1 Introduction

In classical migration theories, family dynamics play an important role for describing mobility patterns (Lee 1966), and the introduction of the family life course perspective to mobility and migration research has increased attention towards this interdependence of family and migration (Kley 2011; Kulu and Milewski 2007). The prevailing perspective considers family events as critical determinants of residential relocations and international migrations. It suggests strong associations between, e.g., the timing of a marriage, parenthood, childbirth, or divorce and the timing of spatial mobility. Spatial mobility here basically functions as a mechanism for adjusting to changing household and family situations (Vidal and Huinink 2019, p. 596), and this may include within- or across-country migration processes. Another perspective of this interdependence entails the effects of migration on family events such as partnership formation patterns, parenthood, or union dissolution. Moving is a potentially stressful life event and challenging circumstances of the migration or the settlement process may lead to changes in partnership and family life (Boyle et al. 2008; Cooke 2008).

The current chapter deals with this latter perspective, focusing on partnership dissolution in the course of an international migration. Thus, different from research studying mobility outcomes following the dissolution of a partnership (see e.g. Cooke et al. 2016; Mikolai and Kulu 2018; Wall and von Reichert 2013), we ask under which conditions an international migration increases the risk of separation among couples. In our exploratory analyses, our specific interest lies in the impact of the couples’ migration patterns and the way the migration decision was made by the couple, with reference to the concept of the “tied migrant” (Mincer 1978). Our sample includes individuals in marital or non-marital relationships. So far, the role of family migration on the subsequent stability of a union is largely underexplored; existing knowledge is surprisingly scarce as only a handful of studies have addressed this issue at all. In light of increasing international mobility and related demands on the labour market, however, it seems highly crucial to shed more light on this topic. In the remainder of this chapter, we outline the theoretical background and discuss existing findings, then we turn to our data and models, and ultimately describe and discuss the findings. The chapter ends with concluding remarks.

2 Theoretical Background

Studies on divorce summarise numerous predictors such as a young age at marriage, low educational resources, unemployment, parental divorce experiences, higher-order or interethnic relationships, as well as relationship characteristics such as frequent conflict or low levels of trust (Amato 2010, pp. 651–652; Wu and Penning 2018). The underlying mechanisms can be specified with reference to the two theoretical approaches of social exchange theory and the micro-economic theory of marital instability (Hill and Kopp 2015; Wagner and Weiß 2003). Social exchange theory suggests that all social relationships, including partnerships, entail the exchange of material (e.g. money) and immaterial (e.g. care, love) resources as well as certain costs (e.g. conflicts), and in micro-economic theory the focus is on the partnership utility for the individual (Becker et al. 1977). Based on these perspectives, partnership stability (or its dissolution) can be considered as the result of perceived alternatives to the partnership and the partnership quality, i.e. the net benefit of the beneficial resources and the costs of the partnership. Partnership quality itself varies depending on the matching of the two partners, the investments of the partners in the relationship (e.g. the existence of children, the degree of institutionalisation, the division of household labour) and the relational maintenance (e.g. interaction patterns and conflicts).

International migration is rarely addressed in research on partnership stability. One argument refers to higher risks of union dissolution in case of stressful events and the proposition that moving across borders is such a stressful event (Boyle et al. 2008). Accordingly, the need for changes in routines, roles, and identities in the acculturation process (Berry 1997) or the lack of social networks and support following migration (Nauck 2007) put a strain on the couple. Alternative explanations suggest that higher dissolution risks may result from freeing the couple from those social networks that discouraged separation, from acculturation processes in new contexts in which separation is more common and socially acceptable, or from the better opportunities for new partner choice in the new place of residence (Boyle et al. 2008).

