1 Introduction

In (post-)industrialised societies, paid work plays a central role and affects two important dimensions of an individual’s life—one material and one socio-cultural. Economically, employment affects autonomy and living conditions (Gundert and Hohendanner 2014), whereas on a socio-cultural level, work can define an individual’s identity, social role and status (Schöb 2013). Through both aspects, employment defines working and leisure time, shapes daily activities, networks and commutes, and thereby influences the people we meet and the places we visit (Pohlan 2019; Schöb 2013; Kunze and Suppa 2017). As paid labour is of marked importance in modern societies, it is a substantial endeavour to study the consequences of labour market difficulties for social integration.

At the societal level, studies have investigated how economic inequality (Wilkinson and Pickett 2009) or the labour market affects social integration and cohesion (Whelan and Maître 2005). As society is made up of individuals and their interactions, social relations form the basis of societal cohesion. Therefore, in this study, I focus on the individual level to explore how experiencing unemployment affects personal relations to family and friends. Close and stable relations with others are an important aspect of an individual’s social integration, as they yield embeddedness and belonging through the provision of emotional and financial support (Russell et al. 2013; Lelkes 2010). These relationships can build a foundation for cultural, civic and political practices encouraging trust and recognition (Gundert and Hohendanner 2014). Analysing the relationship between unemployment and social integration at the micro level can shed light on social integration at the societal level.

The literature suggests two main mechanisms linking unemployment to personal relations and hence micro-level social integration. The first focuses on material losses and financial problems caused by unemployment, which directly affect social life by reducing one’s capacity to afford activities or through the indirect impacts of shame and discrimination (Rogge and Kieselbach 2009; Vogel 2000; Gallie et al. 2003). Until now, research has produced mixed results. Some studies detect negative effects attributable to income losses and material deprivation (Russell et al. 2013, p. 249) initiating a vicious circle of social exclusion (Gallie et al. 2003). Others find that financial resources explain negative effects of adverse labour market situations only in part (Gundert and Hohendanner 2014) or to a minor extent (Winkelmann and Winkelmann 1998; Schöb 2013).

The second mechanism concentrates on identity struggles and difficulties living up to society’s normative expectations (Stryker and Burke 2000; Petersen 2011; Rogge and Kieselbach 2009; Schöb 2013). It suggests that those for whom success is particularly important and whose identity and social roles depend on performing well in the labour market are more strongly impacted by difficulties related to their work than those who have access to alternative social roles (Gurr and Lang 2018; Schöb 2013; McDonald 2000; Schmitt 2008). Owing to the complexity of measuring social identity, shame, and discrimination quantitatively, there are fewer studies testing this hypothesis extensively. Rözer et al. (2020) differentiate effects of unemployment by age, gender, and education and find varying effect sizes and directions depending on the type of social contact. In addition, there are only a few studies analysing both mechanisms (see for an exception Winkelmann and Winkelmann 1998; Heyne and Voßemer 2023).

Drawing on insights from gender theory, I explore two categories of socio-structural position: gender and household composition. Gender is one of the key dimensions structuring socialisation and identity. Moreover, it influences both labour market opportunities and family responsibilities.

Research on the effects of labour market marginalisation frequently points to traditional gender roles to explain gender-specific effects of unemployment (Schmitt 2008). As an example, women often perform a larger share of care and household tasks (Samtleben 2019; Allmendinger 2010; Nitsche and Grunow 2016; Grunow 2013). Care and family tasks can function as an alternative source of identity when confronted with unemployment (Schöb 2013; Schmitt 2008). For many women, being a homemaker and caring for relatives is not only an accessible social role, but also normatively accepted by mainstream society (Kronauer 2010, p. 45). In contrast to this, men are more often normatively responsible for financially providing for the family (Grunow and Baur 2014), which ties their identities to their jobs and occupations (Allmendinger 2010; Josselson 1998). This tendency is often more pronounced after family formation and with the presence of children (Grunow and Baur 2014), and it depends on the centrality of work to an individual’s identity, which is reflected by attitudes and values concerning work and family life (Schöb 2013).

