1 Introduction

Migration provides people with new opportunities. For example, migration can be beneficial for an individual’s career by bringing about income gains (Clemens 2013; Gibson and McKenzie 2012; Gibson et al. 2018; Hendricks and Schoellman 2018) However, migration also brings about costs. By definition, moving across borders implies financial costs. These include costs shortly before or during the migration event, such as costs for transportation, visa fees, or new furniture. Furthermore, financial costs can arise after migration (e.g. travel costs to visit friends and family at home). However, migration might not only be costly in terms of economic capital. For example, human capital could be devaluated as it might not be applicable to the labour market in the destination country (Chiswick 1978; Chiswick and Miller 2007). In this context, Chiswick and Miller (2007) talk about “less-than-perfect transferability of skills acquired on the job or through formal schooling” (p. 2) to underline that this devaluation not only refers to language skills. Furthermore, social networks might be broken, leading to a loss of social capital (Lesage and Ha 2012; Wahba and Zenou 2012). In particular, accessibility of social capital is crucial for its mobilisation (Lin 1999), which might be hindered after international migration.

Social capital understood as interpersonal ties (Granovetter 1973) involves both instrumental and emotional support. With international migration and, thus, increased geographical distance, the loss of instrumental support seems plausible. However, emotional support is also important to analyse as O’Flaherty et al. (2007, p. 819) point out that “migrants are often in the situation where many of their most emotionally significant relationships are conducted internationally” (p. 819). To what extent and by what means migration affects emotional support appears unclear as only a few qualitative studies exist (see e.g. Baldassar 2007; Guo et al. 2009). These studies focus on family relationships. Migration, however, also implies friendships across borders. There is even less scientific knowledge concerning these friendships than transnational family relationships. This lack of understanding of a potentially impactful social experience provides the starting point for this chapter, which asks three research questions:

  1. (a)

    How does the quality of relationships with friends in Germany differ from the quality of other relationships after migration?

  2. (b)

    How is friendship quality after migration related to socio-demographic or socio-economic factors?

  3. (c)

    Is there a link between friendship quality and the subjective well-being of emigrants?

The first question can be seen as the starting point and can be answered using descriptive statistics thus providing the basis of this analysis. The second question builds on the first. To address the second question, determinants of friendship quality are identified by applying multivariate regressions. Finally, the third question looks at correlations between friendship quality and different aspects of subjective well-being. Subjective well-being can be conceptualised as an important outcome variable of the migration event (Bartram 2013; Baykara-Krumme and Platt 2018; Erlinghagen 2011; Guedes Auditor and Erlinghagen 2021; Safi 2010). With this in mind, by focussing on friendship quality the third research question takes a step towards analysing differences in subjective well-being between emigrants. In doing so, this article provides new insights concerning individual psychosocial consequences of emigration from industrialised countries as data from German emigrants is used.

The remainder of this chapter is structured as follows. First, I outline some theoretical considerations concerning determinants of contact frequency and its link with subjective well-being. This is followed by a short overview of (both qualitative and quantitative) findings concerning these factors. Then, I describe the methodology and outline the sets of variables used to answer the three research questions. Next, the research questions are answered in sequence. Finally, I discuss the findings and draw a conclusion.

2 Theory

2.1 Determinants of Friendship Quality

Friendships can be defined as voluntary, informal and intimate social relationships (Bowlby 2011). Friendship quality, in turn, can be conceptualised as the quality of relationships with friends. Relationship quality is usually measured by contact frequency, whereby scholars assume that ties are closer the higher the contact frequency (Bauernschuster et al. 2010; Blömers and Letschert 2011) This contact frequency can encompass different modes of contact: personal visits, calls, messages, etc. Although the German Emigration and Remigration Panel Study (GERPS) includes all modes of contact in their frequency measures, some scholars focus on one aspect, for example personal visits. Below, I summarise theoretical considerations concerning different modes of contact frequency.

O’Flaherty et al. (2007) assume that economic integration (wealth, income, economic status, education, etc.) has a positive impact on visits to one’s own home country as financial resources are crucial for travel opportunities. Still, some aspects of economic integration that are time-consuming appear to work in the opposite direction: someone who has a full-time job and thus a limited number of days of annual leave might be less likely to return home frequently than a student or a pensioner who has fewer obligations in the host country. Also, higher economic integration increases the attractiveness of staying in the host country (Portes et al. 1999). Concerning other forms of contact (letters, information and communication technologies or ICTs, etc.), the direction of the effect is not so clear either. Education might, in general, be correlated with higher ICT handling skills and thus foster contact using this channel. However, given that the overall level of education in the GERPS sample is already very high, education might rather correlate with employment obligations and thus have the opposite effect. Furthermore, it might foster the speed of overall integration (Guarnizo et al. 2003) and thus negatively impact contact frequency. Social integration implies closer ties to the host country’s society and thus is assumed to have a negative impact on frequency of contact with friends in the home country. Furthermore, social integration is likely to negatively affect intentions to remigrate, and not intending to remigrate is assumed to decrease contact with friends in the home country.

