1International migration is still mainly analysed as a one-time, one-way movement from an origin country A to a permanent destination B. Yet migration trajectories are often more complex, as migrants may travel through and successively settle in several countries, or engage in circular mobility. However, the factors that shape individuals’ migration trajectories remain little known. In particular, even though qualitative studies suggest that multiple international moves have become a common mobility strategy (Paul 2011; Schapendonk, 2010) increasingly adopted in times of economic crisis (Cingolani and Ricucci, 2013; Sacchetto and Vianello, 2012), onward intra-European migration is still an under-researched area.
2In the African migration context, for instance, qualitative research points up the increasing complexity and fluidity of migration flows and routes towards and within Europe, with a subsequent fragmentation of migrants’ journeys (Castagnone, 2011; Schapendonk, 2010). Partly in response to border controls, step-by-step migration (Bredeloup and Pliez, 2005) is progressively developing as an emerging migration strategy, with transit migration playing an increasing role in migrants’ trajectories. Return and circular migration patterns are also common practices (Dia, 2009; Flahaux et al., 2011). Finally, African migrants reaching Europe appear to engage in further onward remigration within the European space (Nekby, 2006; Schapendonk, 2011), although the factors driving this phenomenon remain little known (Lindley and Van Hear, 2007).
3This article extends the literature by examining, in a quantitative framework, the drivers of onward mobility within Europe. In the context of this study, onward migration refers to migration from a European country to another in a two- or multi-step process. We focus on Senegalese migration between France, Italy and Spain, taking advantage of recently collected longitudinal data on migration between Africa and Europe (MAFE). It further contributes to the literature by adopting an innovative life-course approach to the study of factors driving remigration. Thanks to the retrospective nature of our data, we are able to examine how the dynamic processes of economic and legal incorporation in the country of settlement lead to onward mobility. Last, we take into account a dimension that has so far been neglected by research on stepwise mobility: the role of kin and friendship ties. While the role of ties to prior migrants has been extensively studied with respect to first international moves, we still know little about the extent and the ways in which they may influence stepwise mobility. In this analysis we examine how the location and composition of migrants’ networks affects their likelihood of moving onwards to another country.
4The article starts by reviewing the still limited theoretical and empirical approaches to onward migration before introducing the specific context of Senegalese international migration flows. A third section presents the data and the methodology employed, while the results are described and discussed in the last two sections.
I – Theoretical background and state of knowledge
Migration as a continuous, stepwise process
5Migration research has been predominantly guided by assumptions whereby migration is a one-off move from a departure country A to a destination country B, mainly directed towards Europe (revealing a strong Eurocentric bias), entailing a permanent settlement at destination (Agunias, 2006), and involving few or no subsequent steps after arrival in Europe. Ways of theorizing and studying migration have been paradoxically informed by a desire to “fix” migration processes within a clear spatial and temporal framework, in order to make it knowable (Cresswell and Hoskins, 2006). Methodological nationalism, as “an ideological orientation that approaches the study of social and of historical processes as if they were contained within the borders of individual nation-states” (Glick Schiller, 2009, p. 4), has largely influenced this way of conceiving migration, taking national borders as the natural unit of study (Wimmer and Glick Schiller, 2002).
6Furthermore, besides spatially fixing the phenomenon, studies of migration have mostly failed to really take into account its dynamic, ever-evolving character (Meeus, 2010). When studying migration and its drivers, a cross-sectional approach is still dominant, often disregarding previous trajectories and backgrounds of migrants and reducing them to dichotomous categories, such as permanent or temporary migrants (King et al., 2006). As both statistical and conceptual frameworks tend to privilege separate analyses of different segments of individuals’ mobility process, the full complexity of migration patterns is often neglected.
7In this respect, a promising tool is the life course approach, developed in the social sciences as a means of examining the evolution of individuals’ life trajectories over time and across social processes. This approach focuses on life events or transitions, their sequencing representing individuals’ life trajectories (Elder, 1975, 1985), also referred to as “life careers” or “paths” (Köu et al., 2009). Due to the interdependence of trajectories in different domains of an individual’s life, an event in one path can bring about status changes in other life domains (Dykstra and Van Wissen, 1999). The most important contribution of event history analysis to the study of migration has been to conceptualize it as an inherently dynamic phenomenon and resituate it within the broader life path of individuals (Courgeau and Lelièvre, 1997). Examining migrant biographies provides insight into how individuals construct their life course in terms of geographical, but also social, economic and labour mobility (King and Ruiz-Gelices, 2003).
8To our knowledge, this article is among the first to carry out a life-course analysis of onward mobility. This approach allows us to take into account migrants’ economic, legal and family trajectories over the course of their stay in their first destination and to connect these with their geographic mobility trajectories. The use of event history data thus introduces a diachronic perspective to the study of migration, enabling us to conceptualize it as a process and allowing individuals to shift from a status to another in a continuum of changes (Collyer and de Haas, 2012).