Against this background, I expect that unemployment does not lead to impoverished social relations for each person to the same degree, but rather that gender, parenthood and individual values with respect to work and family influence the degree to which unemployment affects close personal relations. Unemployment can reduce available material resources, cause financial strain and result in identity struggles and psychological distress. Using German Socio-Economic Panel data (SOEP), this study investigates the effects of labour market disadvantage in the form of unemployment on close social relations. More specifically, through individual fixed effects regressions I assess to what extent financial strain and social identity struggles explain the connection between unemployment and close social relations.

2 Theoretical Considerations

The literature discusses two broad explanatory mechanisms linking unemployment to stress in social relations and potential social exclusion: the loss of income and financial resources (Rogge and Kieselbach 2009; Vogel 2000; Gallie et al. 2003; Russell et al. 2013) and difficulty performing a social role and identity struggles (Stryker and Burke 2000; Petersen 2011; Rogge and Kieselbach 2009; Schöb 2013).

2.1 Financial Strain—Economic Resources Impact Possibilities for Social Relations

Neo-classical economic theory and social exclusion theory both focus on the materialist dimensions of job loss and unemployment. These theoretical explanations are based on the belief that an “essential function” of paid employment in industrialised societies is to provide economic resources to maintain an adequate standard of living and to enable “individuals to satisfy socially defined needs considered important in consumption-oriented societies” (Gundert and Hohendanner 2014, p. 137; Ervasti and Venetoklis 2010). Furthermore, employment grants autonomy. The scarring caused by unemployment on future careers (Brand 2015; Gangl 2006) can impact these functions because of its association with income losses. When multiple experiences of unemployment occur or unemployment duration increases (Brand 2015), more financial resources are consumed (Kronauer 2010). Thus, unemployment often causes financial distress, insecurities, and increased poverty risks (Gallie 2013). The risk of poverty following unemployment has been shown independently of context (Gallie 2004) and social class (Vandecasteele 2011). The reduction of disposable income (Kunze and Suppa 2017; Schöb 2013) can lead to difficulties maintaining one’s previous standard of living, forcing individuals to move to more affordable places of residence (Pohlan 2019) and reduce cultural activities and social events (Pohlan 2019; Kunze and Suppa 2017). As such activities affect networks as well as a sense of belonging (see Gundert and Hohendanner 2014), the lack of resources to participate can negatively influence social relations. However, it might be that people reduce costly activities and spend more time with less expensive ones while remaining socially connected (Kunze and Suppa 2017).

In addition to this, economic theory takes time as a valuable and limited resource into account. Fewer hours spent in paid employment translate to more available leisure time (Schöb 2013), which can increase social interactions and participation. In line with this argument, some research shows that unemployed individuals increase their leisure activities and socialise more often (Pohlan 2019; Kunze and Suppa 2017). What is rarely addressed, however, is that time might be needed to find a new position (Krueger and Mueller 2012) and resolve existential problems when financial resources are sparse, cancelling out additional free time.

Social exclusion theory analyses the consequences of unemployment from a sociological perspective, directing its attention to exclusion as a dynamic process (Rogge and Kieselbach 2009; Vogel 2000; Kronauer 2010; Bude and Lantermann 2006). Following this approach, labour market difficulties, economic disadvantage and problems within social networks mutually reinforce each other and solidify over time, causing a downward spiral or a vicious circle of exclusion (Rogge and Kieselbach 2009; Gallie et al. 2003). In this process, financial difficulties are a key mechanism linking unemployment to a heightened risk of social isolation (Rogge and Kieselbach 2009). With fewer resources at hand, favours are difficult to return, inviting people can be a financial burden and abstaining from invitations might violate norms of reciprocity. Research has shown that financial difficulties can result in exclusion from “status groups and social circles” where affiliation and belonging do indeed depend on economic resources and consumption patterns (Mood and Jonsson 2016, p. 636). Consequently, people facing economic difficulties might withdraw from social interactions or be excluded from their networks.

