For the Sake of Our Children? A Mixed-Methods Study of the Family Dynamics of Intra-European Mobility Among Somalis

Abstract

This chapter adds to a young body of literature exploring the patterns of and motivations for the intra-EU mobility of non-EU migrants. As explanatory factors of these mobilities, recent studies point to the constrained access to work, networks and institutions in the country of residence. The role of children in onward mobilities has received little attention thus far. While offspring are generally theorised as a ‘binding’ factor, research suggests that children can also be central in onward-mobility decision-making. Parents take children’s specific needs and their hopes for their children’s future into consideration within the decision of onward mobility. Children can also move on their own account or stay put while their parents move again. We present a case study to unveil the dynamics within Somali families around the decision to move onwards to the UK. This provides a rich and suitable case, because of the relatively high onward-mobility rates and the great variety in family composition and characteristics. Using a mixed-methods design, we develop, test and contextualise theoretical expectations concerning the relationship between the timing of arrival of children in the host country and their onward migration, with or without their parents. In addition, we empirically address the expectation that independent migration is more common in larger families. A comparison with other young people of refugee background is provided to shed light on the particularities and generalities of our findings.

3.1 Introduction

In the literature, the intra-EU mobility of citizens born outside Europe is predominantly understood as a response to the limitations of their lives as economically and culturally marginalised citizens (Van Liempt, 2011a, b; Toma & Castagnone, 2015; Ahrens et al., 2016; Della Puppa, 2018a; Ortensi & Barbiano di Belgiojoso, 2018; Ramos, 2018; De Hoon et al., 2019; Moret, 2020). While arrival in Europe meant a loss of socio-economic status for many, onward mobility¹ provides migrants with opportunities to access better positions (Van Liempt, 2011a; Moret, 2020). Most empirical contributions assume that these mobility practices are part of a family strategy. It is therefore surprising that little attention has been paid to intergenerational dynamics within families around onward-mobility projects within Europe (see also Chap. 6 by Mingot, Chap. 9 by Della Puppa and Sredanovic and Chap. 11 by Formenti).

The existing literature (e.g. Djajić, 2008) often highlights the desire of parents to keep the family united when (re-)migrating. This is in keeping with neoclassical migration theory, in which female spouses and their children are seen as ‘assets’ that simply move together with the male breadwinner. Recent studies, however, paint a more complex picture, in which all members of the family can be proactive in migration decision-making, with their unique roles, aspirations, needs and opportunities (Kelly, 2013; Toma & Castagnone, 2015; Serra Mingot & Mazzucato, 2017; Ortensi & Barbiano di Belgiojoso, 2018). Such dynamics may result in temporary family separation and the creation or continued existence of a transnational family.

Even when a more complex picture of migration planning is provided, the position of the children in this process remains underexposed. Studies that do highlight the role of children in migration predominantly point to the ‘binding effect’ of children in the host country, in particular when they are enrolled in school (Dustmann, 2003a; Djajić, 2008; Trevena et al., 2013; Erdal & Ezzati, 2015). Yet, an emerging body of literature argues that children may also form an important reason to migrate again, especially when relocation opens up new educational opportunities that foster intergenerational social mobility (Hagelskamp et al., 2010). Moreover, parents may envision a transnational (or even a ‘cosmopolitan’) future for their children that is not always catered for in the country of initial settlement (Van Liempt, 2011c; Della Puppa & King, 2018). We seek to contribute to this debate by focusing on the intergenerational dimension within families and presenting a case study that increases our understanding of when, why and with whom Somali children and their family members move onwards.

The onward mobility of Dutch Somali families to the United Kingdom (UK) provides a rich and suitable case, because of the relatively high onward-mobility rates among both single persons and families. Moreover, there is a great variety in family composition and the presence, age and number of children upon arrival in the Netherlands. Although their onward mobility has been extensively documented, both quantitative research designs (e.g. Van den Reek & Hussein, 2003; Lindley & Van Hear, 2007) as well as qualitative contributions (Van Liempt, 2011b) have largely overlooked the intergenerational dimensions of onward-mobility planning. Using a mixed-methods design, we bring together two high-quality bodies of data to explore the drivers of and dynamics within onward-mobility projects in Somali families.

This chapter continues, firstly, with a discussion of the phenomenon of onward mobility in relation to children’s education and upbringing. Secondly, we briefly explain the historical and institutional context of Somali migration to the Netherlands and, thirdly, explain the quantitative and qualitative data that we employ in this study. In the penultimate section, analyses of register data and interview data in combination provide empirical insights into the intergenerational dimensions around onward-mobility planning. These findings are summarised and discussed in the last section.

3.2 Family Onward Mobility

Various studies examine how migration is informed by gendered obligations and expectations – for example, the idea that men are expected to provide for the family (Van Dalen et al., 2005; Cooke, 2008; Jolivet, 2020). Men’s preferences, however, do not always dominate in the migration decision-making, as illustrated by cases in which women are both the carers for and the breadwinners of the transnational family (Baldassar & Merla, 2013; Serra Mingot & Mazzucato, 2017). Women are shown to be active participants in onward migration decision-making, not only because of their employment (potential) (Ortensi & Barbiano di Belgiojoso, 2018) but also in their voicing of concerns regarding their offspring (Lindley & Van Hear, 2007).

