Keywords

4.1 Introduction

With an increase in global migration (at least until the onset of the Covid-19 pandemic) and with recent technological and communication developments that facilitate migrants’ transnational engagement, there is a growing interest in understanding the determinants of economic transnationalism. Monetary remittances are a form of transnational economic activity that is part of the relations which migrants may keep across nation states’ geographical, political and cultural boundaries (Al-Ali et al., 2001a; Glick Schiller, 2008; Vertovec, 2009). The amount of transnational monetary transfers has increased rapidly: in 2018, remittances to lower- and middle-income countries were estimated to have reached US$529 billion (then around €448 billion), an increase of almost 10 per cent compared to the previous year (World Bank, 2019).

Besides their potential adverse effects (e.g. increased inequality) it is, overall, often argued that remittances have great developmental potential (De Haas, 2005). Monetary remittances can contribute to improving the living conditions of households in migrants’ origin countries, as they provide a safety-net in poorer areas and go directly to people in need. Moreover, remittances can contribute to the increased economic activity and investment propensity of families abroad, potentially leading to an increase in the general prosperity of migrant-sending regions in the long term (Taylor et al., 1996; De Haas, 2003, 2005).

While there has been great interest in remittances for decades – due, inter alia, to their potentially positive impact on the livelihoods of migrants’ families (e.g., Taylor et al., 1996; De Haas, 2003) – the research field remains limited in certain aspects. Particularly, as we argue in this chapter, most studies on economic remittances are dominated by a static view of migration: it assumes that people leave their origin countries behind, settle down in one reception country and then start remitting (Malkki, 1992). This leads to a limited understanding of remittance-sending behaviour because it assumes that migrants remit only when they consider themselves to be settled in the destination country. However, many migrants may also remit during their journeys. While the vast majority of research focuses on migrants who consider that they have reached their final destination, little is known about the economic remittance-sending behaviour of migrants who are on the move.

In fact, the increasing restriction of legal migration and intensified controls at European borders have led to more complex, fragmented and diverse migration trajectories (Dahinden, 2010; Collyer et al., 2012; Schapendonk & Steel, 2014; Snel et al., 2021). Migrants who spend longer periods of time en route may have diverse (and more dangerous) experiences between departure and arrival (Schapendonk, 2009). Thus far, there is little scientific understanding of how these experiences may affect migrants’ transnational engagements such as remittance-sending. If we assume that remittances mostly support families in migrant-sending countries by providing a safety-net or an income enabling them to maintain a basic standard of living, it is relevant to gain more insight into how these situations influence migrants’ remittance-sending behaviour. Our research contributes to filling this gap in the literature by exploring differences in remittance-sending behaviour among migrants in relation to their mobility patterns. More specifically, we aim to understand how and why migrants’ remittance-sending behaviour differs when they consider themselves to be on the move and when they are settled. Using retrospective, longitudinal data from the Migrations between Africa and Europe (MAFE) project, we investigate how being on the move relates to the remittance-sending behaviour of Senegalese, Ghanaian and Congolese migrants along their migration trajectories.

4.2 Sub-Saharan Migration Routes to Europe

As a response to increased border controls or changing situations in transit countries, migration routes, trends and patterns between sub-Saharan Africa and Europe have shifted continuously over time (Lessault & Beauchemin, 2009; Schoumaker et al., 2018). While a large number of sub-Saharan migrants entered Europe by air, a significant number reached Europe’s territories overland and by sea (Caarls et al., 2021). Among this latter group, a wide variety of modes of transport and routes were used to enter Europe. Around the turn of the twenty-first century, the main departure points for sub-Saharan migrants were from Northern Morocco for travel to Spain, Melilla or Ceuta, from Libya and Tunisia to Lampedusa, Sicily and Malta and from the West African coast for migrants travelling to the Canary Islands (Schoumaker et al., 2018). Senegalese migrants mainly set off either directly to the Canary Islands by boat or overland to Morocco or Mauritania. Ghanaians either headed to Africa’s West Coast or crossed the Sahara to Morocco or Libya, while the Congolese commonly migrated to Europe via North Africa.