Theoretically, economic hardship can affect social relations in ways that are not purely negative. The thesis of compensation argues that financial difficulties might evoke solidarity and increase support from friends and family (Böhnke 2008). Yet, empirically, accumulation of social disintegration and financial hardship prevails (Böhnke 2008). Most studies found that poverty and social exclusion were mutually reinforcing (Devicienti and Poggi 2011) and material hardship had negative effects on social relations (Böhnke and Link 2017; Mood and Jonsson 2016). Furthermore, studies found that financial strain played an important role in linking unemployment to life satisfaction, which is often used as an indicator for social inclusion (Russell et al. 2013, p. 249).

Following these theoretical considerations, I assume that financial difficulties explain the effects of unemployment on personal relations (H 1).

2.2 Social Identity—Gender, Household Composition and Individual Values

Although measuring income loss and financial strain is comparatively straightforward, social identity is more difficult to operationalise and identity struggles are complex to assess empirically. Yet, an individual’s identity is often tied to social roles that depend on socio-demographic characteristics such as gender and familial situation. Gender is one of the key dimensions structuring socialisation and identity; it influences one’s position in the labour market as well as family responsibilities. Gender acts on different levels of society: on the macro level “gender involves cultural beliefs and […] distributions of resources”, and at the individual level it shapes behaviour as well as “selves and identities” (Ridgeway and Correll 2004, pp. 510–511). In Germany, as in many other industrialised societies, the “male breadwinner and housewife” family model has been dominant culturally since the urban bourgeoisie imposed it in industrializing Europe (Pfau-Effinger 2004, p. 377). In practice, this has evolved into a “male breadwinner and female part-timer” arrangement, but the male breadwinner family model remains rooted in German culture, social policy and family norms (Trappe et al. 2015).

Research shows that, in environments where traditional family norms prevail, the gender-specific division of labour and attitudes towards men financially providing for the family are strong (Grunow and Baur 2014). In the context of this cultural norm, male identities are bound to paid labour and tied to their professional success, with fathers normatively responsible for providing for the family (Allmendinger 2010; Josselson 1998). An important way for men to preserve their dignity is to provide for their families, which acts as a “proof of moral fortitude and masculinity” (Lamont 2000, p. 35). This makes male identities particularly vulnerable to unemployment. In contrast to men, women were and partly still are socialised with other standards of self-evaluation and external perception. This puts them in a position to define more individually what work means to them (Josselson 1998). In accordance with this, research suggests that men might suffer more from unemployment, especially when it occurs multiple times (Oesch and Lipps 2013).

Moreover, individuals are embedded within the lives of family members and partners (Giele and Elder 1998). The family and household not only shape individual life trajectories, but also the stratification of society. The household “is an important source of welfare and plays a […] fundamental role in (re-)shaping and generating social inequality by supplying wealth, resources, and support” (Grotti and Scherer 2014, p. 628). A partner can buffer financial difficulties. However, if both partners face difficulties at the labour market, it may be that problems accumulate at the household level (Grotti and Scherer 2014).

So far, most empirical evidence supports the persistent importance of gender for labour market participation and household division of labour (Samtleben 2019; Allmendinger 2010; Nitsche and Grunow 2016; Grunow 2013; Gonalons-Pons 2015). Research indicates that, although gender norms changed in favour of paid work for women, they did less so in favour of men engaging in housework (Grunow and Baur 2014). Traditional gender roles are stronger the more traditional the household situation: men living with a partner perform less housework than men living by themselves, and particularly so the more children live in the household (Grunow and Baur 2014). Paternal engagement tends to be limited in Western societies (Fuwa 2004). Often, even couples that initially adopt a gender egalitarian division of labour develop more gender-specific division practices when forming a family (Grunow 2013). In line with this, research on Germany shows that many mothers have “discontinuous employment experiences” (Lersch et al. 2017) and reduced labour participation (Allmendinger 2010).