Another line of argument stresses increasing differences and inequalities between the partners in the course of a migration. Accordingly, spatial relocations and international migrations are often undertaken primarily for the benefit of one partner (see Erlinghagen 2021). Whereas this partner’s career is enhanced, the career of the “tied spouse” might suffer. Mincer (1978) was the first to examine the phenomena of “tied moving” and “tied staying”. He defined “tied” persons in the family as those whose gains from migration were dominated by gains or losses of the spouse, i.e. by net personal loss but net family gain (Mincer 1978, p. 751). In such cases of “tied migration”, the partners may experience their new lives differently, including differences in enthusiasm about the new context, different hardships in the course of acculturation, and diverging experiences in establishing new social networks. This inequality in the context of “tied migration” might put strain on the relationship (Boyle et al. 2008; Cooke 2008). A specific situation prevails when the couple is geographically separated following migration of one partner. Partners then spend less time with each other, and the perception of alternatives to the partnership may change, resulting in higher dissolution risks (Caarls et al. 2015; Vidal and Huinink 2019).

In the following, we aim to address the above-mentioned “tied migrant” hypothesis in an exploratory analysis for explaining union dissolution in the course of an international migration. The human capital approach considers migration as an investment that is undertaken if the expected value of benefits exceeds the costs, and the spouse with the greater earning potential will have the greatest influence in the migration decision. Thus, one partner, the lead migrant, initiates a migration, while the other partner, the “tied migrant”, follows, despite potential penalties to his or her career. Given the still-prevalent gender differences in the domestic division of labour, careers, and earnings, “tied migrants” are mostly women (Cooke 2008), even though recent research suggests that rates of “tied migration” are becoming more similar for men and women (Cooke 2013; see also Erlinghagen 2021). Clearly, with increasing women’s employment, occupational characteristics of women may play an increasing role in family migration decisions (Shapira et al. 2019).

Research shows that “tied migrants”, specifically those with higher labour market resources, suffer from career and income losses after relocation (see the literature review in Vidal and Huinink 2019). Anticipating these disadvantages, spouses may refrain from migrating altogether. Research suggests, for instance, that families have a lower probability of moving when the wife is employed and the couple shares egalitarian gender role attitudes (Cooke 2008). Cooke (2013) further shows that this kind of “tied staying” is indeed more common than “tied migration”. Alternatively, spouses may decide that one spouse stays behind (resulting in a transnational relationship) or to somehow deal with this migration-caused asymmetry and migrate subsequently or together.

We argue that it is reasonable to assume that migration decisions will be negotiated between the spouses rather than automatically structured by the labour market characteristics of one spouse or the couple’s established gender roles (Shapira et al. 2019, p. 3). In line with the family micro-economy theory, and depending on the individual resources and the bargaining power of both spouses, couples may unanimously decide in favour of the migration even if their anticipated individual gains differ, accepting (temporary) career and income disadvantages and gender-specific divisions of labour due to overall higher family utilities.

When the migration decision has not been made unanimously, however, we assume a specific pattern of “tied migration” with problematic differences in commitments, diverging motivations for migration between the spouses, and a strain on the spousal relationship, resulting in higher risks of union dissolution (Boyle et al. 2008; Shapira et al. 2019). Thus, by reference to the ways a) the migration decision was made (egalitarian or non-egalitarian) and b) the migration itself was patterned (synchronised or non-synchronised), we suggest a way to identify “tied migrants” “who moved but did not want to” (Cooke 2013, p. 817) and we aim to test the influence of this pattern of “tied migration” on union dissolution.

Accordingly, we hypothesised that those couples in which the driving force behind the migration decision was primarily one spouse (non-egalitarian) would display higher risks of union dissolution following international migration than couples in which the migration decision was made by both spouses (egalitarian) (hypothesis 1a). Furthermore, we predicted that couples who migrated subsequently and remained separated transnationally (non-synchronised) would display higher risks of dissolution than couples who migrated together (synchronised) (hypothesis 1b).

