition. This illustrates that the hazards of internal migration are not proportional across time, and different determinants are at stake at the exit of the reception system versus later in the life course in Belgium. Indeed, for individuals staying in collective reception facilities, there are strong incentives to change municipalities as they are forced to leave the reception accommodation shortly after recognition.
Table 4 Cox regression models on the first internal migration from 2.5 months post-protection onward
|
|
Hazard ratio
|
|
|
Model 1
|
Model 2.1
|
Model 2.2
|
Model 3
|
Model 4
|
|
Sex
|
|
|
|
|
|
|
Male
|
ref.
|
ref.
|
ref.
|
ref.
|
ref.
|
|
Female
|
0.94**
|
0.93**
|
0.93***
|
0.93***
|
0.79***
|
|
Region of birth
|
|
|
|
|
|
|
Europe non-EU
|
0.83***
|
0.83***
|
0.82***
|
0.84***
|
0.71***
|
|
West Asia
|
ref.
|
ref.
|
ref.
|
ref.
|
ref.
|
|
East Asia
|
0.86
|
0.94
|
1.02
|
1.04
|
0.66***
|
|
North Africa
|
0.99
|
0.93
|
0.94
|
0.97
|
1.03
|
|
Sub Saharan Africa
|
1.09***
|
1.03
|
1.02
|
1.04
|
1.19***
|
|
Latin America
|
0.82
|
0.74*
|
0.73*
|
0.76*
|
0.56***
|
|
Age group at arrival
|
|
|
|
|
|
|
18-19
|
1.02
|
1.01
|
0.98
|
0.97
|
0.96
|
|
20-29
|
ref.
|
ref.
|
ref.
|
ref.
|
ref.
|
|
30-39
|
0.87***
|
0.87***
|
0.87***
|
0.87***
|
0.85***
|
|
40-49
|
0.87***
|
0.85***
|
0.83***
|
0.83***
|
0.82***
|
|
50+
|
0.87***
|
0.86***
|
0.87***
|
0.86***
|
0.75***
|
|
Year of arrival
|
|
|
|
|
|
|
2012
|
0.99
|
1.00
|
1.01
|
1.03
|
0.96
|
|
2013
|
0.99
|
1.02
|
1.05
|
1.05
|
0.91**
|
|
2014
|
1.08**
|
1.11***
|
1.1***
|
1.10**
|
1.01
|
|
2015
|
ref.
|
ref.
|
ref.
|
ref.
|
ref.
|
|
2016
|
0.85
|
0.98
|
1.00
|
0.97
|
0.65***
|
|
Type of protection
|
|
|
|
|
|
|
Refugee (before July 2016)
|
ref.
|
ref.
|
ref.
|
ref.
|
|
|
Refugee (after July 2016)
|
0.79***
|
0.79***
|
0.79***
|
0.85***
|
|
|
Subsidiary protection
|
0.99
|
0.99
|
0.98
|
1.00
|
|
|
Type of reception at protection recognition
|
|
Collective reception
|
ref.
|
ref.
|
ref.
|
ref.
|
|
|
Individual reception
|
0.86***
|
0.96
|
0.86***
|
0.89***
|
|
|
No reception ("no show")
|
0.35***
|
0.34***
|
0.36***
|
0.35***
|
|
|
Characteristics of residence at protection recognition
|
|
Region
|
|
|
|
|
|
|
Brussels
|
|
0.95
|
1.29***
|
1.29***
|
|
|
Flanders
|
|
0.67***
|
0.70***
|
0.70***
|
|
|
Wallonia
|
|
ref.
|
ref.
|
ref.
|
|
|
Level of urbanisation
|
|
Urban
|
|
|
0.57***
|
0.57***
|
|
|
Suburban
|
|
|
1.02
|
1.02
|
|
|
Rural
|
|
|
ref.
|
ref.
|
|
|
Duration of asylum procedure
|
|
0-6 months
|
|
|
|
ref.
|
|
|
6-12 months
|
|
|
|
0.92***
|
|
|
12 + months
|
|
|
|
0.88***
|
|
Note: N=16,749 individuals who obtained international protection before the end of 2020 and who did not migrate internally during the first 2.5 months post-protection. Among the entire cohort of recognised refugees who arrived between 2012 and 2016, 49.3% of the cohort had not yet migrated internally within the first 2.5 months post-protection. Each model tests a different hypothesis with a different main explanatory variable highlighted in bold.
*** p<0.01 ; ** p<0.05 ; * p<0.1
Source: own calculations based on Belgian National Register
30In this study, we examined the internal migrations of asylum-seeking migrants in Belgium after their international protection is recognised, focusing on the effects of asylum procedures and locations of the reception system. The use of longitudinal register data made it possible to place internal migration events in their institutional and residential context, considering the residence at recognition and the moment when the event occurs in the life course of individuals. Internal migration was measured as a change of municipality of residence. While such moves may occasionally reflect short-distance relocations, they nonetheless represent a meaningful change in context, as they entail new administrative registrations and renewed interactions with local welfare institutions. It also interrogates the capacity of the reception policies to retain refugees locally.
