Gender and Social Contacts: Labour Market Entry Among Refugee and Family Reunion Migrants in Sweden

Gender Differences and Social Contacts

A substantial literature reveals gender differences in access and returns to social contacts in the labour market (Lin 2000; Ridgeway 2011). Compositional differences in men and women’s networks are largely accounted for by structural factors and gender homophily (Hofstra et al. 2017; Kalmijn 2012; McPherson, Smith-Lovin & Cook 2001). Yet, studies reveal that men’s networks are often more gender-segregated than women’s (Bilecen & Seibel 2021; Huffman & Torres 2002). While men’s social networks tend to include colleagues and business partners, women’s networks more often centre around kin and community-based contacts (Kalmijn 2012; Smith 2000). Research also reveals that male faculty members appear to be more effective in using friends and counsellors to generate resourceful ties in male-dominated university settings than their female colleagues (van Emmerik 2006).

Differences in network composition are important, since they can impact the kinds of jobs that men and women obtain through their networks. Bourdieu (1985) used the term social capital to describe the resources or profits that can be accessed through membership or participation in groups, such as families, parties or associations. For Bourdieu, social capital is the product of time and energy directed towards a series of material and/or symbolic exchanges among members. These help to reproduce social relationships and lead to economic, cultural or symbolic profits. Similar to Bourdieu, Lin (2001: 25) defines social capital as ‘resources embedded in a social structure that are accessed and/or mobilized in purposive actions’. Both conceptualisations of social capital emphasise the resources, such as information or help that contacts provide individuals with, while the contacts themselves remain secondary.

Social capital has been operationalised in a number of ways. Consistent with the theorists’ focus on information, much of the stratification research has measured social capital in terms of social contacts (Smith 2000). Still, some social contacts have novel information to pass on, while others are less influential. To the extent that men have more resources in the labour market, differences in the gender composition of social networks structure access to resourceful contacts (DiMaggio & Garip 2012; Lin 2001).

Beyond network size and resourcefulness, the relevance of information that contacts have access to and their willingness to share advice play important roles in the returns to social contacts (Livingston 2006). Family has been argued to provide social support, while friends are more likely to have new information on job openings (Elrick & Lewandowska 2008). Prior work from the United States also shows that women predominantly receive job information from other women, whereas men tend to find out about jobs from other men (Fernandez & Sosa 2005). Considering gender divisions across employment sectors, women may have more relevant information for other women than men.

Individuals may also be reluctant to share information with contacts who are in the outgroup, as they signal to hiring managers that their job-seeking tie is suitable for the post. Indeed, prior work shows that referrals can have consequences for the individual’s prospects in the firm. For instance, Swedish youth appear to benefit from their parents’ referrals through faster access to jobs and better labour market outcomes, but their parents’ wage growth drops when one of their children enters the firm (Kramarz & Nordström Skans 2014; Marin 2012). Findings based on a sample of professional, technical and managerial workers in California indicate that women tend to receive less work-related help and worse job information from their contacts than similarly positioned men (Huffman & Torres 2002).

In sum, previous findings reveal differences in the composition of men and women’s networks, which may shape the resourcefulness of their networks. The pertinence of information that contacts have access to and their willingness to share advice can also impact men and women’s labour market outcomes differently.

Migrants’ Integration and Social Contacts

According to network theory, the decision of whether and where to migrate is often based on social contacts who have previously migrated and established themselves in the destination country (Garip & Asad 2016). In particular, family reunion migrants relocate to join their family members abroad. Among refugees, the decision to migrate is predominantly driven by war or political unrest. Still, they can, to some degree, choose the country of destination.

Upon arrival, social contacts tend to assist newly arrived migrants by helping them find housing and employment (Lancee 2010). In light of language difficulties, reduced insight into the labour market and discrimination, migrants rely more frequently on their social contacts in the job search than natives (Åslund, Forslund & Liljeberg 2017; Carlsson, Eriksson & Rooth 2018). Still, different aspirations and decisions of where to settle can imply variation in the returns to social contacts. Some individuals apply for family reunification or asylum in Sweden rather than another European country due to perceived opportunities, while others follow or join their family despite uncertainties about work opportunities in the destination country (Crawley & Hagen-Zanker 2019).