Individual characteristics that have been considered include gender, age, and foreign roots. O’Flaherty et al. (2007) find that women are more likely than men to visit home. They argue that this finding might reflect expectations concerning gender roles: women are traditionally expected to do caregiving and provide emotional support but men focus on political and economic concerns. In line with this argumentation, Kaasa and Parts (2008) argue that it is easier for women to find emotional support, for example when depressed. This would imply that they not only invest but also receive more in terms of social support. Age might increase contact frequency as older people are usually more attached to their home country (Iarmolenko et al. 2016). Also, the nature of friendship changes while aging: Fox et al. (1985) state that, with age, men develop more concern and thoughtfulness regarding friends and women become more tolerant and less confrontational. Different behaviour or attitudes towards friends might also impact contact frequencies in friendships. Additionally, foreign roots might play a role in defining contact behaviour. In this context, ‘foreign roots’ refers to having a migration background in Germany and not in the host country as German emigrants by definition have a migration background if they have moved abroad. In order to prevent confusion, I use the term foreign roots. For example, Iarmolenko et al. (2016) hypothesise that differences in home visits among ethnic Germans, Russian Jews, and Turkish people living in Germany depend on distinct exit and entry conditions. Consequently, as resources are limited, these differences might in turn affect home visits in Germany (and thus contact frequency with friends living there).

Distance to the home and time in the host country are assumed to affect contact frequency. In particular, I hypothesise that contact frequency decreases with distance. This is for two reasons: first, travel costs increase with distance, making home visits more expensive. Second, a longer distance is associated with a larger time difference, which narrows possible time slots for communication and thus complicates transnational contact (Ryan et al. 2015). Connections with the home country decrease with time in the host country, and this holds especially true for friends (Ryan et al. 2015). Thus, length of stay is assumed to decrease contact frequency.

Number of friends is assumed to play a crucial role as having a large number of friends simply provides more opportunities for contact frequency. Furthermore, partnership, marital status, and household size (i.e. whether there are children in the household) might be important. Ryan et al. (2015) argue that family obligations negatively impact the likelihood of home visits, one possible mode of contact. Such obligations are higher for people with a partner and/or children and with the presence of the respective person(s) in the household. Furthermore, family obligations might affect network size, which is consistent with the finding that married persons have fewer informal networks (Kaasa and Parts 2008).

2.2 Contact, Friendship, and Subjective Well-Being

Prilleltensky (2008) analyses different risk and protective factors of migrants’ subjective well-being. Friendship is categorised as being protective and thus assumed to positively affect subjective well-being. But what exactly is the link between these two phenomena? Lee and Ishii-Kuntz (1987) argue that individuals choose their friends and at the same time are chosen as friends. The latter demonstrates to the individuals that they possess positive attitudes or qualities that are valued. This is assumed to have a positive impact on subjective well-being (Lee and Ishii-Kuntz 1987). Furthermore, friendship is linked to subjective well-being via support and disclosure (Cuadros and Berger 2016) and mediated via loneliness–a subjective feeling of social integration (Liang et al. 1980). The latter motivates the inclusion of different loneliness measures to assess the link between friendship and migrants’ subjective well-being. Furthermore, contact with friends in the home country might correlate not only with overall life satisfaction, but also with satisfaction concerning different life domains.

3 Literature Review

In general, literature on cross-border relationships of emigrants mainly focusses on Asia and Latin America (Apitzsch 2014) and economically disadvantaged groups (e.g. Berítez 2012; Madianou and Miller 2011). If Germany is the subject of study, literature usually approaches it as a destination country (Baykara-Krumme 2014; Pusch 2013). This means that the behaviour of migrants in Germany, not German emigrants, is analysed. Two exceptions exist. First, Décieux and Mörchen (2021) analysed differences regarding the number of close friends between German emigrants and non-mobiles, including both cross-border and within-country friendships as they assessed the overall number of friendships. Second, Mau and Mewes (2007) studied transnational social relationships but focused only on non-mobile individuals living in Germany and their contact with people living abroad. Also, not much research has been published with respect to other industrialised countries. Still, it was found that 98 per cent of Irish migrants in Melbourne have contact with their family in Ireland (O’Connor 2010). The same holds for 93 per cent of migrants of English-speaking background and 91 per cent of migrants of non-English-speaking background in Australia (Ang et al. 2002) and 95.4 per cent of migrants from different countries in Vancouver, Canada (Hiebert 2003).