This shows that the composition of the household, including whether someone lives with a partner and the presence of children, is an important factor in explaining the effects of unemployment. The “male breadwinner, female homemaker” model tends to be reinforced by the presence of children, and traditional gender roles can explain gender- and household-specific effects of unemployment (Schmitt 2008). Women often have to handle competing and contradictory expectations within the family and professional realm, such as being a caring mother and wife and a committed professional (Sabelis and Schilling 2013). This can lead to economic dependency and increased risks of old age poverty, particularly for women of lower economic backgrounds. However, care and family tasks might function as alternative sources of identity that aid in coping with unemployment (Schöb 2013; Schmitt 2008), as children can provide meaning and offer opportunities to participate socially (McDonald 2000). Although single women experience a stronger norm to be employed, for women living with a partner it is “easier to self-categorise as ‘housewife’ or ‘mother’ rather than ‘unemployed’” because the environment is less likely “to distinguish between stigmatised unemployment and voluntary inactivity due to intra-household division of labor” (Schöb 2013, p. 171). However, for men in traditional family settings, unemployment is a potential threat to their identity. Research shows that men (Heyne and Voßemer 2023), and particularly partnered men, suffer more from unemployment (Schöb 2013), and that it is primarily male unemployment that puts relationships to the test (Kinnunen and Feldt 2004).

Whether unemployment results in conflicts with an individual’s social role and negatively affects his or her social integration depends on the standards attached to the respective role. Socially and culturally prioritised norms achieve a hegemonic power and become intuitive. They represent “the implicitly or explicitly shared abstract ideas about what is good, right, and desirable in a society” (Schwartz 1999, p. 25; Groh-Samberg et al., this issue). Social norms define the standards for social roles and hence have an impact on individual attitudes and values (Schwartz 1999). To understand why and how unemployment can translate into role and identity conflicts, it is useful to take a close look at values around work and the family. In many industrialised societies, the prevailing work ethic assigns a central value to work, which in turn becomes a vehicle for meaning, purpose and fulfilment (Schöb 2013; Grimm et al., this issue). However, social roles vary in the extent to which paid labour is central to their fulfilment. Social roles entail different culturally shaped work and family values and societal expectations, and the set of associated behavioural standards varies accordingly. Therefore, the importance of paid labour and living in a (traditional) family context is not the same for all members of society. Put another way, individual work values refer to aspirations at work and determine its centrality relative to other aspects such as family and community life (Schwartz 1999).

These theoretical arguments and previous findings suggest that an individual’s social role and values regarding work and family life are influential in explaining this relationship (H 2). More specifically, the social identity approach assumes that the effect of unemployment on social relations is stronger for men than for women (H 2a). Moreover, the literature indicates that family and household composition shape the experience and social consequences of unemployment. Two basic dimensions characterise a “traditional” family: the presence of children and living with a partner. This implies that having a partner and children intensifies negative unemployment effects for men and mitigates them for women (H 2b). Finally, although strong work norms should intensify negative unemployment effects for all, as this indicates that an individual’s identity is tied to employment, strong family norms should, in accordance with gendered social roles, mitigate negative effects of unemployment for women and intensify them for men (H 2c).

3 Data and Methods

3.1 Analytical Strategy

To test my hypotheses empirically, I compute individual fixed effects linear regression models. The models assess the impact of within-individual changes in unemployment on within-individual changes in social relations. Individual fixed effects account for all time-invariant individual characteristics such as social background and stable character traits. This modelling strategy produces a conservative estimate of the effect of unemployment on changes in close social relations. As SOEP data sample individuals in households and the analyses consist of both individual and household-level variables, standard errors are corrected at the original sample household level to account for mutual dependence of individuals from the same household.