Additionally, ‘traditional’ destination countries have changed for sub-Saharan African migrants (Mazzucato et al., 2015). Although increased border controls have made migration to Europe more difficult, the increasing lack of opportunities in African destination countries and the growing demand for cheap labour in Southern Europe in particular have increased migration from sub-Saharan Africa to Europe and people have emigrated to new destination countries. Whereas around 3 million sub-Saharan African migrants were living in Europe in 2000, this number had increased to 3.9 million by 2011 (Schoumaker et al., 2018). In the 2000s, Senegalese destinations in Europe were mainly Italy and Spain, even though they primarily used to migrate to France. Ghanaians mostly migrated to the UK, the US and new European destination countries such as the Netherlands, Belgium and Germany. Congolese migration itineraries shifted differently. With the abolition of apartheid in South Africa and the ending of the Angolan War in the early 2000s, new economic opportunities were created in the region, which decreased the number of Congolese migrating to Europe.

4.3 Conceptualising Migration Trajectories

In order to show that migration experiences go beyond initial points of ‘departure’ and ‘arrival’, scholars have often used various terms to acknowledge that migration does not usually have a clear end-point but can involve secondary, onward, return or circular migration (Ehrkamp, 2019). In Chap. 2, Erdal et al. suggest the term ‘complex migration trajectories’. As such, migration is characterised by multidirectionality and complexity, with deep entanglement between mobility and immobility along migrants’ routes (Düvell, 2006; Collyer, 2007; Ehrkamp, 2019). To recognise this (im)mobility – aggravated by strict state immigration regimes (Düvell, 2006) and therefore, at least to a certain extent, involuntary – the term ‘transit migration’ has commonly been used. This is defined as ‘the situation between emigration and settlement that is characterised by an indefinite migrant stay, legal or illegal, and may or may not develop into further migration depending on a series of structural and individual factors’ (Papadopoulou-Kourkoula, 2008, 4) and is often thought to be anticipated before migrants depart from their origin countries (Castagnone, 2011).

However, as transit migration is often not a choice, nor is it always intended or planned, the ‘initial’ aim of reaching a specific country is not always realised and changes along their journeys (Düvell, 2006; Collyer, 2007). Rather, transit stays are characterised by indefiniteness that may result in (involuntary) settlement (Papadopoulou-Kourkoula, 2008). In addition, merely categorising transit migration as a phase between departure and arrival does not reflect migrants’ changing aims and different experiences along their trajectories, whereby migrants change their plans, adapt to new environments and explore opportunities in countries of stay (Andersson, 2016; Massey et al., 2016; Staring, 2018). A broader interpretation that reflects migrants’ embodied experiences along their journeys is therefore needed. Hence, we turn to Coutin’s (2005) term ‘en route’Footnote 1 to explore how transit migration can be experienced as a ‘liminal state that positions migrants simultaneously outside (in transition, not yet arrived), yet inside (travelling through), national space’ (Coutin, 2005, 196). Along migration journeys, migrants can be physically present in ‘precarious transit zones’ (Hess, 2012) while being excluded from the rights and protection mechanisms which citizens of that state enjoy.

Being en route/on the move is often characterised by particular living conditions that increase migrants’ vulnerability and poverty – conditions such as semi-protection, social exclusion, a lack of state protection or outright hostility (Coutin, 2005; Papadopoulou-Kourkoula, 2008). This precariousness is mainly driven by two mechanisms (Düvell, 2006). First, it is produced by nation states, international regulations or border regimes (e.g., the increased securitisation of the EU’s outer borders) as they exclude certain migrants from conventional protection regimes and restrict them from legally reaching their territories, thereby creating challenges for the protection of human rights and access to basic services. Second, this precariousness is driven by social class. Whereas wealthier migrants can afford to pay for a direct flight or provide savings to obtain a visa, poorer migrants are inclined to use less expensive modes of travel, usually overland or by sea (Düvell, 2006).

In this chapter, we explore how Senegalese, Ghanaian and Congolese migrants engage in sending economic remittances to their origin countries while considering themselves to be on the move rather than being settled. We understand being en route as a lived experience (Düvell, 2006), meaning that migrants themselves are ‘experiencing subjects’ (Eastmond, 2007) who have defined, in hindsight, what they experienced as stays to settle or stays with the intention of moving onwards. This overcomes the constraints that previous research raises, arguing that concepts – such as transit migration – should be studied not only as a socio-political condition but also as a way of experiencing the world (Willen, 2007). It allows moving beyond over-generalised notions of transit migration and acknowledges the wide diversity in transit zones and migrants’ different realities in transit (Collyer et al., 2012; Hess, 2012).