Regarding gender differences, research from the United States shows that social contacts are associated with a higher likelihood of employment for migrant men than for migrant women (Aguilera 2005; Hagan 1998; Livingston 2006). Likewise, in the Netherlands, female migrants have been found to be less likely to use social contacts in the job search than male migrants, though men and women have access to similarly resourceful networks (Bilecen & Seibel 2021). Turkish male migrants also appear more likely to rely on their social contacts to find employment than female migrants in Germany (Drever & Hoffmeister 2008), whereas other research from Germany and the Netherlands indicates no gender differences (Kanas, van Tubergen & van der Lippe 2011; Lancee 2010). Heterogeneous findings may be in part related to differences across migrant groups. Studies from the United States focus on one or two migrant origin groups from Latin America. By contrast, prior research from Europe analyses heterogeneous migrant groups, running only a few analyses separately by region of origin. Despite some mixed findings, these studies point to gender differences in the relationship between social contacts and labour market outcomes. They furthermore suggest considerable heterogeneity across migrant groups.

Differences across migrants are important to the extent that they shape integration patterns and returns to social contacts. According to the segmented assimilation theory, distinct integration pathways result from an interaction between migrants’ human capital and pre-migration experiences, on the one hand, and the policies and prejudice in the receiving society, on the other (Portes & Zhou 1993; Zhou & Gonzales 2019). Indeed, male and female migrants with high education levels tend to have a smoother transition into the labour market than those with low education (Åslund, Forslund & Liljeberg 2017; Bevelander & Irastorza 2014). Prior work has also found that social contacts assist migrants and natives in finding work in comparatively low positions in the labour market. In contrast, social contacts do not appear to be as helpful in gaining access to skilled positions (Kalter & Kogan 2014; Oesch & von Ow 2017).

Another aspect that may shape differential returns to social contacts is migrant origin. Migrants from geographically distant countries often receive lower returns for their educational qualifications and work experience on the destination labour market (Andersson Joona, Gupta & Wadensjö 2014; Dahlstedt 2015). Some origin groups also experience a particularly high threshold to entering the labour market, in part due to employer discrimination. Prior work indicates that non-European migrants are more likely to face discrimination in the labour market than European migrants (Drouhot & Nee 2019; Quillian & Midtbøen 2021). For these groups, social contacts can be important in overcoming the threshold to entering the labour market, whereas a lack of social contacts may imply a double burden.

The theoretical and empirical literature reviewed above leads me to a number of expectations. According to Lin’s (2000) theoretical framework, contacts can assist the job search through information and help. However, prior work on gender differences in the returns to social contacts suggests that women’s contacts may have less relevant information than men’s or may be less willing to share advice with them (Huffman & Torres 2002; Livingston 2006; Marin 2012; Smith 2005). This leads me to expect that social contacts will assist entry into employment among men, while the returns to contacts are expected to be lower for women. The type of contact may also play a role. While family often provides a support network, friends may be more likely to have new information on job openings (Elrick & Lewandowska 2008).

Following the segmented assimilation theory (Portes & Zhou 1993), I further expect heterogeneity in the integration pathways by migrants’ human capital and origin. According to this theory, the segmented character of assimilation stems from an interaction between migrants’ human capital, policies and prejudice in the receiving society (Zhou & Gonzales 2019). This leads me to expect that the role of social contacts differs by migrants’ educational attainment and that these patterns are gender-specific. Male and female migrants’ exposure to discrimination also differs across regions of origins, leading me to expect that the importance of social contacts differs across migrants’ origin and gender.

Gender Divisions in the Labour Market

Sweden is commonly characterised as having a dual-earner or earner-carer family policy model, that is, supporting female labour force participation by providing public day-care services and paid parental leave (Korpi, Ferrarini & Englund 2013). In 2022, the female relative to male labour force participation was 91% in Sweden when compared to the European average of 82% (ILO 2023). Still, gender differences in terms of income, industry and employment sector prevail (Statistics Sweden 2022). For instance, 70% of employees in the public sector were women in 2015. In contrast, in the private sector, 37% of employees were women (see Table A1 in the Appendix). In the 1970s, the gender wage gap declined, but this trend did not continue into the 1980s or after (Boye, Halldén & Magnusson 2017). Differences are partly related to lower wages in female-dominated occupations, the unequal division of childcare between men and women, and a large proportion of women working part-time when they have children.