I employ an analytical approach exploring both financial strain and identity struggles as potential explanatory mechanisms linking unemployment to the quality and frequency of personal relations. The analytical setup proceeds in three main steps. First, I explore the effects of cumulative unemployment on social relations, with three distinct specifications of the unemployment variable that capture different durations and degrees of accumulation of unemployment (I). Second, I conduct a mediation analysis and compare the results from models with and without controlling for the household’s financial situation to test for financial difficulties as an explanatory mechanism (II). Third, as identity struggles are more difficult to examine empirically, I approximate identity in four different ways (III). First, I consider the part of the effects that cannot be explained by financial difficulties as a first indication for the social identity mechanism. In a second step, I differentiate the effects of unemployment by two dimensions of structural inequality, gender and household composition. These socio-demographic characteristics shape an individual’s identity, social roles and normative expectations. Third, to capture how important paid labour is to the individual’s identity and to what extent family and care tasks may be an alternative source of identity, I analyse the importance of attitudes to work and family for shaping the effects of unemployment on personal relations with family and friends.

3.2 Operationalisation of Key Concepts

This article investigates the effects of labour market marginalisation from a processual perspective, as well as its consequences for social integration through close personal relations. Through all models, the dependent variable is an additive index that captures close personal relations and is composed of two variablesFootnote 1: the frequency of meeting and the frequency of helping friends and family. This indicator is intended to measure individual embeddedness through the quantity and quality of close social relations. The question behind the variables is: which of the following activities do you take part in during your free time? Please check off how often you do each activity: at least once a week, at least once a month, less often, never. To construct the index about personal relations I chose variables capturing the two following activities: “Meeting with friends, relatives or neighbours” and “Helping out friends, relatives or neighbours”. This not only considers the frequency of interactions, but also adds a qualitative measure of intensity through the variable about helping others. The composite measure is based on the theoretical aim of coming closer to “meaning-based measures of structural isolation” rather than employing a measure based solely on the frequency of meeting others (Parigi and Henson 2014, p. 162). The index ranges from 1 to 7. An individual who meets friends and family frequently but never helps have a medium score; the same is true for someone who helps often but does not meet others regularly. People who often meet and help others have the highest score, and those who rarely engage in either of the activities have a low score. Table 1 shows the mean frequencies of the social relations index by gender. As gender is one of the core analytical dimensions, it is crucial to compare baseline levels of the outcome variable. The mean frequency of meeting and helping friends and family is almost identical for men and women. Furthermore, to account for potential differences based on the construction of the dependent variable, I compute the baseline model for each indicator separately. The results in Table 7 in the Online Appendix show that using only meeting or only helping friends and family as the outcome variable yields similar results, with social meetings showing a slightly stronger gender differential than helping friends and family.

Table 1 Mean frequency of social relations by gender
Full size table

For my main explanatory concept, I measure cumulative unemployment in five distinct ways. The baseline variable captures the cumulated time an individual has experienced unemployment over their entire career in years and months in decimal form up until the point of the interview. In the regression model, this continuous variable provides the effect of an additional unit of unemployment (1.2 months), a short additional experience within an individual’s working life, which probably produces conservative estimates of unemployment effects. Moreover, including this simple continuous variable assumes effect linearity, which means that each additional 1.2-month period adds the same effect, whereas it could be that there is a threshold after which unemployment becomes more of a burden.

To explicitly capture the effects of longer or multiple unemployment experiences and distinguish them from short-term or singular events, I employ four additional measurements of cumulative unemployment. First, I transform the baseline variable into a categorical one containing six categories. The first contains respondents without any unemployment experience, the second those who have over their working lives accumulated up to half a year of unemployment, the third those with up to 1 year, the fourth those with 2 years, the fifth those with 5 years, and the sixth those with more than 5 years of cumulated unemployment. The categories are based on the distribution and frequency within the sample. This operationalisation is based on the distribution of unemployment within the sample and allows me to detect turning points, habituation and intensification of effects, and to visualise thresholds after which effects may change their impact. However, the variable does not show whether someone has had multiple short spells of unemployment or a single long experience. Therefore, to better capture continuous long-term experiences in the near past, I construct three variables that capture the amount of unemployment that someone has experienced within the past 2 years. Each variable indicates whether the respondent has had no unemployment in the last 2 years; or less than, greater than, or equal to 12, 18 or 24 months of unemployment respectively. This captures the effect of unemployment at the critical point of switching from unemployment benefits with above 60% of the previous income to a basic and means-tested social assistance, implying potentially more severe income losses.