Against this backdrop, we argue that our regard as researchers should move towards the experiences of migrants en route who are also likely to be sending remittances from different places which they do not consider to be their final destination. These remittance flows can be both to countries of origin or elsewhere where they have social ties (as also demonstrated in Patterson’s Chap. 10 in this book). In fact, previous research has also highlighted the importance of reverse remittances, which support migrants on the road and abroad (Mazzucato, 2011). While we acknowledge that remittances can be sent to and from diverse places, due to data limitations this chapter focuses on the remittance-sending behaviour of migrants only towards their country of origin.

4.4 Remittance-Sending Capacity and Incentives of Migrants En Route

Although being en route is initially expected to be a temporary experience without the intention to settle down in the country of stay, it is often unclear how such a stay will develop (Düvell, 2006). Migrants might be hindered in moving onwards and their future decisions might be dependent on the available and affordable choices, a local network or migrant smugglers or brokers (Collyer, 2010). Following the life-cycle theory of consumption (Dustmann, 1997), this insecurity and precariousness could lead migrants in temporary stays to save their money in order to ensure their future consumption. This implies that they are more likely to prioritise money as a means to contribute to reaching their future destinations, instead of investing it in relationships with the origin country. Consequently, money sent through remittances is less likely be prioritised over more direct expenditure on their onward migration, leading migrants en route to (temporarily) refrain from sending remittances. Therefore, we expect that migrants en route are less likely to remit compared to migrants who consider themselves to be settled.

Several factors could further explain the expected difference between remittance-sending behaviour en route compared to that when settled. The particular living conditions and opportunities of being en route are likely to influence migrants’ remittance-sending behaviour, as engagement in transnational activities (i.e. sending remittances) is closely related to migrants’ transnational capabilities to engage in these exchanges (Bilgili, 2014, 2015). These transnational capabilities include two elements. First, migrants need the capacity to engage in transnational activities (Al-Ali et al., 2001b). When it comes to monetary remittances, this depends, for example, on migrants’ wages, on alternative forms of income or on their savings. These capacities are influenced by the local context and the characteristics of the country of origin, as the skills and resources available to migrants to engage in transnational activities are influenced by their social networks, opportunities and the length of time spent in the country of stay (Carling & Hoelscher, 2013). Second, engagement in sending remittances depends on the migrants’ incentives to remit and their willingness to prioritise money spent on these remittances over other expenditure (Bilgili, 2015). These incentives depend on migrants’ connections and attachment with both their origin country and their countries of stay. In order to remit, it is a prerequisite that migrants identify with the social, economic and political processes in their origin countries (Al-Ali et al., 2001b).

In this research, we first investigate whether migrants’ employment status in their country of stay, as a transnational capacity, can explain possible differences in remittance-sending behaviour. Previous research has shown that income and employment status are closely related to sending remittances, as migrants struggling to cover their living expenses in the host country are limited in their possibilities to remit (Hagen-Zanker & Siegel, 2007; Carling & Hoelscher, 2013). Several studies found that those with higher incomes or with permanent contracts were more likely to remit (Carling & Hoelscher, 2013; Bilgili, 2014, 2015). Similarly, those with more stable economic positions are likely to be in less precarious situations and are, therefore, more likely to remit.

Migrants who are en route rely more on informal labour opportunities and social networks influencing their strategies and objectives (Hess, 2012). They might be more likely to work in precarious jobs and therefore have more unstable labour-market positions. Moreover, being in a perceived situation of limbo – neither ‘here’ nor ‘there’ – can be paralysing. It might prevent them from actively looking for employment, as the stay is probably expected to only be temporary but could, meanwhile, develop in a longer or even a permanent stay (Al-Ali et al., 2001a). Being in a more precarious situation might make it more difficult to save money and send remittances (Cox et al., 1998). Therefore, we expect that migrants who are en route are less likely to be employed and are therefore less likely to remit, compared to migrants who are settled.