Onward intra-European mobility, an overlooked phenomenon
9Against a background of increasing complexity, fluidity and reversibility of migration flows, onward mobility has been studied mainly in relation to transit migration, conceived as a temporary stay in one or more intermediate countries with the aim of reaching a final destination. Most studies focus on migrants’ intermediate steps on the way to Europe and limit their analysis to European Union (EU) neighbour countries or North African countries (Brachet, 2009; van Moppes, 2006; de Haas, 2006; Nyberg Sørensen, 2006), failing to explore onward movements once migrants have arrived in Europe.
10The limited research on intra-European migration has mostly focused on the mobility of Eastern Europeans, who, following the 2004 EU enlargement, moved to countries that chose not to restrict access to their labour markets. Intra-EU mobility has also been an object of attention in relation to secondary movements of asylum-seekers who, once in Europe, pass through one or more “third countries” with the aim of reaching destinations where network members are settled or where reception conditions, opportunities and welfare provisions are more generous (Koser, 1997; Weine et al., 2011).
11Recent studies suggest that secondary movements within and from Europe are an increasingly common mobility practice (Nekby, 2006; Takenaka, 2007). However, there is a surprising absence of systematic attention to the intra-European mobility of third-country nationals (Benton and Petrovic, 2013; Pascouau, 2013), due partly to the fact that available data are scarce and limited to localized qualitative studies. Information collected by EU countries includes administrative statistics, data drawn from the national population census or population registries, and survey data, which usually do not distinguish between non-EU citizens arriving at destination as their first step in Europe or via another first member state (EMN, 2013). Undocumented mobility within Europe is an even more hidden and unknown phenomenon, despite its role in migrants’ mobility strategies.
12Intra-EU mobility of third country nationals is regulated by EU migration directives, providing rights of entry and stay within the EU to certain categories of migrants such as students, long-term residents or highly-skilled workers. Besides excluding the larger groups of low- and medium-skilled workers, the EU migration directives leave significant areas of discretion to member states – and therefore to national laws – in regulating mobility. As a consequence, freedom of movement and settlement within the EU is governed by a fragmented legal landscape and is obstructed by considerable barriers. Reports by the European Commission show that few people have in fact been admitted under the schemes provided for under the directives, and consequently few have taken up the opportunity to move within the EU (Pascouau, 2013).
13Given its growing importance within the EU acquis, with policy proposals to strengthen intra-EU mobility for some groups, and given that mobility of the labour force is a possible solution to employment imbalances within Europe, intra-EU mobility of third-country nationals is a phenomenon that needs to be explored and understood in its different forms, determinants and implications. Using new and original survey data, this article reconstructs Senegalese migrants’ international mobility trajectories and examines the prevalence and drivers of onward mobility within Europe.
Drivers of intra-European onward mobility: the role of human capital and socioeconomic integration
14Most work challenging the paradigm of migration as a permanent, one-off movement has focused on return migration (Cassarino, 2004; Constant and Massey, 2003; Dustmann, 1996, 2003; Flahaux, 2013). Few quantitative studies distinguish between migration to a third country and movements back to the origin country (among the few exceptions see Larramona, 2013; Nekby, 2006; Rezaei and Goli, 2011; Schroll, 2009). Yet we may expect the reasons and circumstances behind onward mobility to differ from those underlying a return move (Kelly, 2012). Thus, it is important to distinguish the two phenomena, and that is what this article sets out to do. Given that the factors driving return migration are better known, we focus mainly on drivers of onward mobility.
15The so far limited research on onward migration has mostly focused on the impact of human capital and occupational status in this mobility, reaching somewhat contrasting findings. A strand of quantitative work analysing remigration from Nordic countries, such as Denmark (Rezaei and Goli, 2011; Schroll, 2009) or Sweden (Nekby, 2006), or from the United States (Takenaka, 2007), finds that those who engage in multiple migration are positively selected on education and income. Their findings show that highly-skilled immigrants and those with graduate education have a higher probability of leaving for third-country destinations than returning to countries of origin or staying at destination. According to these studies, onward mobility is facilitated by these migrants’ high skill levels and allows them to make better use of their human capital (Kelly, 2012).
16A different reading of onward mobility is apparent in a set of qualitative studies, mostly focused on remigration from Southern Europe. These studies emphasize the precariousness of migration careers, both in terms of deteriorating conditions in the European labour markets and of a concomitant tightening of rules relating to migrants’ legal status (Van Nieuwenhuyze, 2009). In this perspective, the fragmentation of the migration paths in Europe may reflect the impact of macro-structural changes in migration policies and labour markets on individual socioeconomic patterns of integration at destination (Larramona, 2013). It is argued that onward migration is the result of unsuccessful socioeconomic integration in the first European destinations and is pursued by the more vulnerable migrants. Furthermore, recent works suggest that the economic crisis has increased levels of secondary migration among long-time residents in Europe, such as Moroccans in Italy (Benton and Petrovic, 2013; Cingolani and Ricucci, 2013; Sacchetto and Vianello, 2012).
17Our article contributes to this debate by taking into account the dynamic nature of migrants’ human capital, and their economic and legal status in Europe, thus offering a more fine-grained analysis of their influence on onward mobility.