Even though obtaining a job is strongly linked to searching through social contacts in certain occupations, there are notable differences in typical recruitment methods across employment sectors (Helgertz 2011). Employers in the public sector are over-represented in the use of formal channels, while the private sector predominantly uses informal recruitment channels such as employees’ referrals (Lundin 2011). In light of gender divisions across employment sectors, this variation in recruitment methods likely rewards men’s use of social contacts more than women’s.

Migration and Integration

Since the 1970s, migration to Sweden has been dominated by refugee and family reunification migration (Migration Agency 2023). They came predominantly from Latin America and the Middle East in the 1970s and 1980s and from Somalia and ex-Yugoslavia in the early 1990s. The economic recession in the 1990s collided with refugee immigration and impacted particularly the employment rate of foreign-born individuals (Bevelander 2011).

The Swedish government has undertaken active integration efforts by providing refugee and family reunion migrants with transition programs and free language classes (Mangrio, Carlson & Zdravkovic 2020). Yet, employment rates were about 10% lower for male migrants and about 20% lower for female migrants compared to their Swedish-born counterparts in 2020 (Eurostat 2023). The larger employment gap among women is only in part accounted for by differences in parental leave, as migrants tend to use less parental leave than natives (Mussino & Duvander 2016). Alternative explanations have underlined female migrants’ limited possibilities for work-related development and gendered as well as ethnic discrimination in the labour market (Helgertz 2011).

Among migrants, gendered sorting into occupation and employment sector parallels patterns observed in the native population. Whereas 69% of foreign-born employees in the public sector were women in 2015, in the private sector 38% of foreign-born employees were women (see Table A1 in the Appendix). This suggests that, similar to Swedish-born women, migrant women have more difficulties in using social contacts to get a job than men. In sum, factors from the origin and destination play a role for entry into employment and returns to social contacts.

Variables

Migrants’ ability to find employment is an important benchmark for integration. Economic integration is measured by the years elapsed between migration and entry into employment, building on the survey questions What year did you immigrate to Sweden? and What year did you start your first job in Sweden? Using a single dependent variable allows me to model the independent variables with greater complexity across various specifications. The variable is censored at ten years. In this way, I capture initial differences in entry into the first job as well as longer-term distinctions in employment chances.

Social contacts are difficult to measure and have been operationalised in a number of different ways: (1) by firm or neighbourhood, generally focusing on the percentage of individuals from the same origin in the firm or neighbourhood (Bygren & Szulkin 2010; Dustmann et al. 2016), (2) social capital indices (Lancee 2010), (3) position generator items regarding work-related and friendship networks, often analysing second rather than first-generation migrants (Nygård & Behtoui 2020), (4) egocentric or sociometric network information (Hofstra et al. 2017), and (5) social contacts at arrival (Kalter & Kogan 2014; Kanas, van Tubergen & van der Lippe 2011).

This paper takes the fifth approach and differentiates between having family, friends, both family and friends or no contacts in Sweden. Specifically, the survey includes the question, Did you know someone in Sweden before you arrived? And if so, who? Respondents can answer if they knew family, friends, others, nobody or a combination of the former. For clarity, the question refers to individuals who were already in the country when the respondent arrived. When answering others, respondents specify who they knew. Their written responses are manually recoded as family or friends. Among those recoded as having family contacts, spouses and cousins are common answers. Among those who are recoded as friends, responses range from members of an organisation to work colleagues.

Despite providing less fine-grained information on social contacts than some of the other approaches used in the literature, this measure has two advantages. First, it pertains to contacts that respondents had at arrival in Sweden and thus precedes entry into employment, as refugees and family reunion migrants generally start looking for employment upon arrival. Second, it provides a strong anchor point (the year of migration) to improve recall (Hipp et al., 2020). Remembering the circumstances surrounding migration, which often represents a substantial change in the life course of individuals, is likely easier than remembering other less salient events (Smith and Duncan 2003).

I also control for a set of individual-level characteristics that have been shown to be determinants of migrant integration in the literature (van Turbergen 2022). They include age at arrival, region of origin, residence permit and year of arrival. I also control for marital status, parenthood and educational attainment as time-varying variables.