Another factor influencing migrants’ capacity to remit is their legal status in the host country (Al-Ali et al., 2001a). Although there is no consensus in the literature, there is a common view that the insecurity of migrants’ legal status poses obstacles to their engagement in remittance-sending behaviour (Van Meeteren, 2012). Irregular migrants who are structurally excluded might experience (legal) barriers to formal institutional participation, hampering their ability to engage in cross-border activities (Portes, 2001; Bloch, 2008; Vickstrom & Beauchemin, 2016) and undermining a stable situation compared to regular migrants. Since not having a legal status could also make migrants reluctant to engage in activities that jeopardise them – as not abiding by the law could lead to fines, imprisonment or deportation (Pinger, 2010; Vickstrom & Beauchemin, 2016) and the insecurity of no legal status also limits the freedom to move around – this situation could also lead to psychological problems, such as apathy and non-commitment (Al-Ali et al., 2001a).

Importantly, being en route is not intrinsically related to illegality, as it is a process intersecting various other migration categories. Irregular migrants, asylum applicants, refugees, documented migrants and migrants who have been trafficked may all experience fragmented journeys (Papadopoulou-Kourkoula, 2008; Collyer, 2010). However, migrants who are en route are more likely to be excluded from conventional protection regimes, challenging the protection of their human rights and access to basic services (Düvell, 2006). These structural barriers might, in turn, hamper their likelihood of obtaining a legal status. Moreover, as being on the move is envisioned to be temporary, migrants might be less inclined to make an effort to obtain a legal status in this host country. Accordingly, we expect that migrants en route are less likely to have a fully regular status and are therefore less likely to remit, compared to migrants who consider themselves to be settled.

Although migrants’ economic position and legal status while being en route are expected to hamper their remittance-sending behaviour, these capacities play no role if there is no obligation or incentive to send money to their origin countries (Bilgili, 2014). Previous research has shown that different incentives can determine monetary transfers. First, remittances can be sent out of pure self-interest, whereby migrants invest and accumulate assets in their origin country (Hagen-Zanker & Siegel, 2007; Ruiz & Vargas-Silva, 2009). Feelings of attachment to the origin country might be higher for those who have assets, resulting in a strong motivation to remit. Second, economic remittances can also function as a contractual arrangement between migrants and their families in origin countries, as a strategy to diversify income resources to support the household (see Lucas & Stark, 1985). In this way, both migrants and households can mitigate their risks and are insured against financial shocks: the household supports the migrant by contributing to their migration costs and, in turn, receives remittances from the migrant as a form of income. Remittances are then sent to repay the loans which migrants took out to pay the costs of migration. Lastly, migrants could also remit because they want to take care of their families for altruistic reasons, as they feel concerned about their families’ well-being (Hagen-Zanker & Siegel, 2007; Ruiz & Vargas-Silva, 2009).

Although previous research found that having both assets and family in origin countries is an important incentive for migrants to send remittances, their effect may be less strong for migrants who are in more precarious situations, such as those who still consider themselves to be on the move. Therefore, we expect that migrants who own assets and/or have family members in their origin countries are more likely to remit but that this relationship will be weaker when they are en route.

4.5 Methods

4.5.1 Data and Participants

This research used data from the Migrations between Africa and Europe (MAFE) project. Identical surveys were conducted in six European countries – Spain, Italy, France, Belgium, the UK and the Netherlands – and the three sub-Saharan African countries of Ghana, Senegal and the Democratic Republic of Congo or DR Congo (Beauchemin et al., 2014). The countries were part of three distinct migration systems (see Fig. 4.1): Senegalese migration was studied in Senegal, France, Spain and Italy, Congolese migration in DR Congo, Belgium and the UK, and Ghanaian migration in Ghana, the UK and the Netherlands. This multi-sited approach allowed us to realise the aim of the MAFE project: to recognise that migration is not a linear process (from Africa to Europe) but, rather, a process between Africa and Europe, by also including onward, circular and return migration (Beauchemin, 2012).

Fig. 4.1
A model diagram illustrates migration systems 1, 2, and 3. It includes Senegal, which includes France, Spain, and Italy,D R Congo, which includes Belgium and Great Britain; and Ghana, which has Great Britain and the Netherlands.

Structure of MAFE data depicting where retrospective surveys have been conducted. (Adapted from Beauchemin et al., 2014)

Full size image

This research only used data from respondents who had been, at the time of data collection, staying in one of the six European countries for at least one year. Data collection only took place in the major cities where the majority of migrants stayed. Except for Spain, quota sampling was used to select respondents (Beauchemin et al., 2014). These quotas were set by age and gender. Additionally, in France, socio-occupational status was included as an extra criterion and in Belgium and the UK the place of residence was a criterion. Different recruitment methods were used, including approaching respondents in public spaces and churches, through snowballing or through interviewers’ contacts. Only in Spain was a random selection of participants possible through the population register. The data were collected between 2008 and 2010 (Beauchemin et al., 2014). Respondents were eligible if they were between 25 and 75 years old at the time of the survey, were born in one of the three African countries and had – or had had – the nationality of their origin country.