Migrant networks, an influential factor in onward migration?
18A factor overlooked in most of the above-mentioned studies is the influence of migrant networks in relocation decisions. Migrant network theory argues that connections to migrants abroad encourage people to move by diminishing the risks and costs while increasing the benefits of moving (Boyd, 1989). Empirical work has generally found that migrant networks are crucial in triggering a first international move but less important in subsequent trips (Massey, 1987; Massey and Espinosa, 1997). Such work argues that as migrants accumulate personal migration experience, they no longer need to rely on migrant social capital.
19Yet most of this (quantitative) research is based on remigrations in the same country, while the role of networks in onward mobility to a third-country has received little attention. Recent qualitative studies suggest that networks might play a key role in shaping mobility trajectories and remigration within Europe. Lindley and Van Hear (2007) find that the presence of relatives and friends in the UK represents a strong incentive in the decisions of Somali migrants to relocate there from mainland Europe. Schapendonk (2012) also shows that migrant connections have both a facilitating and an aspiration-shaping role. Connections assist migrants in their journeys by helping them avoid exploitation or abuse, but they also give rise to new destination aspirations, through the sharing of information (Bang Nielsen, 2004). Schapendonk’s findings emphasize the key role of weaker ties, encounters made en route, or acquaintances in shaping African migrants’ trajectories. Kelly (2012) similarly argues that diasporic connections enabled her Iranian respondents to pursue opportunities across space that were more difficult to reach for those lacking international connections. In this perspective, networks are seen as a resource that enhances mobility opportunities for migrants, allowing a better redefinition and readjustment of the migratory project once in Europe.
20In contrast, the “affinity” hypothesis, as formulated by Ritchey (1976), whereby a dense local network of family and friends acts as a brake on further migration, has received less consideration. This is partly because most qualitative studies only interview onward migrants, thus excluding those who remain in their initial destination. Using data on re-migrants, returnees as well as non-migrants from Sweden, Schroll (2009) finds that living in an area with a high share of immigrants from one’s home country decreases the probability of out-migration from Sweden, especially for migrants from more distant countries.
21This article extends the literature by systematically examining the role of migrant networks in shaping (subsequent) mobility trajectories. In doing so, it disaggregates networks according to their location, the type of relationship between members and the migrant, the gender of the members and their migration experience.
II – Senegalese migration flows to Europe
22Internal and intra-continental migration flows go a long way back in Senegalese history and have involved large shares of the population (Adepoju, 2004; Bakewell, 2009; Bakewell and de Haas, 2007; Beauchemin and Lessault, 2009; Manchuelle, 1997; Ndiaye and Robin, 2010; Trémolières, 2009). Senegalese migration to (and within) Europe has a more recent history. The flows have their origins in the colonial relationship with France, when some Senegalese held temporary blue-collar positions in the French administration or enrolled in the French army as Tirailleurs during the Second World War (Manchuelle, 1997). At the end of the war, some of them settled in France for good, engaging mostly in commercial activities.
23In the immediate post-war reconstruction and subsequent economic growth in Europe, the French public authorities implemented an active recruitment policy to attract foreign workers – mainly men from colonies in Africa, including Senegal. As a result, flows towards France intensified after Independence (1960) to meet the needs of the rapidly developing automobile industry (Pison et al., 1997; Robin, 1997; Robin et al., 2000). Later, in response to the economic crisis of the early 1970s, France followed the example of other European countries and in 1974 stopped all recruitment programmes for foreign workers. This did not lead to a decrease in immigration, however. In terms of numbers, family reunification has since become the most important channel for immigration to France. Moreover, following the halt to foreign labour recruitment programmes in 1974, external and internal controls (visas and residence permits, respectively) were introduced (Devitt, 2012). In 1985 France introduced a compulsory visa for Senegalese nationals.
24Meanwhile, Senegal was facing one of the most serious periods of drought of its contemporary history, with a subsequent crisis of the traditional agricultural system. Propelled by the globalization of the economy, by ineffective national development policies and by accelerated pauperization, more families invested in international migration (Adepoju, 2004). From the 1980s onwards, Senegalese migration flows to Europe intensified considerably, while migration to other African countries decreased (Flahaux et al., 2013). At the same time, destinations in Europe became more diverse, with a switch from France to southern European countries. Italy became the most important destination for Senegalese migrants in the 1990s, after laws legalizing irregular migrants were passed in 1990 and 1994. Here, the new immigrants were able to find work in the informal trade and in the industrial sector. Initially the Senegalese arrived in Italy primarily through secondary migration from France, but they subsequently established direct channels and networks of migration from Senegal. From the end of the 1990s, Spain also became a popular destination, with its booming construction and agricultural sectors attracting Senegalese workers. Figure 1 documents the evolution of these trends based on stocks of (legal) Senegalese migrants in France, Italy and Spain.