Moreover, we assumed certain differences of the “tied-migrant effect” by gender. Despite large progress in gender equality and widespread egalitarian attitudes, gender inequality patterns are still widespread, both in terms of a gender gap in labour force participation patterns and income levels, and in the division of household labour among couples (Wrohlich 2017). In line with the argument of higher divorce risks in hypergamous marriages, i.e. in those couples in which–against the “traditional” pattern–wives have an educational or occupational advantage (Grow et al. 2017; Schwartz and Han 2014), we assumed that union dissolution risks are higher when the female partner initiated and pushed the migration decision compared to those couples in which the migration decision was made by both partners, and we expected no differences when the male partner was the driving force (hypothesis 2a). Likewise, we assumed that the separation risk increases when the female partner was the first to migrate, but we expected no differences when the male partner migrated first (hypothesis 2b). Given the different contexts of emigration and remigration decisions (see Erlinghagen 2021), we conducted separate analyses for emigrants and remigrants.

3 Empirical Background

Most research on the association between a migration experience and union dissolution focuses on internal migration. One of the most-cited studies is an analysis on internal long-distance migration and short-distance residential mobility of the Austrian Family and Fertility Survey from 1995/96, which included detailed retrospective partnership and residential histories (Boyle et al. 2008). These findings suggest that union dissolution was affected by multiple migrations and residential moves, whereas the first migration did not show any effect and the first residential move even decreased the risk of union dissolution. The authors argued that the power imbalance between the partners may widen as the number of moves made by a couple (to support one spouse’s career) increases, potentially increasing levels of stress and dissatisfaction. In a similar study on internal migration in Russia this finding was confirmed (Muszynska and Kulu 2007). Couples who moved frequently over long distances (within one country) had a significantly higher risk of union dissolution than couples who did not move or who moved only once. The authors referred to the high costs for women in the course of repeated adjustments. A similar study in Great Britain supports the finding that geographically mobile couples are at higher risk of union dissolution, with long-distance internal migration and frequent moves increasing the risk and short-distance residential moves being associated with greater union stability (Shapira et al. 2019).

In research on the effects of international migrations, a main focus is on transnationally separated couples and their union dissolution risks in this specific situation. For instance, Davis and Jennings (2018) found for ever-married emigrant men from Nicaragua that migration and a separation from their spouses, and even more importantly, the duration of the separation increases the dissolution risk. Previously, the widespread migrant narrative of “the spousal desertion” triggered research on Mexican couples by Frank and Wildsmith (2005), who found that shorter separations do not increase divorce risks compared to non-migrants, but longer stays abroad do increase risks. Caarls et al. (2015) focused on different spousal migration patterns in the Ghanaian-European migration context. They showed that divorce risks were higher when couples migrated jointly and when the wife emigrated by herself, whereas risks were similar to those of non-migrant couples when only the male partner migrated or the wife followed later, which was explained with the specific cultural contexts of origin and destination.

Moreover, a couple of studies focused on expatriate families who migrated in the context of job assignments. According to a brief literature overview by McNulty (2015), the familial challenges of international relocation are a main reason for assignment refusal and assignment failure (see also Cole and Nesbeth 2014). Yet she stated that “there is not one academic study yet published on expatriate divorce” (McNulty 2015, p. 107). In her own qualitative study she disentangled the reasons that resulted in divorce or separation, including diverging acculturation patterns (to an expatriate culture) resulting in alienation between the partners. A number of other qualitative studies have shown how both men and women deal with their lives as trailing spouses (e.g. Cangià 2018), and Kõu and Bailey (2014) stressed how joining spouses among highly skilled Indians in Europe are no longer passive movers but active agents in the migration process, challenging the notion of tied or “trailing wives” (see also Kõu et al. 2015). So far, however, there does not seem to be any statistical evidence of dissolution rates in these couples.

This brief research overview indicates that migration and spatial mobility do not per se increase the risk of union dissolution, but that the specific characteristics of the migration experience and the union relationship as well as the broader (cultural) context of the migration and union systems have to be considered to understand the underlying mechanisms. With this exploratory analysis, we aim to contribute to this research field with a specific focus on the migrating couple, the relevance of the spousal migration decision-making process, and migration patterns in the context of emigration from and remigration to Germany.