The data comprise individual life histories collected through biographic questionnaires. Respondents were asked to provide retrospective information on their family, economic and residential history. In total, 1450 migrants were interviewed in Europe. After only selecting those who were over 18 years old at the time of emigration from their origin country, our final sample consisted of 1439 respondents, of whom 601 were Senegalese, 426 were Congolese and 412 were Ghanaian (see Table 4.1 for further specification). On average, respondents were 39 years old when they completed the survey. Moreover, most were generally highly educated: at the time of the survey, 41.7 per cent had completed university, with only 5.6 per cent reporting they did not complete any schooling.

Table 4.1 Sample size per European country and migrants’ countries of origin
Full size table

4.5.2 Measures

  • Sending economic remittances. The dependent variable in this research records whether or not respondents sent economic remittances to their origin countries in a specific year – over 95 per cent did so. Given our focus in this research, we excluded remittances that were sent to other countries and focused only on whether respondents remitted in a particular year or not. Since respondents only reported when they did remit as a response to the question, the years when they were abroad and did not report sending remittances were coded as 0. Consequently, this variable had no missing values.

  • Being en route. For the independent variable ‘being en route’, which indicates whether a migrant is considered to be on the move or settled, we combined information from three questions into a binary variable.Footnote 2 We defined being en route as either a short or a long stay retrospectively described as a ‘transit stay’ by the respondent or as ‘not having a clear idea about the next destination’. Being settled included when the respondent considered the country as a place to settle or at the time when the respondent arrived in the country of stay he or she had considered it to be their final destination, where they had planned to go from the outset. From the total number of observed years, information from 5 years was missing (less than 1 per cent).

  • Economic position. Two questions were used as indicators to measure respondents’ economic position. First, it was determined whether respondents were employed or unemployed during a stay. A binary variable was created from a question that distinguishes between 0 (unemployed; also includes being an apprentice, trainee or intern or helping family member in a family business or farm) or 1 (employed). Those who answered an earlier question saying that they were studying, unemployed, homemakers, retired or inactive did not get this question and were coded as being unemployed. For respondents with missing values on this question, the International Socio-Economic Index of occupational status (ISEI) values were used to determine whether they were employed or not. After this, 1 per cent of the responses to this question were missing. Respondents’ subjective wealth was measured by asking whether, all-in-all, respondents would say that, during this period, they had enough to live on day-to-day, with three categories: Yes absolutely; No, not at all; It depended. The answer categories were re-coded so that a higher number reflects an increase in satisfaction. From the total number of observed years, 1 per cent of the responses to this question were missing.

  • Legal status. Two questions about respondents’ residence and work permits throughout their lives were combined to create the variable legal status. For each year, we recorded whether or not respondents had a residence permit or did not need one. The same was done for work permits. These two indicators were combined into a variable that consisted of three categories: precarious status (no residence permit), semi-regular status (only residence permit) and fully regular status (both residence and work permit). As a work permit is usually combined or granted simultaneously with a residence permit or as a prerequisite for obtaining a work permit, only having a work permit but not a residence permit was considered a ‘precarious legal status’ (Vickstrom, 2014) – an insecure status which does not grant similar (basic) rights compared to a residence permit. When respondents included information on one permit but had a missing value on the other permit in the same year, we coded the missing value as 0 to keep the information provided on the other permit. After this, in 8.8 per cent of the total number of observed years, information on neither residence nor work permit was reported and they were therefore coded as missing values.

  • Family in origin country. We used available information about respondents’ children, partner(s), parents, brothers and sisters to determine, for each year, whether or not they resided in the country of origin while the respondent was abroad. From this information, a binary variable was constructed: 1 (has family member(s) in the country of origin) and 0 (has no family members in the country of origin). When no family members were reported, we coded these years as not having family members in the origin country.