To contrast the cumulative nature of disadvantage with current and immediate effects, I also include a variable measuring the individual’s current employment status, distinguishing between unemployment and full-time, part-time, or marginal employment.

The first explanatory mechanism linking unemployment to social relations is financial strain. The main reason is that risks and resources are often shared at the household level, where one partner can support the other but a difficult period for one person will also affect other household members. To account for this, I use objective and subjective measures of the household’s financial situation. The objective measure is the monthly net household equivalence income, which is obtained by dividing the household income by the square root of the number of household members (OECD Project on Income Distribution and Poverty, 2020). The subjective variable measures the satisfaction of the individual with the current financial situation of the household. As fixed effects models measure within-individual change over time, all models including the household’s financial situation estimate the effects of changes in the objective and subjective financial situation.

For my second explanatory mechanism, I follow a multidimensional approach to test for social identity struggles. First, I take two structuring dimensions of socio-economic position, gender and household composition into account. The gender variable distinguishes between men and women and the household composition consists of two variables: partnership status, which distinguishes between having a partner or not (independently of the legal marital status); and the presence of at least one child below the age of 16 living in the householdFootnote 2. In a second step, I try to measure social identity more explicitly using two indices to determine how important family and work are to someone. The index of family-related values is composed of three variables measuring the importance the respondent places on a happy partnership or marriage, on having children, and on generally caring for others. The index for work is composed of two variables measuring the importance the respondent places on self-realisation and a successful career. Each index is composed of a set of variables, with each of the variables having four categories: not important, less important, important, and very important. These are transformed into a scale from 1 to 10 to make them comparable, the different number of underlying variables notwithstandingFootnote 3. The SOEP waves including these variables are from the years 1990, 1992, 1995, 2004, 2008, 2010, 2012 and 2016 and do not coincide perfectly with the waves including the dependent variable on social relations (1990, 1992, 1994, 1996, 1997, 1999, 2002, 2005, 2007, 2009, 2011, 2015, 2017). Therefore, I carried forward the variables to match the exact time point of measurement of the dependent variableFootnote 4. This is only an approximation of the ideal analytical model, as it assumes a more-or-less stable individual identity or set of character traits. Thus, results should be interpreted with this restriction in mind.

To account for the time spent actively in the labour market, all models control for the months worked full-time and part-time as well as individual age. Furthermore, all models include year dummies to consider general time trends such as economic crises or periods of economic growth and prosperity. Table 2 gives an overview of the variables used for the empirical analysis of this study.

Table 2 Overview of variables
Full size table

3.3 Data and Sample

To assess the hypotheses derived in the theory section empirically, I use data from the German Socio-Economic Panel (SOEP) from 1990 to 2017 (Liebig et al. 2019). As the aim of this study is to examine the consequences of unemployment on close personal relations, I have placed a couple of restrictions on the sample to make it more suitable for the analysis. As the setup of the analysis includes individual and household level variables, the analysis is restricted to respondents who have realised both personal and household interviews. Moreover, I have restricted the sample to individuals of working age. This means that all individuals below the age of 16 and above 70 years of age were excluded from the analysis. The age of retirement in Germany is currently at 65.5 years and steadily increasing (IAQ—Institut Arbeit und Qualifikation der Universität Duisburg-Essen 2019). Therefore, I decided to include those individuals between 64 and 70 who are still part of the workforceFootnote 5. This means that they reported working regular full- or part-time jobs or that they are looking for a job. Individuals above 64 who were retired or working minor jobs are not part of the analysis. Furthermore, I excluded all respondents who at the time of the interview were in education, early retirement or military service, independent of their age. Overall, the basic sample was obtained by list-wise deletion of missing values, resulting in a sample of 164,873 cases. I did not impute missing data, but I used (when available) the variables constructed by the data provider, which can have imputed values. For the net monthly household income, for example, about 4% of the values were imputed. Moreover, the variables measuring the importance of work and family are not included at each time point in the questionnaires of all subsamples or part of the SOEP. Therefore, the models analysing the impact of work and family values are based on a reduced sample (121,230 cases).Footnote 6