4 Data and Methods

Our analyses are based on the first two waves of the German Emigration and Remigration Panel Study (GERPS) (Ette et al. 2021). This pooled data set contains 11,897 observations with information about 4928 German emigrants and 6969 remigrants. The data set used for analyses includes only completed online interviews, and respondents who emigrated or remigrated prior to 2017 are excluded because we wanted to concentrate on recently migrated individuals. Furthermore, we excluded respondents with missing information regarding their sex, age, or the dependent variable. In addition, the sample is restricted to migrants with heterosexual relationships that started prior to migration and survived the migration event. In wave 1, 5752 GERPS participants retrospectively reported that they were in a relationship 3 months prior to migration (emigrants: 2334; remigrants: 3418). However, 105 of those also reported that they had separated already prior migration (emigrants: 40; remigrants: 65). Under these conditions, data from 2292 emigrants and 3352 remigrants remained in our data set.

Until now, there has been only two waves of interviews after migration. However, via retrospective questions, it is possible to get information about the relationship status of the emigrants and remigrants 3 month before the migration event (wave 0). Figure 10.1 shows the resulting time-dimensions in GERPS. They include one interval prior to migration (between wave 0 and 1) and two intervals after migration (between migration and wave 1, and between wave 1 and wave 2, respectively). Due to the chosen restriction of our data set to people with partnerships that survived the migration event, we assess separation in the two time intervals after migration. These are the basis for our dependent variable, which takes on the value 1 if a separation occurred in one of these two time intervals and 0 if respondents are in a permanent partnership between wave 0 and wave 2. Consequently, union dissolution is a binary measure. Therefore, binary logistic regression is the appropriate analyses strategy for our purpose (cf. Wooldridge 2009).

Fig. 10.1
A flow diagram demonstrates the resulting time dimensions in G E R P S. It illustrates the migration event for wave 0 during 2017 and 2018 prior to migration, and waves 1 and 2, after migration in February and November 2019.

Timeline of interview waves and construction of time intervals in GERPS. (Source: authors’ presentation)

Full size image

With regard to our hypotheses formulated at the outset, the two main explanatory variables in our analyses are “migration decision” and “timing of migration”. These variables are based on the following questions:

Migration decision: “If you remember your migration decision: Who has been the driving force? You or your partner?” The participants were given the following response categories: (1) “my partner”, (2) “I myself”, (3) “Both in equal shares”, and (4) “don’t know”.

Timing of migration: “What was the timing of your migration like? Which one of you migrated first or did you migrate together?” The participants could choose among the following categories: (1) “My partner already lived there at the time we met”, (2) “My partner migrated first”, (3) “My partner migrated after me”, (4) “We migrated at the same time”, (5) “My partner is still living in Germany” [emigrants only] / “…in the country I have lived before” [remigrants only], and (6) “My partner lives in another country”.

Table 10.1 shows that these two variables, although related, measure different phenomena. Although 53.3% of the respondents who made an egalitarian decision to migrate with their partner also moved at the same time, 41.1% moved subsequently and 5.5% still lived in different countries (transnational relationship). Similarly, when one partner was the driving force, the largest percentage of spouses fell into subsequent migration (47.2%) and 22.5% of respondents still lived in another country than their partners. However, 30.3% moved at the same time even though the decision was primarily shaped by one partner.

Table 10.1 Relationship between the driving force of migration and the actual course of migration (shares in %)
Full size table

Additional to these explanatory variables, further control variables were considered. These included current migration status (emigrant vs. remigrant), gender, age, prior migration background (foreign roots vs. no foreign roots), main activity (economically active vs. not economically active), educational attainment (primary and secondary education vs. post-secondary/short-cycle tertiary education vs. bachelor’s or higher), whether there are children living in the household, and the spatial distance between Germany and the (former) host country (outside Europe vs. within Europe, but not neighbouring vs. within Europe and neighbouring). Descriptive statistics of variables used in this analysis are provided in Table 10.2.