  • Assets in origin country. One question captured whether respondents had been or still were the owners of plots of land, houses, businesses, ventures or commercial premises in their origin countries. If they owned any of these assets, they provided us with the start and end dates. From this information, we constructed a binary variable that captures whether or not respondents owned assets in a specific year. When respondents did not report any assets in a particular year, we coded it as 0.

  • Controls. We controlled for respondents’ sex and country of birth (both recorded by the interviewer), as previous studies found different remittance-sending patterns by national origin and sex (Carling, 2008; Bilgili, 2014). Since Carling (2008) also found that age positively influences remittance-sending behaviour when controlling for time since migration, respondents’ age and length of stay were controlled for. Length of stay was constructed from the number of years someone had spent in a country since arrival. Upon return to a previous host country, the variable continued counting from the years they stayed during their earlier visit. Upon re-migration to another country, the variable started counting at Year One again.

Additionally, educational level was included as a control because a higher educational level can result in a higher income and a more stable job, affecting the propensity to remit (Bilgili, 2014). Respondents reported the number of years of schooling and the last school year they attended. Using this information, a time-variant measure for educational level was constructed. Respondents’ educational level was lowered by one point for each year prior to the final school year. The educational level stayed the same in periods during which education was interrupted. From this range, education was coded in six categories: no schooling, some schooling (1–6 years), primary school (7–10 years), lower-secondary (11–13 years), upper-secondary (14–15) and university (16–23). There were no missing values for the control variables.

4.5.3 Analytical Strategy

To account for the longitudinal data structure, a person period file was created that included yearly information for each respondent. This means that a single respondent could appear in multiple lines of the data file, reflecting several years of their life. This file reflected information valid on the 1st of January of each year (e.g., their employment status, subjective wealth, legal status, etc.). Since we were interested in migrants’ remittance-sending behaviour in the years in which they lived abroad, we excluded the years in which respondents stayed in their origin countries. The 5 years with missing values on the independent variable ‘being en route’ were excluded from the analysis. In total, the number of observed years from all respondents together was 20,590.

For the main analyses, we ran the models in Mplus, taking into account the multi-level structure by clustering the data on respondents. Mplus allowed us to integrate the different paths of our model simultaneously using Structural Equation Modelling (SEM). Missing values on the dependent variables were handled using the Full Information Maximum Likelihood (FIML) option in Mplus. FIML ignores missing values on a dependent variable if there are valid cases on other variables in the same person–year.

We decided to use the complex survey method to analyse our data. Compared to a random effect specification, the complex method makes fewer assumptions about a person over time and is therefore considered to be more reliable, with less bias but also less power. We used the robust weighted least-squares estimator because the dependent variable is binary. For the standard errors of the direct and indirect effects, bootstrapping with 1000 samples was used. For the categorical mediator permit, we calculated the indirect effects through marginal effects while, for the ordinal mediator subjective wealth, a continuous latent variable with categorical indicators was constructed to calculate the indirect effects.

4.6 Results

4.6.1 Descriptive Results

Table 4.2 contains information on both time-variant and time-constant variables. For the time-variant variables, the information presented was an average across the total number of observed years of all respondents. For the time-constant variables, the observations for each respondent did not vary across the years (e.g. sex, country of origin) and we therefore only presented these descriptive results per individual and not across the years (1439 respondents).

Table 4.2 Descriptive results for full sample
Full size table

In almost two-thirds of the total observed years, respondents sent remittances. However, respondents reported being en route in only 10 per cent of the total number of observed years. When comparing the number of temporary stays to the total number of migration trips (1651), 18 per cent of the total number of different migration trips were classified as migrants being en route.

Over the years, respondents were often employed (70 per cent) and in over three-quarters of the observed years, respondents stated they had quite enough to live on. In the majority of observed years (71.1 per cent), respondents reported having a fully regular status, compared to having no residence permit in only 15.4 per cent of the years. In a quarter of the observed years, respondents reported having family in their origin country and in about a third of the observed years said that they owned one or more assets in their origin country.

4.6.2 Analytical Results

We estimated three structural models with control variables (Table 4.3). Following our first hypothesis, we initially estimated the direct relationship between being en route and sending remittances in a model without mediators and predictors (Model A). Then we estimated a model which included them (Model B). Third, we estimated our full model (Fig. 4.2) and included the hypothesised interactions between being in transit and owning assets and between being in transit and having family in the origin country (Model C). As we found one of the hypothesised interactions was significant and the effect of the other one did not change compared to Model B, we turned to Model C for the analysis of our results regarding the remaining hypotheses.