Evolution of Senegalese migrant stocks in France, Italy and Spain
Evolution of Senegalese migrant stocks in France, Italy and Spain
25Contrary to the most recent trends in France, entry and residence policies in Italy and Spain have focused on economic migration channels. These countries are more labour-intensive than those of north-western Europe and their economies rely more heavily on immigrant unskilled or low-skilled labour (Arango, 2012). The underground economy has also played a crucial role in shaping migration patterns in these two countries (with a strong attraction effect), and represents one of the main structural differences with respect to France (Reyneri and Fullin, 2010).
26As flows to Europe intensified and destinations become more diverse, migrants’ socio-demographic profiles also changed. Up to the 1980s, most international migrants came from the rural areas of the Senegal River Valley. The later period saw a diversification of departure regions, with cities in general, and the capital in particular, playing an increasingly key role. The first migrants were Toucouleur and Soninke from the Senegal River Valley, with little or no education, primarily employed in French manufacturing and construction industries. Most of them were men who made use of existing social cohesive networks abroad. From the 1970s, educated migrants started growing in number, some of them with the objective of completing their studies abroad, especially in France. In addition, women began to join their husbands and to establish new families abroad, mainly in France, and to a lesser extent in Italy and Spain, where this type of migration is still under-represented. A smaller share of women engaged in international migration on their own, for economic or educational reasons, but their numbers are still limited (see article by Toma and Vause in this issue).
III – Data and Methods
Data
27Onward migration is still an under-researched topic due largely to the lack of adequate data. The data requirements for examining this phenomenon are indeed quite high: longitudinal data are needed to trace individuals’ geographic mobility over time, at least in a retrospective design; at the same time, surveys need to be carried out in several destination countries (or locations), as well as in the origin country (or communities), in order to capture all possible migration outcomes (those who stay at their first destination, those who remigrate and those who return).
28This article uses a new set of survey data collected between 2008 and 2009 as part of the Migration between Africa and Europe (MAFE) project. This MAFE dataset is both longitudinal and multi-sited, making it one of the few quantitative sources that can be used to explore the factors which shape intra-European migration among sub-Saharan African migrants.
Longitudinal data
29The survey collected detailed retrospective life histories on several domains of the respondents’ lives, such as their education and employment trajectories, their family formation and housing histories, as well as their geographic mobility and the migration trajectories of their social network members. The information was collected on a yearly basis from the respondent’s birth until the time of the survey. This enabled us to identify and date all migration events and to link them to their labour market and legal status transitions, as well as to the migration spells of their kin and friends with international mobility experience, dimensions which are of main interest in this study. While actual dates may not be completely accurate due to retrospective bias and the long period of life recorded, respondents are likely to remember the sequencing of events (i.e. whether their brother migrated before or after their own migration), which is particularly important for the type of analysis we use.
Multi-sited data
30The MAFE survey is based on a transnational research design, as information was collected from non-migrants and return migrants in origin countries, as well as from current migrants in several destinations. In this article we use only the MAFE-Senegal survey, [1] which collected data in the Dakar region of Senegal and in three European destinations. In a first stage, a household-level survey was carried out with approximately 1,200 households, a sample representative of the Dakar region. Subsequently, life histories were collected using a life event history questionnaire from 1,067 individuals (non-migrants and return migrants), who were sampled within the households using a three-stage probabilistic sampling method. In addition, 600 current migrants were interviewed in the three main Senegalese destinations in Europe [2] (200 migrants each in France, Italy and Spain), recruited through a mix of non-probabilistic sampling methods [3] (Beauchemin, 2012), in order to include both documented and undocumented migrants.
31Our study population consists of all respondents having migrated at least once to France, Italy or Spain for a period of at least 12 months and who may or may not have been at destination at the time of the survey. The units of analysis are migration spells and our sample thus consists of 775 migration spells belonging to 668 individuals. By only focusing on three European countries, this analysis cannot claim to provide a full picture of the intra-European migration patterns of Senegalese migrants. However, France, Italy and Spain were deliberately chosen as they are the main destinations, together attracting 42% of all Senegalese international migrants in 2002 according to the Senegalese census. Thus, despite its limitations, the MAFE survey is currently the most comprehensive quantitative source for analysing intra-European remigration patterns among sub-Saharan migrants.
Methods
32Both descriptive and multivariate methods are used in the analysis. As a first step, sequence analysis is used to visualize migration trajectories that involve at least one onward intra-European move, from the first migration until the survey date. The chronological sequencing of migration events by geographical location (countries where they occurred), and nature (outmigration, onward migration, returns, re-departures, etc.) shapes the mobility trajectories of each respondent.
33Our analysis then turns to the determinants of the decision to leave the country of destination, either in order to move elsewhere in Europe or to return to the origin country. Given that the data are longitudinal, the best way to do this is to employ discrete-time event history analysis. This technique makes it possible to measure the “risk” that an event will occur (i.e. remigration) and to follow the evolution of this risk over time, while taking into account the variables that may interact with it. In other words, the method estimates not only whether the event occurs but also when it occurs (Le Goff, 2013).