The literature shows a general preference among mothers to remain in the country of settlement and to avoid disruptions in their children’s schooling (Lindley & Van Hear, 2007; Trevena et al., 2013). More generally, research suggests that, once family unification is established, the presence of children enhances the integration of the family in the country of residence, making the onward mobility of the family or one of its members less likely (Dustmann, 2003a; Erdal & Ezzati, 2015; Sajons, 2016). Mobility is often seen by parents as disruptive and hampering the integration of their children into schooling systems that are confined to national borders (Erdal & Ezzati, 2015; Trevena et al., 2013). Indeed, educational systems are generally not designed to support children who live transnational lives (Van Geel, 2019). This generates the expectation that children who arrived in the host country at a young age will, instead of moving again in the near future, be prone to remain, together with their parents, at least until successful completion of their education (Hypothesis 1a).

A key argument in these studies is that subsequent family migration² is not only driven by the economic perspectives of the parent(s). In addition, concerns about the offspring are included in the equation, with or without their direct involvement in the migration decision (Dustmann, 2003a; Bushin, 2009; Hutchins, 2011). The theoretical mechanism, as developed in economic contributions, implies that parents’ migration decisions are informed by the expected future educational and economic career(s) of their child(ren) (Dustmann, 2003a; Djajić, 2008). Dustmann (2003a) finds that the birth of a child generally reduces the likelihood of parents returning to their country of origin. He also concludes that this effect is stronger for boys than for girls, which is interpreted as evidence that parental concerns about the future of their offspring depend on the sex of the child, with a greater emphasis on the economic performance of male offspring in the host country.³ While these results form an important addition to studies departing from a single (male) breadwinner model, a sole focus on labour-market performance is too narrow to adequately capture the multi-layered family considerations behind migration.

3.2.1 A Family Approach to Onward Mobility

We intend to build on the general theoretical model put forward by Dustmann (2003b) and Djajić (2008) by assessing some of the critical underlying assumptions. First, emigration from a host country is by no means necessarily directed towards the origin country. Migration to a third country appears to be more common than return among specific groups of migrants, as found for forced migrants in the Netherlands (De Hoon et al., 2019). Second, the implicit assumption that children always move together with their parents should be put up for scrutiny. Instead of remaining a family unity, one of the parents may relocate after migration, while the spouse and (some of) the children remain in the host country. Moreover, children can migrate independently of their parents. To address these two shortcomings in the literature, we analyse onward mobility in the context of the family, allowing observations of both independent onward mobility as well as onward mobility with one or both parents. In the remainder of this section, we develop refined hypotheses before introducing our empirical case to study intergenerational dynamics in onward mobility.

3.2.2 Timing of Entry and Educational Trajectories

It is a foregone conclusion that school systems matter for children’s educational trajectories (Entorf & Lauk, 2008; Crul, 2013; Van de Werfhorst, 2019). In a tracking tradition such as the Dutch educational context, children are placed, at a fixed age, into a specific track for all subjects. Track placement is primarily based on a standardised test but is known to also depend on parents’ ability and efforts to influence this outcome (Forster & van de Werfhorst, 2020). Parental resources such as their own educational attainment as well as their knowledge of the educational system all affect their children’s initial placement and their likelihood of successfully (upwardly) navigating the system. Consequently, children with a migration background⁴ are generally placed in lower tracks and more often ones lower than those corresponding to their test scores, compared to their ‘native’ counterparts (Crul, 2013).

Tracking implies that underperforming children, for example those who did not manage to realise their full potential up until the moment of tracking, are sorted into streams that offer poor opportunities to move upwards and access higher education. For migrants, the timing of arrival therefore largely determines the opportunities available to make up for the learning disabilities that arise from language deficits and interruptions in schooling. Arriving in the host country at a later age, particularly at age 12 or older, can thus be detrimental. Moreover, the ‘side-tracking’ of migrant children aged between 12 and 18, which is common in European education systems, tends to negatively affect the school attainment of refugee children in particular (Crul et al., 2017; Folke, 2018; Koehler & Schneider, 2019). In the Dutch context, this side-tracking (Internationale schakelklas, ISK), which targets ‘first-generation’ migrant children, generally offers a poor prelude to mainstream education, limiting their ‘upstreaming’ possibilities. Arriving at an older age may thus spur a desire to increase opportunities elsewhere, for example in a more comprehensive educational system.⁵

There is a more general argument to be made concerning the timing of arrival in the host country and the mobility trajectories of migrant children. The literature reveals a positive association between children’s age at arrival and the return intentions (of their parents). The general theoretical reasoning behind this implies that the younger children are at the moment of entry into the new society, the stronger their orientation towards this country and the less likely they are to intend to forsake it (Erdal & Ezzati, 2015). In what follows, an alternative hypothesis concerning age at arrival and onward mobility is put forward.

3.2.3 Timing of Entry and Onward Mobility

The interplay between migration and the educational trajectories of children alters when considering onward migration, instead of return, as an alternative to remaining in the host country. Where new migration projects are concerned, there are indications that the presence of school-aged children in the household and concerns about their education spur rather than obstruct migration (Lindley & Van Hear, 2007; Ahrens et al., 2016; Della Puppa & King, 2018; Ortensi & Barbiano di Belgiojoso, 2018). Indeed, educational experiences in the country of origin and opportunities in other locations may form the very reason why parents engage in onward mobility.