Table 10.2 Distribution of variables by migration status (percentages)
Full size table

5 Results

In model 1 in Table 10.3, we pooled the data of remigrants and emigrants. Results suggested a lower propensity of union dissolution for emigrants compared to remigrants. Moreover, Table 10.3 shows that female remigrants had a higher risk of separation compared to males but that there was no evidence for any gender effect on emigrants’ propensity for separation (model 2). These findings were stable even after controlling for certain decision patterns and timing patterns of migration in our extended model 3. With regard to the other control variables, younger remigrants under the age of 30 years showed increased separation risks. In addition, relationship breakup was more likely for remigrants with a medium educational degree compared to remigrants with a bachelor’s degree or higher. However, the presence of children under the age of 17 lowered the separation risk for remigrants. The same held true for emigrants but only if children were younger than 6 years. Furthermore, non-egalitarian migration decisions were associated with an increased risk of relationship dissolution for both emigrants and remigrants (model 3), which corroborates hypothesis 1a. In addition and in line with hypothesis 1b, for both emigrating and remigrating couples a subsequent timing of migration or a transnational relationship further increased separation risk compared to couples in which both partners migrated at the same time.

Table 10.3 Logit regression on separation by gender and migration status
Full size table

The results presented in Table 10.3 point to the general importance of couples’ migration decision process as well as of couples’ migration timing pattern as potential union destabilising factors. However, we had predicted that in addition to such overall correlations there should be gender differences with regard to the correlation between decision and timing patterns of couples’ migration on the one hand and separation risks on the other hand (hypotheses 2a and 2b). Therefore, we analysed whether decision and timing patterns are of the same relevance for both male and female migrants’ relationship stability (Table 10.4). First, we estimated separate models, each including either information on the migration decision pattern or information on the migration timing pattern of couples (models 1 and 2). Then, we estimated a full model that included both types of information (model 3).

Table 10.4 Logit regression on separation by gender differences in migration decision or timing pattern
Full size table

As shown in Table 10.4, it makes no difference whether the male or the female partner was the driving force of couples’ migration decision. In any case, non-egalitarian decisions were associated with higher separation risks after migration. This result conflicts with hypothesis 2a, which predicted no increased separation risks if the male partner led the migration decision. However, hypothesis 2b was partly corroborated. At least in the case of emigration, the data showed higher separation risks if women moved in advance of their partners, but no such effect (as expected) if men migrated first. But when it comes to remigration, there are no such gender differences, which does not support our related hypothesis. In addition, there are also no gender differences in transnational couples: Regardless of whether the male or the female partner still lives in Germany (emigrants) or abroad (remigrants), couples’ separation risks are significantly increased.

6 Conclusion

This chapter has focused on partnership dissolution in the course of an international migration. Rather than studying mobility outcomes following the dissolution of a partnership, we asked under which conditions international migration increases the risk of separation among couples. Our sample included individuals in both marital and non-marital relationships who had recently left Germany and moved to another country or who had recently remigrated to Germany. We were particularly interested in the so-called “tied migrant” phenomenon and its relevance for union dissolution. According to this concept, partners may have diverging motivations for migration. Partners who expect to suffer from career and income losses after relocation may be less inclined to migrate and only follow (later) as “tied migrants” (Cooke 2013). These “tied” persons in a family do not directly benefit from the migration themselves, benefiting only in terms of family gains, which might put strain on the migration experience and the spousal relationship (Shapira et al. 2019). We defined couples with “tied migrants” as those cases in which the migration decision was made in a non-egalitarian manner and in which the migration itself did not take place simultaneously. We argued that this situation included additional strain on the spousal relationship, resulting in higher risk of union dissolution: Accordingly, we hypothesised that couples in which the driving force behind the migration decision was primarily only one partner would display higher risks of union dissolution following international migration than couples in which the migration decision was made by both spouses. Furthermore, we predicted that couples who migrate subsequently and/or remain separated transnationally would display higher risks of dissolution than couples who migrate together.