3.1 Determinants of Contact

Literature on the determinants of contact with people in the home country has focused primarily on contact with relatives and has mainly examined either home visits or contact using information and communication technologies (ICTs). Furthermore, research on transnational relationships of relatively affluent migrants is limited (O’Flaherty et al. 2007). However, some exceptions exist: Major between-group differences for migrants in Australia were found concerning migrants’ visits home (O’Flaherty et al. 2007). In particular, they report differences with respect to visa category, country of origin, sex, age, relative life satisfaction in Australia and aspirations to gain Australian citizenship, English language skills, income, housing situation, home ownership, education, financial aid from overseas, and financial aid from within Australia. Also looking at migrants’ visits home, Iarmolenko et al. (2016) assessed different determinants of transnational activity of female migrants in Germany. They found that determinants vary by ethnic background, for example the financial situation is significant for Turkish immigrants but not so for ethnic Germans and Russian Jews. Determinants that are statistically significant (for some female migrant groups) but not considered by O’Flaherty et al. (2007) are length of stay, discrimination, thoughts about returning to the home country, as well as several acculturation, identification, and social network measures. Analysing contact behaviour of highly skilled American and French migrants in London, Ryan et al. (2015) found that geographical distance matters: French people could visit their home country more frequently than Americans. Also, the authors identified competing family obligations as being crucial for determining home visits.

Given that transnational contact became easier with the surge of ICTs (Bacigalupe and Cámara 2012; Pajnik and Bajt 2012) one might assume that economic status (i.e. income, home ownership, financial aid, etc.) is less crucial for ICT-based contact. However, the availability of these new technologies differs (Wilding 2006). Bryceson and Vuorela (2002) showed that indeed income and material assets are crucial determinants. Analysing phone and e-mail contacts of migrants in the Netherlands, Schans (2009) identified similar influencing factors as those for home visits. Furthermore, she added age at migration to the list of determinants.

3.2 Cross-Border Contacts and Subjective Well-Being

Even though transnational ties are crucial for the emotional well-being of every migrant (Ryan et al. 2015), literature on the link between cross-border contacts and subjective well-being has focused mainly on migrant parents and left-behind children (e.g. Dito et al. 2017). One exception is O’Flaherty et al. (2007), who found a significant effect of relative life satisfaction in Australia on the odds of home visits.

More generally and without focusing on migration, there appears to be a relationship between friendship and subjective well-being (Heady et al. 1991). Among children, Gauze et al. (1996) found that friendship and subjective well-being were correlated. More specifically, Cuadros and Berger (2016) found that two aspects of friendship (support and disclosure) affect subjective well-being of female children. Concerning elderly people, friendship significantly increases morale and decreases loneliness, with effects being stronger than those associated with family (Lee and Ishii-Kuntz 1987).

4 Data and Methods

In order to answer the three research questions, I present analyses of data from the first wave of GERPS. GERPS assesses consequences of international migration of German citizens and is based on a random sample drawn from local population registers (Ette et al. 2021). It covers both German emigrants and remigrants but for this analysis only emigrants are considered. Additionally, only respondents who indicated both their sex and age and who emigrated from Germany in either 2017 or 2018 were included. The latter facilitates focusing the analysis on short-term effects of migration. In total, 3536 observations remained.

Friendship quality is measured using contact frequency with friends who still live in Germany. This approach is common in literature (see e.g. Bauernschuster et al. 2010; Blömers and Letschert 2011). Contact frequency is measured not only with friends, but also with seven groups of relatives, allowing for comparisons between types of relationships. In particular, GERPS respondents are asked to indicate their contact frequency with (a) partner/spouse, (b) parents/parent-in-law, (c) siblings, (d) children, (e) grandchildren, (f) grandparents, (g) other relatives (e.g. aunts, uncles, cousins), and (h) close friends. If respondents were in contact with more than one person in one category (e.g. siblings, friends), they were asked to answer the question based on the person with whom contact is most frequent. The resulting measures include different modes of contact (e.g. visits, phone calls, messages). Contact frequency is measured on a four-point scale and was rescaled so that higher values indicate more frequent contact:

How often are you in contact with the following people in Germany?