Methods

I use an event history set-up, which allows me to assess both the likelihood and the timing of the event. I run piecewise constant exponential models and let the baseline hazard of entering a job to vary on a yearly basis (Blossfeld et al. 2019). Persons enter the risk set (i.e. the observation window) at immigration. The event of interest is entry into the first job.

I present hazard ratios from two specifications that include different subsets of the independent variables. Model 1 controls for demographic characteristics (age and age squared at the time of arrival), region of origin, residence permit and the year of arrival in quintiles. Model 2 additionally includes time-varying variables. They include marital status, parenthood and educational attainment. The indicator for marital status switches from zero to one in the year the migrant marries and subsequent years. Likewise, the indicator for parenthood is coded one in the birth year of the first child and subsequent years, and zero otherwise. Individuals who married and became parents prior to immigration are coded as one throughout the observation window. Educational attainment distinguishes between four levels of schooling: missing education (incomplete compulsory schooling), low (<11 years), intermediate (11–12 years) and high (13–17 years). Among migrants who do not continue their education in the destination country, this variable refers to the highest level of education completed in the home country. Missing education refers to the group with the lowest education, as incomplete compulsory schooling was coded as missing. Still, it is possible that some individuals did not disclose their education level at the interview. I run separate models for men and women to assess the gender-specific relationship between pre-migration contacts and time to first job, as well as allowing the associations between control variables and time to first job to vary by gender.

In a second step, I estimate stratified models by educational attainment and region of origin to assess heterogeneity in the relationship between pre-migration contacts and time to first job among men and women. I stratify the sample by educational attainment, distinguishing between two groups – missing or low education and intermediate or high education at arrival. I have also run sensitivity analyses stratifying by missing, low, intermediate and high education, finding no differences in the patterns observed for missing and low education, and intermediate and high education, respectively. The region of origin distinguishes between Europe, Central Asia and North America; Middle East and North Africa; South and East Asia and the Pacific; sub-Saharan Africa; and Latin America.

Methodological Considerations

This study focuses on associations and relationships. However, when trying to interpret the relationship between social contacts and job entry, endogeneity presents a fundamental challenge. For one, the phenomenon that individuals choose their friends and often prefer contact with similar others, as referred to by the term social homophily, implies that associations in part reflect selection effects based on the non-random ways in which people become friends (McPherson, Smith-Lovin & Cook 2001; Mouw 2006). Second, reverse causality or the tendency of co-workers to become friends can be important. In other words, individuals may find work and then form a new contact. Retrospective assessments that focus on one time point may not be able to disentangle this temporal issue.

Among migrants, these processes are yet more intricate as migration decisions are often linked to social contacts who have previously migrated (Garip & Asad 2016). While some migrants move to Sweden to join family or friends, others relocate due to perceived opportunities in the destination country (Crawley & Hagen-Zanker 2019). This implies that associations between social contacts and the labour market entry of migrants may reflect not only selection into friendships and reverse causality, but moreover selection into migration. Using information from one setting makes it difficult to disentangle selection into migration from other types of selection.

Notwithstanding these concerns, this study has several strengths in relation to previous studies. It focuses on refugee and family reunion migrants, for whom work is not the main driving factor of migration. In addition, it examines contacts that were formed before migration. This indicates that contacts were formed before entry into employment and reduces concerns regarding reverse causality. It can still be that there is selection into friendships and selection into migration, which are partly addressed by controlling for migrants’ characteristics and demonstrating that the coefficients remain robust across models. While this implies that the results present associations that are likely to reflect some degree of endogeneity, it is worth noting that they provide insight into real-life networks. Friendship selection and selection into migration are inherently endogenous processes. Identifying the effect of contacts that individuals did not choose themselves is a different exercise that, while worthwhile, may be difficult to use to understand migrants’ actual experiences in finding a job.

Descriptive Results

Table 1 presents descriptive statistics separately by gender, with the dependent variable shown in the top panel. On average, women take longer to enter their first job in Sweden (4.24 years) than men (3.43 years). Likewise, the proportion who have entered a job within 5 years is 61% among women and 68% among men. However, after 10 years, gender differences are relatively small (74% of men have entered a job versus 69% of women).