34As discussed by Allison (1982) and Yamaguchi (1991), this method divides the time into discrete intervals (in our case, calendar years) and estimates the probability of observing the event within each interval, given that it has not occurred up to that point. It is more suited to data where the information is collected in larger time units (such as years), which is why it was preferred over continuous-time duration models, such as the Cox model. Migration events are only recorded once a year so there may be many observations with the same spell length. This may bias coefficients and standard errors in a Cox model but is not a problem in discrete-time duration models.
35Using the respondents’ detailed migration histories, a categorical measure is constructed, indicating whether the individual is still in the country of destination, whether he or she has remigrated to France, Italy or Spain [4] or has returned to Senegal. We distinguish the case of return since we expect it to be associated with different factors from those behind remaining at destination or remigrating. Our focus is on remigration, however, and we will mostly discuss its drivers. Out of the 775 migration spells in France, Italy or Spain that form our analysis sample, 608 were still ongoing at the time of the survey, 76 ended with a remigration and 91 with a return to Senegal. The number of remigration and return events is therefore low and results need to be interpreted with care.
36The individuals enter the risk set at the beginning of their European migration spell and are followed until either the time of the survey (if still at destination), until their departure for another destination (France, Italy, Spain) or until their return to Senegal. Since we distinguish between these two different types of events (return and remigration) we analyse the data in a competing-risk framework using a multinomial logistic regression model. This model assumes that for an individual i in the population, the log odds of experiencing an event of type r rather than an event of type s (the reference category – here, non migration) at discrete time point t are given by:
38where πr is the hazard of an event of type r occurring at time t for an individual with covariates Xri that are constant over time (e.g.: gender) and a matrix of time-varying covariates Zi(t); the βr are the respective vectors of coefficients; ε is the residual.
Covariates
39The great advantage of this method is that, unlike cross-sectional regression analysis, it can be used to examine the influence of characteristics that vary with time. Indeed, most of the factors that we expect to drive remigration chances are dynamic and are accordingly captured here using time-varying variables. The respondents’ level of education is introduced as a categorical variable distinguishing those with no schooling, primary education, secondary education, and tertiary education. Another categorical variable measures occupational status, and distinguishes persons not in employment, semi-skilled or skilled wage-earners, unskilled wage-earners, self-employed, and students. A large proportion of Senegalese migrants engage in small trade activities on a self-employed basis, as also discussed in Section 3. These activities are highly precarious and those who undertake them run the risk of being deported while also compromising their chances of being regularized. It is thus important to distinguish this status from other forms of unskilled work undertaken with some form of contract. Legal status is also a categorical, time-varying variable with three categories: visa or no formal documents, residence permit or permit not required, [5] work permit only.
Access to migrant networks in Europe
40One of the innovative features of the MAFE survey is the longitudinal information it collects on the respondents’ migrant networks. Interviewees are asked whether any of their parents, siblings, children, partners or other kin or friends have migration experience (either past experience or are still currently abroad). The relationship to ego, the gender, the year of acquaintance (if spouse or friend) are also recorded for each member. Based on this information, four variables are constructed, capturing: 1) ties in the country of settlement [6] (networks at destination); 2) ties in other European countries excluding the country of settlement (networks elsewhere in Europe); 3) ties outside Europe (mostly Africa); and 4) returnees (network members who returned to Senegal after at least one year abroad). These variables take the value 1 if the respondent has at least one such tie. Furthermore, three aspects of the composition of networks located in other European countries are taken into account: the type of relationships (close family ties versus extended kin and friends), the gender of the tie and the level of migration experience (recent, experienced and long-term migrants). [7]
41Family status and the location of the partner and children are also taken into account with two separate variables. The partnership status distinguishes between those who are single, those whose partner(s) is/are in the same country, and those whose partner(s) is/are located elsewhere. A very similar variable is constructed with respect to children.
42The models also control for several time-varying contextual and individual characteristics which have been shown to shape mobility, such as the period, the country of settlement and the sex of the migrant. [8] The time spent at destination is the basic duration variable, and we expect a negative relationship with the probability of remigration. The respondent’s age is also included in a continuous manner. Whether the current country of residence was considered the final destination or whether the migrant was unsure about his or her final destination are also included in the models as a categorical variable.
43Appendix Tables A.1 and A.2 present descriptive statistics for all of the explanatory variables used in this analysis. Since the great majority of them are time-varying, they are measured at the last year of the migration spell, distinguishing between our three outcomes (non-migration, remigration and return). Statistical tests (chi2, t-tests and Bartlet’s tests) examine whether the differences between these three categories are significant or not and are reported in the last column.
IV – Results
44Three main types of mobility trajectories are adopted by those who spent at least one year in a European country. [9] First, unique migrations, involving only one migration step in Europe, which were still ongoing at the time of the survey, represent more than three-quarters of the cases (79%). Second, 10% of the Senegalese who migrate to Europe follow trajectories involving at least one remigration within Europe. Finally, 11% of the migrants to Europe eventually return to Senegal at least once. While these more complex trajectories represent only about a fifth of the cases, it should be borne in mind that we are dealing with right-censored observations: those still in their first European destination at the time of the survey may remigrate to another European country or return to Senegal at a later date. This is all the more to be expected given that the survey was carried out before the economic crisis (2008-2009). If, as previous qualitative work has shown, the crisis has increased the likelihood of secondary moves within Europe, the share of trajectories involving at least one onward migration within Europe is probably under-estimated with respect to the situation prior to the survey.