Della Puppa and King (2018) illustrate how Italian Bangladeshi families decide to move to the United Kingdom because of their dissatisfaction with the Italian education system and their aspirations to raise and educate their children in an English educational system in anticipation of a global future. We expect such aspirations to be shared, particularly among families in protracted refugee situations, as their near and presumably also long-term futures will unfold in exile, in a ‘transnational social field’ (Levitt & Glick Schiller, 2004). One can argue that migrating once again may be the most fruitful for children when they are young, as they adapt relatively easily to new (educational) environments. This generates the expectation that the younger children are at arrival, the more likely they are to engage in onward mobility (Hypothesis 1b).

While family unity may be the preferred option of most families, this condition is neither easily achieved nor automatically maintained. As such we also explore the possibility of children migrating on their own.⁶ In line with Djajić (2008), we expect such forms of independent migration of children to be the most common in the case of larger families, because of the variation in the timing of migration between the different members of the family (Hypothesis 2). These independent migrations are likely to take place at critical moments in the educational trajectory, resulting in twists in migration propensities over time, rather than a linear relationship between the age of the child and migration propensities.

3.3 Case Description: The Onward Mobility of Dutch Somalis

Migration from Somalia to the Netherlands has a relatively short history, starting from the late 1980s when the Somali civil war forced large numbers of the population to flee to neighbouring or more-distant countries. More people were forced across the borders after military dictator Siad Barre was overthrown by a coup in 1991. Largely motivated by the search for asylum, Somalis went to live in various regions and countries across the globe, from the Middle East to North America, Australia and Europe. The Netherlands was a main destination for Somalis⁷ within Europe as a result of its asylum policies and procedures⁸ but in other respects this country was not necessarily the most obvious place to settle (see also Van Liempt, 2011a). Many Somalis speak English and are used to the British (education) system as a result of a shared colonial past. The Somali population in Europe is a young population, due to the arrival of many young asylum-seekers and the high birth rate of the settled community.

Families may travel together to exercise their right to asylum. Often, however, one member travels ahead to lodge an individual asylum claim; upon receiving a positive answer, he or she then applies for family reunification. For Somalis – and refugees more generally – the papers necessary to prove their identity are not easily obtained, which may complicate family reunification or even completely rule out this possibility. About a quarter of all Somalis who arrived in the period concerned came under a family reunification regulation. In contrast to other migrant groups, family reunification often implied that the husband and one or more children joined the female first applicant (Van Liempt, 2007). Moreover, the widespread practice of shared parenthood results in children being reared and supported by member(s) of the family network who are not necessarily their biological parents (Orellana et al., 2001). This combination of Somali familyhood and the imposition of the ‘nuclear family paradigm’ through Western family-reunification policies has resulted in the creation and maintenance of transnational families. Research in various contexts indeed indicates that Somali families are characterised by transnationalism, with family members often moving back and forth between different nation states (Al-Sharmani, 2004). We thus assume families to be only partially (and perhaps temporarily) reunited in the new country of residence, which has implications for their onward mobility process.

Being granted a residence permit in the Netherlands marks the end of a period of insecurity and forced passivity and of the educational interruptions experienced by children – caused by both their forced international migration and the multiple involuntary relocations that followed between asylum places across the country. Based on the institutional school arrangements (see Sect. 3.2.2), we mark four critical stages in children’s careers: pre-school (up to age 5), obligatory primary schooling (age 5–12), secondary education (from age 12) and, finally, the transition from education to the labour market (different ages, depending on the type of education). Studies concerning the educational attainment of refugee groups demonstrate that young Somali adults mostly obtain a diploma of the lower levels of vocational education – generally resulting in limited labour-market prospects – and that the share of dropouts is remarkably high in comparison to other pupils with a refugee background (Dourleijn et al., 2011). Whether the educational careers of Somali children are (consequently) re-routed across the national border is empirically explored in the next sections of this chapter.

3.4 Data and Methods

Population register data and data from ethnographic fieldwork are used to gain broad as well as in-depth insights into families’ onward mobility decision-making and outcomes. One may at first question the value of using population register data for studying onward migration, given its inherent national character. In what follows, we explain how we make use of these population registers in dialogue with ethnographic data, to overcome what is often referred to as ‘methodological nationalism’ (e.g. Wimmer & Glick Schiller, 2002; Anderson, 2019).The two data sources are discussed separately in this section, before being brought together in the analysis section.

3.4.1 Quantitative Data

Our research sample is based on the total population of Somali children who were registered as asylum migrants in the Netherlands in the period 1995–1999, before reaching the age of 16 (N = 2457). These data are available through the System of Social-Statistical Datasets (SSD), a dataset that contains individual-level information on the population residing in the Netherlands (Bakker et al., 2014).

It is noteworthy that almost three in ten children (N = 735) had no parent (whether registered biological parent or legal guardian) residing in the Netherlands throughout the observation period.⁹ We restrict our quantitative analysis to children who had at least one registered parent or legal guardian. Furthermore, children who had not acquired Dutch citizenship over the observation period (N = 273 or 15.7 per cent) are left out of the analysis for two reasons. First, children without citizenship may have disappeared from the population register because of non-voluntary emigration, events that we do not aim to theoretically and empirically consider. Second, the interviewee sample exclusively consists of Somalis who hold a Dutch passport, hence omitting those without citizenship makes for a more adequate match between our data sources. This leaves a sample of 1467 Dutch Somali children and information on 2182 parents which we incorporate into the analysis. Children and their families are followed over a period of 15 years since their first registration in a Dutch municipality.