Our leading independent variable is contacts at arrival. Family contacts are the most common among men and women. More than half of women have family in Sweden, and 45% of men have family contacts. Nearly a fifth of men and women have friends in Sweden, while having both family and friends is relatively uncommon (5%). Arriving without contacts is more common for men, among whom one in three has no contacts in Sweden. The remaining panels in Table 1 show mean values for the other independent variables.

To gain insight into gender differences in the timing of entry into employment, Figure 1 provides raw survival estimates on time to first job separately for men and women. The top panel reveals that about half of male migrants enter a job in the first year in Sweden. Male migrants, who have family or friends in Sweden, tend to enter employment faster than their peers. They are followed by male migrants, who have both family and friends in Sweden, whereas male migrants who do not have contacts take somewhat longer to locate work. The bottom panel shows that female migrants with friends enter the labour market faster than the other groups during the first 5 years, but are then overtaken by migrants with both family and friends. Female migrants who have family or no contacts tend to take longer to enter employment than their counterparts.

Figure 1

Gender Differences in Time to First Job among Refugee and Family Reunion Migrants

The next section assesses gender differences in the returns to pre-migration contacts at entry into employment. Table 2 provides hazard ratios from models estimating time to the first job. Hazard ratios above one indicate a shorter time to the first job, and hazard ratios below one indicate a longer time to the first job.

Among men, family and friends are associated with a shorter time to first job when compared to no contacts, as indicated by hazard ratios of 1.26. With respect to demographic characteristics, the association between age and time to first job is curvilinear, with the time to first job declining across younger ages and increasing at older ages. We observe little variation across regions of origin, while refugee status is associated with a longer time to first job than family reunification. Model 2 additionally controls for time-varying variables. Results show that the main relationships for pre-migration contacts hold. While the coefficient for family decreases somewhat, the coefficient for friends remains stable. This model also reveals that married migrants tend to have a shorter time to their first job than unmarried migrants. Parenthood and educational attainment are not associated with time to first job, apart from missing education, which is negatively related to time to first job.

Among women, pre-migration contacts are not associated with an equivalent decrease in time to first job. The curvilinear association between age at arrival and time to first job is similar to that observed among men. In contrast, among women, we observe considerable differences across regions of origin, where female migrants from the Middle East and North Africa and sub-Saharan Africa take longer to enter work than their counterparts from Europe, Central Asia and North America. Residence permit does not appear to be related to time to first job. In model 2, married individuals and female migrants with intermediate and high education tend to enter employment more quickly than their unmarried and low-educated counterparts.

This analysis suggests that different mechanisms are at play for men and women. While men appear to benefit from their contacts, women seem to gain from their education. Among women, we also find considerable differences across regions of origin.

Differences by Educational Attainment and Region of Origin

Having established gender differences, the next part assesses what educational attainment and region of origin can tell us about the differential relationship. Table 3 presents results from models that use a simplified measure of contacts. Specifically, we assess a binary indicator that is equal to one when migrants report having contacts at arrival in Sweden (family, friends and/or family and friends) and zero otherwise. This is done as stratifying the sample leads to small sub-samples. For completeness, results from models using the more detailed indicator of contacts are provided in Tables A2 and A3 in the Appendix.

Table 3 shows that men with low educational attainment and missing information on education tend to benefit from their social contacts. In contrast, contacts are not related to labour market entry among men with higher education. This indicates that the positive association observed among men (in Table 2) is mainly the result of the smoother labour market transition among low-educated men with social contacts. Among women, we find no difference in the returns to social contacts by educational attainment.

Table 3 further shows that men from Sub-Saharan Africa and Latin America benefit from their social contacts, whereas contacts are not related to labour market entry among the other origin groups. Among women, having contacts is related to a shorter job search among Sub-Saharan African migrants. In contrast, female migrants from the Middle East and North Africa and Latin America tend to have a longer job search when they have contacts in Sweden.

Together, these results reveal that social contacts promote labour market entry among low-educated men. Social contacts also appear particularly beneficial for men from sub-Saharan Africa and Latin America. In contrast, among women, social contacts are associated with a shorter job search for some origin groups, while they are associated with a longer job search for others.