Moreover, we assumed certain differences in the “tied-migrant effect” by gender. Despite large progress in gender equality and widespread egalitarian attitudes, gender inequality patterns are still widespread. In line with the argument of higher divorce risks in hypergamous marriages, we assumed union dissolution risks to be higher when the female partner was the driving force of migration, compared to those couples in which the migration decision was made by both partners. When the male partner initiated and pushed the migration, we expected no difference. Likewise, we assumed that separation risk would increase when the female partner was the first to migrate, but we expected no differences when the male partner migrated first.

Referring to data from the first two waves of GERPS, we found that emigration is accompanied by a lower separation risk than remigration. Whereas previous research has started to differentiate between varying lengths of spousal separation, showing that union dissolution outcomes are more likely in case of longer separations, we can add that emigration and remigration contexts differ as well, with the latter being more prone to result in union dissolution. These findings underscore that research on family dynamics must differentiate carefully between “directions of migration” and increase efforts to better understand the underlying mechanisms (see also Erlinghagen 2021).

The main finding of the presented analyses, however, is that non-egalitarian, non-synchronised migration patterns are important predictors of union dissolution for male and female emigrants and remigrants. In cases of “tied migrants” the risk of a dissolution is significantly higher. Thus, our results mainly support our hypotheses and stress the importance of the pre-migration spousal negotiation process as well as the migration patterns (timing of migration) for the understanding of changing family dynamics in migration contexts. In addition, the data showed that it makes no difference whether the male or the female partner was the driving force or the person who moved first. Although women are still more often “tied movers” than men, we can conclude that tied moving is significantly correlated with a higher risk of partnership dissolution for both gender groups. This holds particularly true for non-egalitarian migration decisions but to some extent also for non-synchronised migration patterns. However, we find one interesting gender difference: Only among emigrating couples in which the man moved first we do not find increased dissolution risks compared to couples who emigrated together. This effect even remains when we control for the decision-making process. This finding may hint at a certain persistence of traditional gender norms in shaping the timing patterns of emigrating couples and may indicate that traditional gender norms help couples to cope with uncertainty in the emigration process (cf. Erlinghagen 2020). Women, even if they are “tied movers”, may not feel additional strains on their relationship.

Beyond these rather general statements we can only speculate about the reasons for the observed patterns. Since we do not have information on pre-migration partnership quality or prior spousal conflicts, we cannot rule out that the couples’ migration decision processes as well as their migration patterns are consequences of relationship strain rather than its causes. If this is the case, it is not the dissolution which follows the migration. Rather, changes in family dynamics (here: lower partnership quality and spousal conflicts) result in specific (r)emigration mobility patterns and subsequent union dissolution.

Although these exploratory analyses provide some interesting new findings, further research is needed to better understand the determinants of partnership dissolution in the context of migration. First, future analyses should include more information on the characteristics of the two partners (e.g. their human capital resources, pre- and post-migration economic situation, gender role attitudes) to test whether the observed pattern applies to all couples with “tied migrants” alike or whether risks of dissolution vary by the partner’s resources, attitudes, and interspousal patterns of homogamy and heterogamy. Second, a comparison between emigrants, remigrants, and non-mobile couples (“stayers”) would help to address the issue of selectivity, namely whether migrants have higher separation risks per se due to certain personality traits and individual characteristics that have an impact on the propensity to migrate and on the propensity to split up at the same time (cf. Shapira et al. 2019). Further research may then address possible consequences of union dissolution for the two partners (in terms of labour market outcome, subjective well-being, new partnerships, subsequent mobility, etc.), and shed further light on the short-term and long-term impact of an international migration on family lives and individual life courses.

Given the little existing scientific knowledge on the complex interplay of family dynamics and international migration processes, the data of GERPS offers a new, great opportunity to improve our understanding of family- and partnership-related issues. The analytical potential of GERPS will further increase with the subsequent third and fourth waves that will become available in 2020/2021. This is a major opportunity because migration-related panel studies are still extraordinarily rare. Both the future panel data and the option to consider comparisons with the non-migrant population will allow a rich contribution to the research field of determinants and consequences of family dynamics in the course of migration.