We chose to take children as the ‘reference point’ in the family because following the parent(s) would obscure the mobility of children who move with only one parent or independently of their parents, which happened in the majority of the cases (see Table 3.2, below). The observation that in only 52.3 per cent of the cases did all siblings (registered in the Netherlands) migrate together further validates an empirical approach that takes into account the differences in outcomes between siblings by following each of them separately.

Population register data provide a unique opportunity to link multiple generations. For the analyses presented in this chapter, children are linked to their parents as well as to siblings – through the parent(s). In so doing, the family as observed in the data excludes members who are not (registered) in the Netherlands. It is important to keep this in mind while interpreting the results, as we agree with other scholars that family relations – and transnational familyhood in particular – often incorporate several households within and across national borders (Borell et al., 2014; Johnsdotter, 2015). The number of family members who are identified should be regarded as in the lower range of the actual number of nuclear family members.

3.4.2 A Comparison Group: Protracted Refugees

To assess whether our findings are specific to people of Somali descent or are, more broadly, generalisable, we run separate models for a comparison group of refugees, sharing the transnational or ‘scattered’ feature of Somalis. The literature suggests that onward mobility is considered a viable alternative for return or long-term settlement in the country of asylum, in particular when migrants have pre-existing links to other destinations. Onward mobility may, in such cases, reflect a family (re-)unification or the search to join one’s own community. We approximate ‘scattered families’ with a refugee background by means of the UNHCR definition of protracted refugees¹⁰ and, as such, marked 14 countries or regions that produce(d) large numbers of refugees living in exile for an extended period of time (see Table A3.1 for countries and frequencies). Although this list in principle concerns forced migrants who applied for asylum in developing countries, a substantial share of these origin groups made it to ‘the West’. The selected sending countries are indeed also among the main origin countries of asylum migrants in various European countries.

3.4.3 Qualitative Data: Interviews with Somali Parents and Young Adolescents

In addition to the register data, we draw upon qualitative data, collected in a study on Dutch Somalis’ onward mobility to the United Kingdom. At the core of this project was fieldwork with Somalis in Leicester and London between 2008 and 2010 (see Table A3.2 for a sample description). London and Leicester were chosen because both cities house a substantial number of Dutch Somalis (Van den Reek & Hussein, 2003). Not only do these qualitative data provide in-depth insights into onward mobility decision-making but they also allow us to unveil factors, practices and outcomes that are situated across borders or are inherently transnational.

The 33 in-depth interviews were conducted with a mix of parents and children (14 female and 19 male Somalis).¹¹ The age at the moment of interviewing ranged from 20 to 62 years, that at migration to the Netherlands from 1 to 39 and at onward mobility from 12 to 48. All interviews were conducted in Dutch or English and subsequently transcribed.

3.5 Adopting a Multi-methods Approach

The two bodies of longitudinal data and research methods are brought together in a mixed-methods approach (Creswell, 2014; Hesse-Biber & Leavy, 2006). The type of questions we address here on the intergenerational dimensions around onward mobility and the role of children guided us in the development of our research design. Within mixed-methods research, this approach – where the type of information is central – is called ‘pragmatic relativism’ (Tashakkori et al., 1998).

In addition to the patterns which we structurally identify based on the registers, the stories collected through interviewing serve to tease out the motivations behind the family migration decision-making of Somali parents and the role of children within these decisions.

The literature on combining methods indicates that there are different ways in which data sources can be brought together in terms of timing. In our case the qualitative data were already collected before we started analysing the quantitative data. As such, the quantitative data collection was inspired by and analysed in conjunction with insights from the qualitative data. In the analysis we integrated the results from both methods as a form of triangulation, focusing mostly on the complementarity of the data.

3.5.1 Register Data Sample Characteristics

The statistics in Table 3.1 provide a general overview of the composition of the Somali children in our register sample. Both sexes make up close to half of the population, with a slight over-representation of boys (52.6 per cent). New-borns and one-year-olds are quite low in number, which is a direct result of the restriction of our sample to children who were born in Somalia. Both the migration trajectory as well as the path to municipal registration are lengthy processes, which results in children entering the registers at a later age. The great majority of our sample (ages 0–16), eight out of ten children, arrived before reaching secondary-school age.

Table 3.1 Register data description, Somali children and comparison group (protracted refugees)

Somali families in the Netherlands are relatively large, with almost half of the children entering the registers with three or more siblings. Another remarkable feature is the absence of their fathers in the Dutch context. About half (48.1 per cent)¹² of the children of Somali origin in our sample had lived separately from their father during their entire period of residence in the Netherlands, while this amounts to 12.3 per cent for other children (protracted refugees) who arrived in the same period. This situation of absent fathers fits the picture of ‘scattered families’ that was painted in many of the interviews. (Unofficial) divorces are common and some fathers have second wives but, even if the marriage of the parents is still intact and not shared, fathers are frequently physically (and emotionally) absent for Somali children (see also Haga, 2020; Ismael, 2020).