  • Daily

  • At least once a week

  • At least once a month

  • Less than once a month

Thus, contact frequency is an ordered categorical variable. Often with categorical variables, ordered logit models are estimated. These models rest on the proportional odds assumption that does not hold in this sample. This is not surprising as scholars argue that it almost always needs to be rejected, especially with many explanatory variables (Brant 1990), the inclusion of continuous explanatory variables, and a large sample (Allison 2012; Brant 1990; Clogg and Shihadeh 1994; O’Connell 2006). Consequently, I estimated generalised ordered logit, which allows the coefficients to differ across categories of the dependent variable. For each variable, the auto-fit option–in an iterative process–tests whether its coefficients actually differ across categories and Stata estimates only one coefficient if this is not the case (Williams 2006).

Table 13.1 describes the coding and underlying classification of each explanatory variable used to assess the second research question. It also shows summary statistics.

Table 13.1 Descriptions, coding, and summary statistics of explanatory variables
Full size table

In order to address the third research question, different measures of subjective well-being were taken into account:

  1. (1)

    Individual items of the GSOEP loneliness scale

    • Miss the company of others

    • Feel left out

    • Feel isolated

  2. (2)

    GSOEP loneliness scale overall

  3. (3)

    General life satisfaction

  4. (4)

    Perceived change in the quality of different life domains

The inclusion of (1) and (2) in this list is based on the findings by Liang et al. (1980) concerning the mediating effects of loneliness on subjective well-being. In the present study, first, participants’ responses to the individual items of the GSOEP loneliness scale, including “miss the company of others,” “feel left out,” and “feel isolated” were analysed. These items cover different dimensions of loneliness and are measured on a five-point scale, ranging from 1 (never) to 5 (very often). Second, the three-item GSOEP loneliness scale, which was developed building on the 20-item UCLA loneliness scale (Luhmann and Hawkley 2016) is included. It is based on the three single items in (1). Scholars have shown its strong correlation with the original scale (Hughes et al. 2004) as well as the validity of the German version (Hawkley et al. 2016). The overall score is computed as the sum of the individual items. Third, general life satisfaction as a measure of subjective well-being is considered. Respondents were asked about their overall satisfaction level on a scale from 0 to 10. Fourth, changes in the quality of different aspects of life are considered. These include changes in family life, the situation with friends, health, and neighbours. Respondents were asked to compare the situation before migration to the present situation and indicate whether it got better or worse. In particular, respondents could answer on a five-point scale ranging from 1 (much better than in Germany) to 5 (much worse than in Germany).

As with contact frequency, these measures are categorical variables. As explained above, ordered logit models rest on the proportional odds assumption, which in most cases does not hold. However, the measures of subjective well-being consist of 5–15 categories, which would lead to a very high number of equations to be estimated as well as a very low number of observations included in each estimation equation. When pooling these categories, information would be lost. Furthermore, the third research question is not so much about how exactly different determinants affect subjective well-being but rather whether or not there is a link between friendship quality and different dimensions of subjective well-being. Thus, I stick to simple ordered logit regression and interpret estimation coefficients as “average effects” of contact frequency on subjective well-being (keeping in mind that this effect might differ across the categories of subjective well-being and that causality cannot be established).

5 Results

5.1 Patterns of Contact Frequency

Table 13.2 shows emigrants’ contact frequencies with different relatives as well as friends living in Germany. Overall, frequency of contact with friends fell in the middle of this frequency scale, with most of the respondents having contact with their friends on a weekly (41.1%) or monthly (37.7%) basis. A total of 10.8 and 10.4 per cent accrue to the two extreme response options, less than once a month and daily, respectively. Contact was most frequent with the partner (if he or she stayed in Germany), followed by children and parents (including parents-in-law). Least frequent contact was observed to be with other relatives, followed by grandparents and grandchildren. The differences in the numbers of observations are also noteworthy, which stem from the fact that not all respondents had all different types of relatives and/or a partner living in Germany.