31Our findings reveal that most refugees leave the reception municipality shortly after recognition, with the highest migration rates occurring within the first three months. Nearly 75% of the cohort had relocated within six months post-protection, showing that mobility trajectories continue after the forced migration and applying for protection. This is even more rapid for those hosted in collective centres, as 75% had changed municipalities within 3 months. Indeed, individuals staying in reception facilities, particularly collective centres, show significant differences compared to those not hosted: the likelihood of changing municipality is higher among the former across all our models. These results underscore the significant influence of the initial reception location on internal migration trajectories, as those not living in a reception facility are much less likely to migrate internally after recognition.
32Characteristics of the municipality of residence (urbanisation and region) influence internal moves after protection, as residing in Wallonia and/or in rural areas appears to increase the hazard of internal migration. Refugees commonly relocate to urban areas and/or to Flanders and Brussels. The attractivity of the Flemish and Brussels regions might be related to several factors: e.g. proximity with co-ethnic communities [de Hoon, Vink and Schmeets, 2021 ; Saeys, Vandervoordt and Verschraegen, 2018 ; Stewart and Shaffer, 2015]; differences in socio-economic context (higher employment rates in Flanders [Statbel, 2024a], higher risk of poverty or social exclusion in Wallonia compared to Flanders [Statbel, 2024b]); differences in integration policies across regions [Rea et al., 2014]]. We also find that the duration of the asylum procedure impacts subsequent internal migrations. Particularly among recognised refugees who did not migrate during the first 2.5 months of observation, longer asylum procedures are associated with a decreased likelihood of individuals leaving the reception municipality after protection recognition.
33Our analysis underlined that the determinants of internal migration vary over time, with different mechanisms shaping mobility at the exit from the reception system and at later stages of the life course in Belgium. In the first 2.5 months, internal moves are largely driven by institutional constraints linked to the reception system. Descriptive statistics support this interpretation, as individuals who migrated early are significantly more represented in the reception network at protection recognition. 90.5% of the early movers were residing in a reception facility, either collective or individual (Table 5). By contrast, the second set of models concerns a selected population that did not move early and is therefore likely more territorially anchored. Among the latter, mobility is more strongly associated with the duration of the asylum procedure and with characteristics of the place of residence, suggesting that accumulated experiences and local contexts play a greater role once the immediate institutional “push” has faded. At this later stage, internal moves may also reflect strategic relocation choices aimed at improving integration prospects.
Table 5: Comparative table of the sub-populations considered in the Cox models
|
Full sample
|
Migrated within the first 2.5 months
|
Did not migrate in the first 2.5 months
|
|
N
|
33,973
|
17,224
|
16,749
|
|
Type of reception at protection recognition
|
|
No reception
|
20.2
|
9.5
|
31.7
|
|
Collective reception
|
56.3
|
73.2
|
38.1
|
|
Individual reception
|
23.6
|
17.3
|
30.2
|
|
Total
|
100
|
100
|
100
|
Source: own calculations based on Belgian National Register
34Based on this research, we argue that reception facilities' locations often do not fit the needs of recognised refugees, as they move out of the reception municipality once recognised. The mobility of recognised refugees could be interpreted as an adaptation or integration strategy, contrasting with the location of the reception facilities. On another note, a long stay in this reception accommodation, induced by a long asylum procedure, increases the likelihood of staying in this municipality of reception. Because of the numerous negative effects of long procedures on asylum migrants’ lives shown in previous studies [Bakker, Dagevos and Engbersen, 2014 ; Brekke, 2010 ; Hainmueller, Hangartner and Lawrence, 2016 ; Sarolea, Raimondo and Crine, 2021a], our contribution indicates that policies should strive both for shorter asylum procedures as well as reception accommodations more adapted to the refugees’ needs. Our quantitative analysis of detailed register data provides valuable insights into how the reception system—and particularly the dynamics of the 2015 reception crisis—affects recognised refugees’ subsequent settlement. Again, the results show that a reception system not suited to the asylum migrants’ needs (e.g., mainly in collective structures, located in rural areas, etc.) increases the likelihood of internal movement post-protection, potentially delaying integration processes. This is even more relevant in contexts of reception crisis, such as in 2015 or since 2021 [Tatti and Doutrepont, 2024], during which places are allocated based on availabilities rather than on specific needs.
35Our research focuses on recognised refugees and considers only their first relocation after receiving protection, leaving out other relevant periods. For instance, future research should expand on this by examining internal moves during the asylum procedure itself. This would allow for a more comprehensive understanding of the long-term spatial integration of asylum migrants in Belgium, strengthening the life course approach, and providing deeper insight into the specific situation of individuals not residing in reception facilities. Furthermore, other determinants not considered here may play a role as well: employment and family; proximity to co-ethnic communities; and availability on the housing market, among others. By focusing on asylum procedures and the role of the reception system, this research contributes to the body of literature on internal migration among refugees in Europe. Despite the different ways in which European Union directives on reception of asylum migrants are implemented in the member states, the implications of this research are not only relevant to the Belgian context, but also to other EU member states. Our findings underscore the complex interplay between administrative factors and spatial mobility, suggesting that both the location and duration of asylum seekers’ initial reception have long-lasting effects on their residential trajectories in the host country.
Flux RSS