We follow the research population over an extended period of time, starting at 31 December of the year of entry onto the municipal registers up until the move or over a period of a maximum of 15 years (180 months). A crucial and unique asset of our data is the information concerning emigration. If residents decide to leave the Netherlands for a period of at least 8 months, they are expected to inform the municipality about the destination and migration date. As these records do not capture unreported emigrations, we deduce an additional category of emigration based on removals from the registers – that is, when the authorities have ascertained that a person or family is no longer residing at the recorded address. In our analysis we focus on emigration from the Netherlands, under the assumption that both registered onward migrants and children who have left without deregistering have resettled in a third country over the course of time.¹³ The figures in the bottom rows of Table 3.1 indicate not only that onward mobility is highly common among Somali children in an absolute sense (63.5 per cent) but also that, by the end of the observation period, this percentage is more than twice the share of onward migrants for the comparison group (28.7 per cent). Note that both Somali children and children in the comparison group have formally become Dutch citizens over time.

3.5.2 Analyses: Event History Models and Covariates

We first present non-parametric event-history models (Kaplan-Meier curves) to provide a general picture of the timing and magnitude of onward mobility among Somali children. These outcomes are complemented with a set of models to predict the probability of a single event, namely the onward mobility of the child versus their continued stay in the Netherlands. Cause-specific hazard models are performed to estimate the role of covariates in differentiated, competing events: migration with one or both parents (A) or independent migration, without parents (B).

By means of Cox proportional hazard models, we estimate how the probability of onward mobility is conditioned by a set of key characteristics. Sex is included as a dummy variable that takes value 1 for females; their male counterparts form the reference category. In line with the educational stages, as discussed in the case description section, age at arrival is included as a categorical variable with the reference category made up of children who arrived at pre-school age (0–5) – when school is not yet mandatory – and two additional categories of children who arrived at primary-school age (5–12) or aged 12–16 years. The family/parental situation in the Netherlands is also measured as a time-independent categorical variable, differentiating between children who had only their mother registered in the Netherlands (reference category), those who had only their father registered and those who had both parents registered in the Netherlands. It should be noted that this variable is fixed, meaning that it does not vary over the observation period.¹⁴ In addition, we incorporate a set of control variables to increase the accuracy of the estimation of the main effects.

We include the number of foreign-born siblings in the Netherlands (continuous), a dummy capturing whether the child is the oldest in the family (in the Netherlands), a time-varying dummy for mothers who gave birth to a child (a younger sibling) born on Dutch soil and a time-varying categorical variable measuring the employment situation of the parents. The number of years until Dutch citizenship acquisition (the speed of naturalisation) is also included in the extended models. Lastly, related to the refugee dispersal policy and residential locations in the Netherlands, we control for a set of time-varying residential variables, including the years between municipal registration and allocation to a regular dwelling, internal mobility (inter-municipality mobility up until time t) and residency in one out of five municipalities with the largest Somali community.¹⁵

3.6 Findings

3.6.1 Differentiated Onward Mobility: A Family Approach

To begin with, the interviews clearly show that onward mobility does not naturally imply a relocation of all (nuclear) family members simultaneously – on the contrary, that appears to be an exception (see Table A3.2, columns 7 and 8). Drawing upon the register data, the wide occurrence of differentiated family migration becomes visible (see Table 3.2). Of all Dutch Somali children who left for a third country during the observation period, about half (51.5 per cent) migrated with only one of their parents – mostly the mother – and more than a quarter (30.8 per cent) left the Netherlands without a parent or legal guardian. Patterns of ‘differentiated’ mobility are not exclusively found among Somali families. Although children in the comparison group more often than Somalis move with two parents instead of one, migrating without parents is the most common outcome among children in this subpopulation (48.3 per cent).

Table 3.2 Onward mobility (OM) composition

Our family approach reveals the variation in timing of onward mobility across members of the family. The qualitative data additionally reveal how Somali mothers often stayed behind with the youngest child in order for him or her to finish elementary school, while older children can travel on their own and/or will be taken care of by (extended) family members in the third country. Ekram (ID#31),¹⁶ a young Somali woman, for example, explained how her father had come first to the UK, while her mother decided to stay behind in the Netherlands with the youngest children: ‘My mother decided to stay with us. We were still at a point in our education where transferring could have had a tremendous impact.’

3.6.2 Age at Arrival and Timing of Onward Mobility

In this section, we discuss the empirical results concerning the role that the age of children at arrival in the first country of settlement plays in onward-mobility projects. We test the hypothesis (1a) that children who were young (particularly of pre-school age: 0–5) when arriving in the Netherlands are the least likely to migrate again, with or without their parents.

Before discussing the results from the multivariate analysis, we present a descriptive picture of the magnitude and timing of the onward mobility of children of Somali origin, by means of Kaplan-Meier cumulative hazard curves (Fig. 3.1). For each point in time, this curve gives the cumulative hazard rate, which is the event rate (onward mobility) at time t conditional on surviving (staying in the Netherlands) up to or beyond time t. The relatively flat curves in the first months indicate that onward mobility hardly ever occurs before 3 years after initial registration. After about 4.5 years (54 months), the hazard rates clearly increase, with peaks in the period between five and 7 years after registration, marking the 2 years following citizenship acquisition for most individuals in the sample.