Table 4.3 Results of logistic regression analysis on sending remittances (N=20,590, estimated with WLSMV)
Full size table
Fig. 4.2
A network diagram illustrates four levels. It indicates the hypothesized interactions between being in transit and owning assets and being in transit and having family in the origin country.

Conceptual model

Full size image

As in our first hypothesis, we expected that migrants who were on the move would be less likely to send remittances, compared to migrants who considered themselves to be settled. However, we found that the relationship between being en route and remitting was insignificant (Model A).Footnote 3

Secondly, we expected migrants’ economic positions to partially explain the main relationship. We indeed found that employment status and subjective wealth were both significantly and positively related to the probability of sending remittances (resp. β = .260, p < .001; β = .076, p = .042). This means that being employed or being more satisfied with their financial situation made them more likely to remit. Moreover, we found that migrants who were on the move were less likely to have a satisfactory subjective wealth (β = −.286, p = .003) but, on the other hand, were more likely to be employed (β = .271, p = .003) compared to those who considered themselves to no longer be on the move.

Contrary to this hypothesis, we found that employment status was significantly and positively mediating the relationship between being en route and sending remittances (total indirect effect: β = .071, p = .007). This means that migrants who were on the move were more likely to be employed and were, therefore, more likely to remit. We did not find that migrants’ subjective wealth negatively mediated the relationship between being en route and sending remittances.

Moreover, we hypothesised that migrants who were en route were less likely to have a fully regular status and would, therefore, be less likely to remit. However, we did not find that having a more secure legal status was significantly related to sending remittances. We did find that those who were en route were less likely to have a fully regular status, compared to those who were settled (β = −.331, p = .002).

According to our fourth hypothesis, we expected that migrants who owned assets or had family in their origin country would be more likely to remit but this relationship would be weaker when they were en route. We in fact found that the odds of remitting were 1.3 times as high for those who had assets in their origin country compared to those who did not have assets (β = .230, p = .001). Contrary to our expectations, we found that being en route significantly and positively moderated this relationship (β = .319, p = .046) as the relationship between owning assets in the origin country and sending remittances was stronger for those who were en route. However, we did not find that having family in the origin country was significantly related to sending remittances and we also did not find support for the notion that being en route significantly moderated this relationship.

4.7 Discussion

This chapter aimed to explain the extent to which and how Senegalese, Ghanaian and Congolese migrants’ remittance-sending behaviour is affected by being en route (or not) along their migration trajectories. Overall, we hypothesised that migrants en route would (temporarily) refrain from sending remittances but we did not find evidence of this. Although we did not find overall support for the hypothesis that being on the move decreased migrants’ probability to remit, our results support the idea that migrants’ capacities influenced their remittance-sending behaviour. Those who were employed and were more satisfied with their financial situation were more likely to remit. This is in line with previous research and supports the capabilities approach. In other words, migrants’ engagement in transnational activities depends on their transnational capacities (Al-Ali et al., 2001b; Bilgili, 2014, 2015). However, migrants’ legal status did not affect their remittance-sending behaviour. This is not uncommon, as there is no consensus in the literature that having a more secure legal status is always positively related to migrants’ remittance-sending behaviour. Previous research, for example, found that undocumented migrants remitted more compared to documented ones (Sana, 2005; Amuedo-Dorantes & Pozo, 2006).

As expected, we found that being on the move was negatively related to having a full and regular status. Migrants in temporary stay situations were more likely to be excluded from conventional protection regimes which, in turn, imposes structural barriers that hamper their chances to obtain a legal status (Düvell, 2006). Moreover, being en route was also negatively related to migrants’ subjective wealth. This, together with a reduced legal status, is in line with previous research stating that transit zones are characterised by precariousness (Hess, 2012).

Contrary to our predictions, we found that being on the move was positively related to being employed, compared to having reached the final destination. This, in turn, also partially explained why migrants would, instead, be more likely to remit while they were en route. As transit countries are characterised by greater informality (Hess, 2012), migrants might indeed be more likely to be employed because of a large informal sector, possibly explaining this positive relationship. Additionally, in many economically developed destination countries, there is an educational mismatch for immigrants on the labour market because, for example, of discrimination and the lack of recognition of diplomas or language skills (Piracha & Vadean, 2013). This could increase the likelihood of (initial) unemployment in economically developed destination countries, compared to areas where migrants stay temporarily.