Table 13.2 Contact frequency by types of relatives and friends (in per cent)
Full size table

Descriptive statistics for contact frequency with friends in Germany also show distinct patterns by socio-demographic and socio-economic variables (Table 13.3). Female migrants had more contact with friends in Germany than male migrants did. In particular, female migrants were more likely to have daily or weekly contact. Contact frequency decreased with age: the lowest contact category (less than once a month) increased, but both daily and at least once a week decreased with age. Regarding foreign roots, only minor differences in contact frequency were observed: People with foreign roots were somewhat more likely to give extreme answers (daily or less than once a month). Similarly, small differences were found concerning the length of stay: Contact frequency appears to decrease with the length of stay. Given that the maximum length of stay in this data set is 25.1 months, even short-term differences could be identified. Contact frequency was slightly higher for emigrants who reported thinking about returning to Germany. The same held for emigrants who live in Europe–which can be seen as a proxy for distance. However, contact frequency was lower for people who reported being employed or holding a bachelor’s or higher degree. Contact frequency increased with the number of friends between one and 20. Having more than 20 friends, however, did not further increase contact frequency. Comparing respondents with more than 20 friends with those having six to 10 or 11 to 20 friends, it is furthermore noteworthy that people with more than 20 friends were more likely to give extreme answers, i.e. have contact daily or less than once a month. Furthermore, respondents who had a partner showed lower levels of contact frequency with friends than those who did not have a partner. A similar picture can be drawn when looking at household constellations: Respondents in single-person households had more contact with friends than did people living with their partner but without children, who in turn have more contact than people living with their partner and children.

Table 13.3 Contact frequency with friends by explanatory variables (%)
Full size table

5.2 Estimation Results: Determinants of Contact Frequency

Generalised ordered logit regression draws a detailed picture of how different factors affect contact frequency (Table 13.4). Its coefficients need to be interpreted as follows (cf. Williams 2006): A positive (and statistically significant) sign implies that a higher value of the explanatory variable increases the likelihood that the respondent is in a higher category of contact frequency than the present one. In contrast, the probability of being in the present or lower category of contact frequency increases with higher values of the explanatory variable if the estimated coefficient is negative. For several explanatory variables (foreign roots, length of stay, intention to remigrate, Europe, main activity, and household constellation), the proportional odds assumption is not violated (i.e. coefficients are the same throughout the categories of the dependent variable). Thus, following Williams (2016), only one coefficient is shown in the first row, which can be interpreted as a coefficient from estimating a simple ordered logit model. As for the other variables, the effect differs throughout the categories of the dependent variable and thus different coefficients are reported.

Table 13.4 Estimation results for contact frequency applying a generalised ordered logistic model using the auto-fit option
Full size table

Being male increases the likelihood of being in the present or a lower category of contact frequency, but this effect decreases with increasing categories of contact frequency. Similarly, age has a negative impact on contact frequency, but the effect is strongest for the highest category of the dependent variable, followed by the lowest category. For the next five variables, the proportional odds assumption holds and thus the effect is constant across the categories of the dependent variables. Although foreign roots and intention to remigrate appear to not be statistically significant, significant effects can be found for length of stay, Europe, and main activity. The longer the length of stay in the host country, the less frequently emigrants were in contact with friends in Germany. This effect is noteworthy as–due to the recentness of the study plus the fact that only respondents who left Germany in either 2017 or 2018 were included–the overall level of length of stay is still relatively low (see descriptive statistics in Table 13.1). Staying in Europe was associated with more frequent contact with friends and being economically active (employed, self-employed, or civil servant) and being in training were associated with decreased contact compared to not being active (unemployed, not employed, or retired). Holding a bachelor’s or higher degree was associated with decreased contact frequency. This effect is stronger for higher categories of the dependent variable (and it is not statistically significant for the lowest category). The number of friends had a positive impact on contact frequency, which is strongest for the lowest and highest category and weaker in the second column of the table (less than once a month and monthly vs weekly and daily). In contrast, having a partner (both married and unmarried) was negatively associated with contact frequency, with the effect being stronger for higher categories of the dependent variable (it even ceases to be statistically significant for the lowest category). Last, not living in a single-person household was associated negatively with contact frequency.