Fig. 3.1

Kaplan Meier curve cumulative onward mobility rate Somali children (N = 1467). (Source: Statistics Netherlands, SSB)

Based on the cumulative onward-mobility rate at the end of the observation period (see Fig. 3.1), we conclude that the youngest children in the sample – aged up to 4 years old at arrival – are more likely to migrate again than those who arrived at mandatory school age (5–12). This gap in cumulative onward-mobility rates starts to manifest itself after 72 months (six years) since registration and remains significant over time. The line for the youngest children is steepest in the period of 4.5–6.5 years after registration, indicating that the odds of onward mobility among these children is highest relatively soon after entry in the Netherlands, when they are in primary education. However, more generally we observe that the bulk of children of Somali origin who left for the UK or another country did so after having spent an extended period in the Netherlands of which they became a citizen.

For children in elementary education, it is arguably relatively easy to switch schools and (even) school systems, compared to children who arrived at primary-school age and are transiting into secondary education in the first years after arrival. This partly supports our alternative hypothesis (1b). The difference in onward-mobility rates between the youngest group and children of primary-school age at arrival remains significant when controlling for a set of other characteristics and appears robust across a number of model specifications.¹⁷ The oldest age group (12–16) shows no significant differences to the middle group, which clearly demonstrates the non-linear nature of the relationship between age at arrival and onward mobility.

3.6.3 Motivations for and Timing of Onward Mobility

We turn to our qualitative data to gain insights into the motivations for the onward mobility of Somali families and the decision-making process. The data clearly demonstrate the dominant role of the parents in decision-making processes that often explicitly concern the educational paths of the child. Much emphasis is placed on foreign educational systems – the British system in particular – which are thought to offer better opportunities for (migrant) children than those in which these latter were raised.

In line with previous studies pertaining not only to Somali (Al-Sharmani, 2010) but also to Iranian (Kelly, 2013) and Moroccan and Bangladeshi (Della Puppa, 2018b) families who migrated to mainland Europe, the parents generally shared the belief that the British education system is more meritocratic and inclusive. They believe that this school system provides more opportunities for the next generation compared to those in other European countries. In addition to the idea that higher education is more accessible, the role of language (English) in education and the broader possibilities for religious education were enticing to parents, which partially stems from a desire to raise their children within their own communities. Related to the language and diasporic motives is the aim to prepare children for a future that is likely to be transnational. This is illustrated by a Somali mother (Fatumo, ID#13), who moved with her four children to Leicester while her husband remained in the Netherlands. When asked what her main reason for moving was, she replied:

Here there are more opportunities. Studying, for the children, is much better; here they can go to university and here you can give your children a religious upbringing. I did not have a clue about opening a shop back then; that came later.

The quote illustrates that mothers often consider the opportunities for their children before exploring their own likelihood of starting a new career elsewhere – like Fatumo, who ran a successful business in Leicester at the time of the interview.

The younger the children are, the more easily they adjust to new (learning) environments. For those who aim to facilitate their children’s chance of obtaining a degree in foreign higher education, it is thus advantageous to have them enrolled in the system before the transition from secondary to higher education. While, in the case of return migration, parents will often align this event with the completion of their children’s (formal) education, the pattern appears different when it comes to onward mobility. This is sensible when return is considered the end of a migration cycle, as opposed to onward mobility – which is merely a continuation of the migratory trajectory. In the presence of a large diaspora network in other countries and the absence of options for safe and sustainable return, families with younger children particularly will try their luck elsewhere. The fact that similar patterns are found for the comparison group (see below) lends further support to this argument.

Based on the extended event-history model coefficients (Fig. 3.2), we find that children who were registered either with both parents or only with their father are less likely to migrate again than those who only had their mother residing in the Netherlands (Fig. 3.2, left-hand panel; Table A3.4 Model 1b). This is remarkable, given that such associations are not found among children in the comparison group (Model 1b, right-hand column; Fig. 3.2, right-hand panel). Somali children not only generally travel together with their mother to arrive in the asylum country but are also accompanied by their mothers when engaging in onward mobility (Table 3.2). This finding resonates with the literature on the rearing practices of Somali parents in various contexts, including the Middle East and the United States, highlighting among other things the high priority given in particular to female members of the family. Female offspring are expected to take over the role of providing for the transnational family in the near future (Al-Sharmani, 2004, 2006, 2010). It is, moreover, argued that women, more often than their male counterparts, possess and rely on formal mobility capital (e.g. refugee status or a European passport) for mobility and residential purposes (Al-Sharmani, 2004).

Fig. 3.2

Hazard ratio plots onward mobility, Somali children (left panel, N = 1415) and comparison group (right panel, N = 9794), other control variables included (see Table A3.4, Model 1B)

In addition to theoretical expectations concerning the age of children and timing of migration, we hypothesised that those in large families are more prone to onward mobility than children in relatively small families. The hazard ratio (HR) for the number of siblings in the Netherlands is positive but insignificant. Being the oldest child in the family does increase the onward-mobility odds but only significantly for childen in the comparison group (HR 1.25). Children who have a sibling born in the Netherlands are more likely to migrate again than their counterparts whose mother did not give birth to a child on Dutch soil. The positive hazard ratio of this covariate for both groups (HR 1.49 and HR 1.42) suggests that new-borns in the family are positively related to onward mobility, not only for Somalis but for children with a protracted refugee background more generally. This constitutes additional support for our argument that parents will seek to migrate again while their children are still young, even if these latter largely or even exclusively grew up in the country of residence.