45Next, we use sequence analysis to focus only on the trajectories involving at least one remigration within Europe in order to analyse its geographic patterns. Figure 2 shows in which of the survey countries migrants are more likely to initially settle, and where they move on from there. The largest share of trajectories (40%) involves a first move to France, followed by a subsequent migration to Italy or Spain. This is probably due to the earlier commencement of Senegalese flows to France, and thus confirms prior findings. However, the opposite trend is also visible as another quarter of migrations involve moves from Italy or Spain to France. Around 10% of trajectories commence with a migration in another African country before reaching Italy and moving further within Europe. Most of those who remigrate within Europe only do so once (72%) but 20% remigrate twice, and the rest three or more times.
Direct intra-European stepwise trajectories (N = 68)
Direct intra-European stepwise trajectories (N = 68)
46Figure 2 gives an idea of the sequencing of events, but gives no indication of their duration. Figure 3 shows the cumulative distribution of remigration spells by time spent by migrants at their previous European destination. Remigrations appear to occur relatively early in the migration trajectory: half occur within 2 years, and 80% within 5 years.
Cumulative distribution of migration spells by time spent at destination before remigration within Europe (France, Italy or Spain; N = 76)
Cumulative distribution of migration spells by time spent at destination before remigration within Europe (France, Italy or Spain; N = 76)
47Previous qualitative studies have argued that multiple migrations are part of a stepwise mobility trajectory where migrants strive to achieve a desired and hard-to-reach final destination (Conway, 1980; Paul, 2011). The MAFE questionnaire allows us to examine this aspect through a survey question that asks respondents whether, at the time of their migration, they considered the country they had reached to be their final destination, whether they had no clear idea about where they wanted to go, or whether they intended to go elsewhere. Figure 4 shows that very few respondents chose the latter case. A quarter of those who eventually moved to a different country were initially unsure about where they wanted to go; their share is higher in this category than among those who stayed at their initial destination or those who eventually returned to Senegal. Yet in a large majority of cases, migrants considered their initial destination to also be the final one. Thus, unlike the Filipino migration studied by Paul (2011), we cannot really talk about a planned stepwise mobility trajectory with respect to Senegalese intra-European mobility, with migrants making their way up through their pre-defined hierarchy of locations.
Intended destination of the migration spell and actual outcome (remigration, return or neither)
Intended destination of the migration spell and actual outcome (remigration, return or neither)
Drivers of intra-European onward mobility
48We pursue our analysis by estimating a series of multinomial logistic models, taking the case of no onward or return migration as our reference category. Coefficients are presented as relative risks and should be interpreted in relation to the reference category, which is remaining in the country of destination. Sampling weights are used in all model and standard errors are clustered by individuals.
49Table 1 presents the results from the first model. Confirming descriptive findings, as time spent at destination increases, the relative risks of remigration and return compared to remaining at destination decrease when other variables in the model are held constant. Furthermore, we see that both remigrations and returns are relatively more likely to occur from France than from Italy or Spain, probably reflecting the earlier onset of Senegalese flows to France. The relative risks of remigration compared to non-migration are significantly higher for men than for women, a fact that has also been documented in other studies (Nekby, 2006; Schroll, 2009). While initially this is also the case with respect to return, gender is no longer an influential factor after accounting for socioeconomic differences between respondents. Confirming the descriptive findings, respondents who were unsure whether the country of settlement was their final destination had substantially higher relative risks of remigration. Interestingly, this is not the case with respect to return migration: the undecided were less likely to return than to stay at destination.
Likelihood of intra-European remigration and return (multinomial logistic regression, relative risks; Ref.: remain in country of settlement)
Likelihood of intra-European remigration and return (multinomial logistic regression, relative risks; Ref.: remain in country of settlement)
Note: The 8,136 person-years are taken from the 775 migration spells, of which 608 ended with no remigration, 76 with a remigration and 91 with return migration. All network variables exclude the partner and children.Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01
50Retrospective data are quite limited when assessing historical trends, but findings seem to confirm previous research on the evolution of Senegalese migration flows. The relative risk of remigration appears to have increased in the 1990s and to have peaked in the first half of the 2000s, while decreasing afterwards. Return migration from Europe illustrates a completely opposite trend: relative risk ratios of return compared to remaining at destination significantly decreased in recent periods, as also found in other work (Flahaux et al., 2013). This may reflect the paradoxical consequence of the restrictive migration policies adopted by European governments. By making it more difficult to cross borders, these policies may encourage permanent settlement and thus increase overall immigrant stocks (de Haas and Czaika, 2013).
Remigration: a way to achieve social mobility?