These findings partly indicate that longer and more complex migration journeys are characterised by increased precariousness, as migrants reported a worse subjective wealth and a more precarious legal status while they were en route. However, the higher likelihood of being employed in these periods – which in turn increased their probability of remitting – could have worked as a safety-net in times of uncertainty. In this way, migrants then remitted to their origin countries to strengthen their connections and ‘purchase’ insurance at times of personal risk (Lucas & Stark, 1985; Amuedo-Dorantes & Pozo, 2006) instead of (temporarily) refraining from remitting. In other words, in light of the increased likelihood of their having to go back home due to their precarious situation and increased future insecurity, migrants on the move may have chosen to send remittances back to their country of origin (Dustmann, 1997).

Following the capabilities approach, migrants’ attachment to their origin country was also an important determinant of sending remittances. As expected, we found that owning assets in origin countries was positively related to sending remittances, compared to not owning assets. This is in line with previous research that found that migrants who are more attached to their origin countries are more willing to prioritise spending money on remittances over other expenditure (Carling & Hoelscher, 2013). However, no such relationship was found between having immediate family in origin countries and sending remittances. Our mere focus on the nuclear family in the origin country could, therefore, possibly explain the absence of a significant relationship, as those without immediate family in the origin country might also remit to other relatives.

Additionally, we expected that, for those who were in transit, the positive relationships between attachment to origin countries and remitting would be weaker, since they were in more precarious situations and would, therefore, temporarily refrain from sending remittances. However, we found the relationship between owning assets and remitting was, in fact, stronger for those who were in transit. This could indicate further support for the argument that insecure or precarious transit stays might lead migrants to intensify or strengthen their connections with their origin countries by sending remittances. Future research should study in more depth these dynamics and also consider reverse remittances for migrants who are en route. Focusing on bi-directional engagements where migrants on the move also receive transnational social and economic support is central not only for understanding the experiences of migrants whose journeys are longer, fragmented and more complex than ever before but also for shedding light onto the effects of these journeys on the lives of those left behind.

4.8 Data Limitations and Multi-sited Transnationalism Research

Firstly, MAFE data reflect the experiences of migrants who arrived in European countries where they considered themselves to be settled. This means that the experiences of migrants who were in transit countries and were hindered from moving on, were not represented in the data. This implies that we might have underestimated the effect of particular living conditions and opportunities for migrants on the move, as we only had information from those who reached one of the six European countries. As such experiences are partially driven by social class (Düvell, 2006), poorer and more vulnerable migrants might experience greater precariousness, possibly preventing them from moving onwards.

Secondly, the use of retrospective data has several limitations. Retrospective information about transitory periods could reflect what respondents believed was happening, instead of what was actually happening (Düvell, 2006). It is, therefore, difficult to determine whether respondents reported their actual intentions upon arrival in a country or whether subsequent events, such as re-migration or settlement, influenced the way in which they classified their host country.

Considering both limitations, using a so-called mobile methods approach could be one way to move away from selecting European destination countries as the starting point for sampling migrants (Amelina & Faist, 2012). Quantitative or qualitative data are then collected by following migrants along their trajectories instead of sampling them in (European) destination countries. By interviewing the same migrants at different moments along their migration trajectories, migration can then be studied longitudinally and capture migration dynamics and changing aspirations (Castagnone, 2011). Such a multi-sited and longitudinal method would move away from ‘methodological nationalism’ and make steps towards a truly transnational perspective for studying complex migration trajectories and transnational engagements.

In conclusion, our research provides a deeper understanding of how sub-Saharan African migrants’ trajectories between Africa and Europe affect their remittances-sending behaviour. By moving away from the simplistic view of migration as a one-time act from point of departure to destination, we aimed to better reflect the realities of migrants’ lives and to bring this approach also to the study of remittances (Collyer & de Haas, 2012; Snel et al., 2021). Although we did not find support for the idea that being en route is negatively related to the probability of sending remittances, our results demonstrated that migrants’ capabilities to remit change along their migration trajectories. We therefore conclude by highlighting the importance of including a(n) (im)mobility perspective in the study of remittances and the need for a fuller understanding of how frictions and experiences along more-complex migration trajectories affect remittance-sending behaviour.