Just like family formation and expansion, employment opportunities generally form a core part of family livelihood strategies. Results concerning the socio-economic position of the parent(s) in the Netherlands indicate that, whereas parental employment reduces the hazard of migration for the comparison group (HR 0.71), the effect is insignificant for children of Somali descent. In other words, whether Somali parents have formal employment or not does not significantly relate to the mobility outcomes of their children. This resonates with the idea that children’s onward mobility is not primarily driven by the economic position of their parents but, rather, depends on considerations concerning various family members – not least their own position and perspectives. On the other hand, it should be noted that employment constraints are not fully (or adequately) captured in our models. Rather than whether or not the parent has paid employment, it is the quality of the job which may actually be driving onward mobility. Entrepreneurial and employment constraints were indeed widely expressed by the interviewees, regardless of their (formal) economic position. So even if parents had been successful in finding employment, the quality of the job was often insufficient, which led them to search for better opportunities elsewhere.

The coefficients of the control variables in the extended models provide some interesting insights that deserve a mention. Internal mobility is positively associated with international migration. Children who changed their residential location in the Netherlands are more likely to leave the country than those who remained in the municipality of initial settlement. This holds true for both Somali children (HR 1.26) and children in the comparison group (HR 1.18). Internal relocation thus presents itself as part of a wider mobility trajectory for children and families that comprises multiple international moves, rather than an alternative for international border-crossing. The lion’s share of the interviewees had, indeed, also relocated internally before moving to the UK, often motivated by employment, educational and housing opportunities and revealing clear similarities to motivations for international relocation. Residing in one of the five municipalities with the largest Somali community does not significantly decrease the onward-mobility propensity.

3.6.4 Mothers as Main Migration Decision-Makers

Our results highlight the transnational character of many Somali families, where the mother typically lives and moves together with one or more of her children. Most Somali children who obtained asylum in the Netherlands migrated only with their mother in the first place (see Table 3.1). This is consistent with previous insights into child-rearing and role divisions in Somali families in the West (e.g. Degni et al., 2006; Haga, 2020; Ismael, 2020). The responsibilities that are usually borne by the women – taking care of the household and family – are often further expanded after migration (Borell et al., 2014), resulting in greater informal power. This is reflected in the considerations of mothers about their own position and location, as well as the position(s) and perspectives of the children. When Nawal (ID#3) was asked who made the decision to migrate again, she responded:

• My mother, she was the main decision-maker, she was the one who said we needed to go, life was getting harder here in the Netherlands and we had lots of relatives in the UK. For my mother it was much easier here, most of our relatives are in the UK, America and Egypt. She was very lonely in the Netherlands. My father came to the Netherlands later – he was in Somalia for a long time – but still most of our relatives were not in the Netherlands.

• Interviewer: Did you agree?

• My brothers did not know much – they were young – and my mother consulted a lot with me. I am the oldest, I am the only girl. They knew – my mother informed us all – but I don’t know if they really understood.

Ibrahim (ID#19) also illustrates how important his mother was in making the final decision to move on to the UK.

In 2001, when I started my exams at high school, my brother wanted to continue his study in the US. He had just finished his higher education (HBO) but in the US they said ‘You are from the Netherlands and we also have a university in the UK, maybe that is a better option for you’. Then he went to have a look and did a bit of research on how it all works. My father thought it was a good idea; he had been a student himself so he knew how it all worked. My mother was the one who eventually decided. She said ‘If the schools are better – no, not better but quicker – if the time it takes you to get a diploma is shorter, and it is easier to get access, then why should we not take this opportunity?’

In the process of ‘rearing devoted and dependable children’ (Al-Sharmani, 2006, 62), Somali parents seek to support their children in navigating the educational field that may transcend international borders, making the mobility of these children essential. One could argue that this applies more generally to those mothers who are part of a large and dispersed diaspora network (see, for example, Jolivet, 2020 for the role of Moroccan mothers). The pattern of children residing and moving with their mothers is, however, not (convincingly) discerned among individuals in the comparison groups, which leads us to provisionally interpret this dynamic as being quite specific to Somalis.

3.6.5 Children Maturing: ‘Independent’ Versus Accompanied Mobility

In this section we delve deeper into the engagement of Somali offspring in onward mobility without their parents, which is observed for three out of ten children (30.8 per cent). The histogram in Fig. 3.3 shows the occurrence of onward migration by composition, for each month of the observation period. Whereas migration with one or both parent(s) peaks in the first 6 years after arrival, peaks for independent migration are observed after about 7 to 8 years and after a period of 15 years.

Fig. 3.3

Histogram time of onward mobility by migration composition (Somali children N = 905, comparison group N = 2438)

What catches the eye are the peaks in onward mobility among Somali children and young adults after roughly 6 years of registered residency in the Netherlands. This clearly corresponds to the required (legal) length of stay for naturalisation eligibility. Regardless of the composition, most onward mobility takes places relatively soon after Somalis have become Dutch citizens. For children in the comparison group, the peaks occur later in time and are somewhat flatter. Interestingly, whereas the occurrence of ‘independent’ onward mobility for Somalis decreases after this first peak of seven to 8 years, onward mobility among members of the comparison group continues to increase over time.