51Previous research reached contrasting findings with respect to the educational and economic profile of onward migrants. Interestingly, we do not find education to significantly influence relative risks of remigration; it appears that individuals of all levels of education engage in onward mobility to a similar extent. In contrast, occupational status has a substantial impact. Migrants having obtained a skilled or semi-skilled employment at destination have significantly lower relative risks of remigrating than those with a more insecure status on the labour market at first destination, such as the self-employed but also the unemployed or unskilled wage workers (Table 1).
52Those currently enrolled in education are also less likely to remigrate, which suggests that having invested in the accumulation of human capital in a particular country discourages from migrating elsewhere in Europe. Previous work (Castagnone et al., 2013; Toma, 2012) showed that having studied at destination increases the chances of attaining skilled employment. Thus, Senegalese students probably increase their chances of achieving social mobility by staying at destination. Occupational status has similar effects on the probability of return, suggesting that both phenomena are discouraged by the attainment of a higher position at destination.
53Furthermore, it is not only migrants’ occupational outcomes at destination that shape their subsequent mobility trajectories, but also their integration in terms of legal status. A weaker legal attachment to the country of settlement increases the relative risk of moving on: those who have no documents or only a visa are more likely to remigrate or to return to Senegal than to stay at destination, compared with those with a residence permit or who do not need one.
European networks, highly influential in onward migration
54Migrant and family networks are an important factor in intra-European remigration, both constraining and encouraging further moves. First, having one’s partner at destination has a strong and significant negative impact on remigration rates compared to being single or having a partner located elsewhere. Coefficients are similar with respect to children, but non-significant: having at least one child in Senegal or elsewhere increases the likelihood of return compared with having children at destination. Furthermore, the discouraging effect of local ties on remigration also extends to other ties beyond the nuclear family, though their negative effect is not consistently significant across models.
55On the other hand, having ties elsewhere in Europe increases substantially and significantly the relative risks of subsequent moves on the continent. No such effect can be observed on return probabilities. We also examined the influence of ties located elsewhere (mostly in Africa) or of former international migrants who returned to Senegal, yet these ties do not appear to matter in the decision to leave the destination country. Only returnees seem to have a positive impact on the likelihood of return, but the coefficient is not significant.
56Having a network in other countries in Europe appears to increase the likelihood of subsequent mobility within the continent, a result that is robust to controlling for other factors. Table 2 further investigates whether the role of these networks depends on their composition. Models 2 to 4 include different specifications of the network variables, while controlling for the same set of factors as Model 1.
Network effects on the likelihood of intra-European remigration or return (Ref.: remain in country of settlement)
Network effects on the likelihood of intra-European remigration or return (Ref.: remain in country of settlement)
Note: All network variables exclude the partner and children.Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01
57We find that only the weaker ties – friends or extended kin – significantly increase the likelihood of remigrating (Model 2), while close kin – siblings or parents – have less of an effect. The gender of the tie also appears to matter, as only male connections in Europe encourage remigration (Model 3). Lastly, and somewhat surprisingly, it is only network members who have recently migrated – within the last 3 years – who affect chances of intra-European mobility (Model 4). None of these ties has any significant effect on the probability of return.
Conclusion
58Onward mobility is a relatively under-researched phenomenon which challenges the idea that migration is a one-off event, leading to permanent settlement in the country of destination. So far, studies have tended to focus on return migration, while the drivers of remigration to a third country remain little known. Partly responsible for this is the nature of most migration data; information is recorded either at origin or at (one) destination, thus preventing the study of more complex mobility trajectories that involve several destination countries.
59Taking advantage of a recent multi-sited survey on migration between the Dakar region of Senegal and France, Italy and Spain, this article examines the drivers of onward mobility within Europe. The biographic nature of the data enables us to adopt a life-course perspective and to analyse, in a discrete-time event history framework, the ways in which processes of economic, legal and social integration at destination shape subsequent mobility trajectories. Furthermore, we are able to systematically investigate a factor of influence that has so far received little attention: the role of family and friendship ties in onward mobility. While the multi-sited and longitudinal nature of our data make it extremely rich for studying drivers of onward migration, our analysis is based on relatively small samples of migrants from the Dakar region, so our findings should be interpreted with caution.
60Findings show that, unlike remigration from Nordic countries such as Denmark (Rezaei and Goli, 2011; Schroll, 2009) or Sweden (Nekby, 2006), or from the US (Takenaka, 2007), Senegalese onward migrants are not positively selected on skills. Those who are more likely to move within Europe do not belong to the categories that have the formal right to do so, such as students, highly-skilled workers, or long-term residents (Pascouau, 2013). Onward mobility occurs, in fact, early in the migration trajectory and especially concerns the low-skilled and the unemployed, as well as those lacking more permanent documents (such as a residence permit). These findings suggest that the development of legal measures to facilitate intra-EU mobility of migrant workers already residing in the member states should also target the low-skilled and recent migrants who, while representing the largest component of the demand for labour force in Europe, are also the most fragile component of the migrant population and the one most affected by the economic downturn.