5.3 Contact Frequency and Well-Being

Table 13.5 shows estimation coefficients of contact frequency on different indicators of subjective well-being. The major finding is that a link between contact frequency and different indicators of subjective well-being exists. In particular, two individual items of the GSOEP loneliness scale, the composite GSOEP loneliness scale, general life satisfaction, and two indicators comparing changes in different life domains before and after migration (“situation has worsened with respect to…”) show statistically significant estimation coefficients. Concerning the GSOEP loneliness scale indicators, statistically significant negative coefficients with respect to “feel left out” and “feel isolated” as well as the composite indicator were found. These variables were coded such that higher values indicate stronger feelings of loneliness. Thus, a negative correlation indicated that higher contact frequency correlated with weaker feelings of loneliness. There were also statistically significant effects concerning general life satisfaction: more contact with friends corresponded to higher levels of general life satisfaction. Additionally, changes in two of the four analysed life domains show statistically significant estimation coefficients: a worsening of the situation with respect to friends and neighbours. Low values of these variables indicate that the situation now is much better than it was in Germany (improvement) and high values indicate that the situation now is much worse than it was in Germany (worsening). Both changes in the area of friends and neighbours have positive signs, implying that more contact with friends in Germany corresponded with a worsening of the situation. In other words, people who still have strong ties to their friends in Germany reported feeling that the situation concerning friends and neighbours after migration was worse than it was in Germany.

Table 13.5 Regression coefficients for contact frequency with friends on different subjective indicators of well-being estimating ordered logistic models
Full size table

6 Discussion and Conclusion

At least in the short term, out of sight does not mean out of mind: German emigrants maintain contact with different types of relatives and friends in Germany. In particular, contacts with friends fall in the middle of this frequency scale: It is less frequent than contact with the partner, children, and parents, but more frequent than contact with siblings, grandchildren, grandparents, and other relatives who remained in Germany.

Several determinants of this contact frequency with friends could be identified. In line with the theoretical considerations outlined earlier, staying within Europe and a higher overall number of friends was associated with increased contact with friends in Germany. The Europe dummy is used as a proxy for distance to Germany and the finding can be explained as follows: Contact decreases with distance as travel costs and time difference increase. Having more friends simply provides more opportunities for contact and thus increases contact frequency. In contrast, being male, increasing age, length of stay, holding a bachelor’s or higher degree, having a major activity, being in a relationship, and not living in a single-person household were associated with decreases in emigrants’ contacts with friends in Germany. Except for age, these effects are also consistent with the theoretical considerations outlined earlier. Different models regarding gender roles, associating women with being more involved with emotional support than men (O’Flaherty et al. 2007), might explain why being male is associated with lower frequency of contact with friends. Both length of stay and education are likely to foster integration, making contact with people in the host country relatively more attractive. Having a major activity, being in a relationship, and not living in a single household appear to influence contact frequency similarly. They not only increase the relative attractiveness of the host country, but are also time-consuming and imply more obligations in the host country. In contrast, previous literature suggested that contact frequency might increase with age as older people are more attached to their home country (Iarmolenko et al. 2016) and friendship was assumed to strengthen as people grow older (Fox et al. 1985). However, in the present analyses I found that contacts with friends decreased with age. A potential explanation is that younger people are more affluent with ICT-based communication and that this type of communication represents an important way of staying in touch. Furthermore, age might correlate with obligations other than those covered: having a major activity, partnership status, and household composition. Assuming that all types of obligations in the host country are associated with decreased contact frequency with friends in the home country, this might further explain the negative effect of age.

Correlations between different indicators of subjective well-being and contact frequencies exist. More contacts with friends implied weaker feelings of loneliness and higher life satisfaction. However, not all individual items of the GSOEP loneliness score correlated with emigrants’ contact frequency with friends in Germany. This emphasises that different aspects of loneliness possibly exist, and the different aspects might have different determinants and consequences. This finding therefore calls for further research. The fact that a perceived worsening in different life areas (compared to the situation in Germany) corresponds to more contacts with friends in Germany appears plausible. Respondents might try to compensate for the lack of friends or contacts with neighbours in the new country of residence with friends in Germany. However, it poses the question of the direction of this relationship. In this chapter, I assumed contact frequency to affect subjective well-being and thus estimated coefficients using subjective well-being as the dependent variable. However, to some extent it might also be the other way around. In particular, the questions are the following: Do people stay in touch because they perceive different aspects of their new lives as being worse than in Germany? Are they unhappy and seeking social support from friends left behind in Germany? Or is it the other way around and more contacts remind people of the positive aspects of their life in Germany and thus lead to glorification? With the next wave of data becoming available in 2020, further investigations regarding the causal relationship between contact with friends in Germany and emigrants’ integration into host country societies will be possible by estimating a model assessing causality using three survey waves (Heady et al. 1991).

As stated earlier, “migrants are often in the situation where many of their most emotionally significant relationships are conducted internationally” (O’Flaherty et al. 2007, p. 819). Thus, analysing these relationships is an important task when assessing migrants’ social capital. In this context, this article helps understand the phenomenon of cross-border friendship and takes a first step towards analysing the link between friendship and migrants’ subjective well-being.