Concerning the age at migration we find, perhaps unsurprisingly, that children who arrive in the first country of settlement at an older age, have a higher hazard of onward migration without their parents than children who arrived before reaching compulsory school age (Fig. A3.2). In addition, engagement in internal mobility (between municipalities) increases the hazard of independent onward migration in particular.¹⁸ Children reaching adolescence move independently of their parents after having changed their residence in the Netherlands. The timing of their migration will depend on their own family formation process, rather than on the family planning of their parents. This is illustrated by Ayaan (ID#22), a young Somali woman who states:

• My brother, he left for the UK very quickly. He could not get access to university in the Netherlands so he left, already in 1991 he left for London, on his own, because he wanted to study. We all stayed in the Netherlands. I stayed in the Netherlands another 10 years and then I also went. When my brother left, I had just started high school. After that I continued my higher education in social juridical services and worked for the Refugee Council and the Migrant Rights Centre. My brother wanted to study here and I believe my mother thought that it would then be best to all move together.

• Interviewer: And one of your sisters stayed in the Netherlands?

• Yes, she wanted to move only after completion of her studies, but she got married, gave birth to a child and now is still there (in the Netherlands).

Our findings show that (the timing of) migration is the outcome of the careful consideration of the opportunities and attachments of multiple members of the family, where close involvement with the children, for mothers in particular, becomes apparent. Moreover, refugee children arriving at high-school age generally spent more time in the Netherlands, before migrating to a third country independently of their parents. For both generations – parents and their children who landed in the Dutch context for asylum purposes – family formation processes are often intertwined with onward mobility.

3.7 Concluding Discussion

This chapter has presented an examination of intergenerational dynamics within the onward-mobility arrangements of Somali families. Register and interview data are analysed in conjunction, offering unique insights into general patterns of family onward mobility as well as the underlying considerations of various family members. We find that, contrary to earlier findings concerning the role of children in migration planning, Somali children who arrived at a young (pre-school) age are more likely to engage in onward mobility with one or both parents and do so at a somewhat faster pace in comparison to children who arrived at primary-school age. This holds true more generally among protracted refugees who arrived in the Netherlands as minors. The qualitative insights reveal the multifactorial motivations of parent(s) to uproot their children once again. Parents generally feel that their children will be better off in the British education system, with enticing elements such as the accessibility of universities, English as the official language and the availability of religious education (see also Chap. 9 by Della Puppa and Sredanovic).

Our results further highlight the key role that mothers play in managing the onward mobility of various members of the family. Children mostly engage in onward mobility together with their mothers, who also ‘pioneered’ the asylum-motivated migration to the Netherlands. This contrasts with previously found patterns of stepwise family onward migration, where mothers and children typically followed the male head of the family (Herrera, 2012; Jolivet, 2020). The ‘reversed’ gendered patterns for both arrival in the Netherlands and onward mobility appear to be specific to Somalis (this pattern was not found for the comparison group). This is in line with in-depth qualitative insights into Somali transnational familyhood, revealing the high priority of mobility by female members of the family who are expected to cater for the migration of their offspring while trying to avoid disruptions in the children’s education.

Children who arrived aged 11 or older – and thus skipped the transition to secondary education – are the most likely to move independently of their parents. Such manifestations of onward mobility without the parents are common among Somali children who arrived as young adolescents. Among the comparison group, being a girl and being the oldest child in the household also increase the likelihood of this scenario occurring. The timing of independent migration differs from migration involving one or both parents, as it only unfolds after a relatively long period of time (with a peak at 15 years since first registration in the Netherlands). These findings merit further empirical investigation into the mobility trajectories of refugee children, the role of their parents and their own family formation and their independent paths towards new destinations.

The interview material provides some indication that the perceived educational perspectives overseas did not always materialise (cf. Patterson, Chap. 10). Anecdotal evidence points to the underperformance and/or drop-out of – mainly – Somali boys, who were consequently ‘sent back’ to Somalia or to their country of citizenship. The continuation or abandonment of educational trajectories is largely beyond the scope of our empirical contribution and deserves more attention in order to better grasp the open dimension and multi-directedness of the mobility trajectories of young Dutch Somalis and other young (protracted) refugees alike. As clearly demonstrated in this chapter, many refugee children will – rather than settling permanently in the country that granted them asylum – remain on the move, facilitated by the presence of family and community members in different parts of the globe. Future research into the settlement and return of refugees should therefore incorporate this long-neglected option of continued mobility, which may under certain conditions provide a viable approach to sustaining a livelihood.

Whereas this chapter represents a clear innovation in onward-mobility research by adopting a family approach, we do acknowledge its limits in accounting for the ties with other family members such as extended kin or community members and the role of social capital, in driving or facilitating the migration of Somali children, particularly in the case of ‘independent’ mobility. Multi-sited longitudinal or retrospective survey approaches are needed to further narrow this gap. Lastly, we acknowledge that no clear causal conclusions can be drawn based on our empirical contribution. Both the age at which children arrive in the Netherlands as well as their onward mobility may be associated with unobserved (family) characteristics. The present chapter provides various points of entry through which to further scrutinise refined theoretical expectations concerning family dynamics in onward mobility.

Appendix

3.1.1 Tables

Table A3.1 Frequencies protracted refugees by origin country

Table A3.2 Sample description, qualitative data

Table A3.3 Cox PH model onward migration, hazard ratio’s Model 1

Table A3.4 Cox PH model onward migration, hazard ratio’s Model 1B

3.1.2 Figures

Fig. A3.1

Kaplan Meier curve cumulative onward mobility rate comparison group (N = 9927). (Source: Statistics Netherlands, SSB)

Fig. A3.2

Hazard ratio plots onward mobility, Somali children (left panel, N = 1415) and comparison group (right panel, N = 9794), other control variables included (Table available upon request)