61This also explains why onward migration within Europe is not, in the Senegalese case, a planned stepwise mobility where migrants seek to attain a final, desired destination, as was shown by Paul (2011) with respect to Filipino migration to Hong Kong and Singapore. Instead, most onward migrants consider their initial European country of settlement to be their final destination or are unsure about their plans. Thus, onward mobility appears to be the product of a constant re-evaluation of opportunities rather than a carefully planned trajectory.
62Whereas employment insecurity and lack of stable legal status appear to be important incentives for moving on within Europe, our findings also suggest that having ties in other European countries is an important resource in triggering this form of mobility, confirming previous qualitative findings. Not all ties have the same influence, however: weaker links to male migrants who have recently migrated in Europe appear to be the most influential. The fact that men are more likely to engage in subsequent mobility may explain the gender finding, since previous work has shown that connections to migrants of the same gender are more influential in the migration process (Curran and Rivero-Fuentes, 2003; Garip, 2008; Toma and Vause, 2014). The fact that weaker ties to extended kin or friends play a greater role than closer family bonds evokes Schapendonk’s (2012) findings, emphasizing the importance of “bridging social capital” in shaping African migrants’ mobility routes. Furthermore, the fact that recent migrants (who moved to Europe within the past three years) represent more influential connections, may reflect the shifting nature of destination aspirations and the spontaneous quality of mobility trajectories. Information provided by a friend who recently migrated to a different European country may shape new aspirations and encourage migrants to try out their luck.
63In contrast, the presence of nuclear family members and other networks in the country of settlement has a retaining effect on Senegalese migrants. Having one’s spouse(s) at destination strongly discourages both onward and return migration. Again, family formation and mobility decisions are highly interdependent, and the decision (mostly concerning men) to bring one’s partner from Senegal, or to join one’s partner in Europe (almost exclusively the case of women, in the Senegalese context) is arguably concomitant with the decision to remain at destination. Yet, the case of families or couples engaging in onward mobility is not so rare, as shown by Kelly’s (2012) research on Iranians moving from Sweden to the UK. Furthermore, other ties at destination beyond the nuclear family have a similar, though less strong, discouraging effect on moving to another European country. This may suggest that maintaining co-ethnic networks leads to a positive attachment to the destination country and to more successful integration outcomes. More research is needed on the mechanisms accounting for the role of local networks in discouraging onward mobility.
64In sum, this article contributes to the literature by examining the drivers of third country nationals’ onward mobility within Europe. Focusing on the case of Senegalese mobility between France, Italy and Spain, our findings reveal the interdependence of socioeconomic, legal and mobility trajectories. Onward migration appears to be a strategy to achieve socioeconomic integration and legal integration for those who fail to attain these at their first destination, while connections to migrants in other European countries are a key resource facilitating this strategy.
Appendix
Descriptive statistics of control variables by migration status at the last year of the migration spell
Descriptive statistics of control variables by migration status at the last year of the migration spell
Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01, n.s.: non-significant.
Descriptive statistics of migrant network variables by migration status at the last year of the migration spell
Descriptive statistics of migrant network variables by migration status at the last year of the migration spell
Note: All variables measured at last year of the migration spell. All network variables exclude the partner/ children.Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01, n.s.: non-significant.
Notes
-
[*]
École nationale de la statistique et de l’administration économique (ENSAE), France.
-
[**]
Forum Internazionale Ed Europeo Ricerche Sull’Immigrazione, Italy.
Correspondence: Sorana Toma, Ensae, Laboratoire de sociologie quantitative, 60 rue Etienne Dolet, 92240 Malakoff, France, email: sorana.toma@ensae.fr -
[1]
For a detailed description of the entire dataset and the MAFE project countries: http://www.mafeproject.com/
-
[2]
While African destinations still attract a large number of Senegalese migrants, their share has been declining in recent years, in favour of European destinations, which is why the MAFE survey decided to focus on migrations between Africa and Europe (Beauchemin, 2012).
-
[3]
A sampling frame was available in Spain (the Padrón municipal register) from which a random sample was obtained. In contrast, quota methods were applied in France and Italy, with recruitment of respondents through a variety of channels to limit biases (snowballing, intercept-points, contacts from origin households or through migrant associations and public places). See Beauchemin and Gonzalez-Ferrer (2011) and Beauchemin (2012) for more information on the survey design and its sampling biases.
-
[4]
The 24 cases of remigration to destinations other than these three countries are excluded from the analysis by censoring them at the last year of the migration spell. Analyses including these cases reach similar results.
-
[5]
When a permit is not required, this is usually because the migrant has, or has acquired, the nationality of the destination country.
-
[6]
The country from where remigration or return may occur (not the country where the respondent has remigrated to).
-
[7]
Recent migrants have been abroad for 3 years or less, experienced migrants between 4 and 10 years, long term migrants for at least 11 years.
-
[8]
Unfortunately, there are very few cases of female remigration, so men and women cannot be analysed separately.
-
[9]
This analysis is situated at the individual level: to each individual corresponds one mobility trajectory.