4 Unemployment and Personal Relations

The bivariate analysis of unemployment and the frequency of social interactions depicted in Fig. 1 gives a first sense of the relationship. The graph shows the mean values of the index for personal relations depending on the total amount of months someone has been unemployed. The hollow dots represent the mean frequency of social interactions. The dotted line is a fitted regression line of the relationship between the two variablesFootnote 7. We see a tendency for social interactions to decline the more or longer someone has been unemployed over their career/working life. However, the more unemployment is accumulated, the more variation appears in the data, suggesting that this negative relationship might not be universal. The bivariate association merely provides a first glimpse of the data. The following analyses inquire into the roles of financial strain and social identity and disentangle the effects of unemployment at the intersection of gender and household composition, also taking individual work and family values into account to carve out results hidden in these descriptive associations. The next sections include graphs displaying results from multivariate individual fixed effects models.

Fig. 1
figure 1

Mean frequency of meeting and helping friends and family by cumulated duration of unemployment. (SOEP 1990–2017—own calculation)

Full size image

4.1 Gendered Effects of Unemployment

Figure 2 displays the average marginal effect of unemployment on social relations. The left-hand plot shows the results for all respondents, whereas the right-hand plot differentiates the effect of unemployment by gender. The left-hand plot indicates that, when controlling for the individual’s current employment status, each additional month of unemployment has a small but significantly negative effect on the frequency of meeting and helping friends and family. As can be seen in models M1 and M2 in Table 3, the effect does not change with the inclusion of the household’s financial situation.

Fig. 2
figure 2

Average marginal effects of metric cumulative unemployment experiences on personal relations by gender. (SOEP 1990–2017—own calculation)

Full size image
Table 3 Individual fixed effects of cumulative unemployment on personal relations
Full size table

Although each additional month of unemployment shows a general decrease in how often someone meets and helps friends and family, this result does not reveal whether financial strain might play a more crucial role for certain social groups or depend on different durations of unemployment. To address this systematically, it is useful to differentiate unemployment effects by gender and unemployment duration.

The right-hand plot of Fig. 2 shows the average marginal effect of unemployment by gender. The grey dots indicate effects of experiencing unemployment for women; the black diamonds do so for men. The results support the second hypothesis (H 2a), as they reveal that, all else being equal, each additional month a man is unemployed reduces the frequency with which he meets and helps family and friends by 0.035 of an index unit. After half a year, the effect approximates 0.2 units and by the time someone has been unemployed for 5 years, the effect amounts to almost two index units, which is a palpable decline in social interactions.

Figure 3 has a similar setup to Fig. 2, with grey dots showing the effects for women and black diamonds for men. The graph further demonstrates how the effect of unemployment varies with different accumulated unemployment experiences. It becomes evident that the effects of unemployment, when analysed over the entirety of an individual’s working life, only become significantly negative at about 5 years of accumulated unemployment in total. Confirming the pattern uncovered in Fig. 2, the effects are only negative for men and cannot be explained by changes in the household’s financial situation. Yet, so far, this measure captures all kinds of scenarios, including individuals whose unemployment experiences were many years ago as well as respondents who have recently been unemployed. Similarly, the measure does not differentiate between multiple shorter episodes and a single longer one. Therefore, it remains unclear whether unemployment has a stronger effect on personal relations if experienced recently or whether the loss of disposable income and potentially related financial hardship are more salient when someone has been unemployed for a palpable amount of time in the recent past.

Fig. 3
figure 3

Average marginal effects of different cumulative unemployment durations on personal relations by gender. (SOEP 1990–2017—own calculation)

Full size image

In parallel to the previous figures, Fig. 4 differentiates the effects by gender, with black diamonds indicating effects of unemployment for men and grey dots for women. Furthermore, the graph shows three distinct plots, each representing whether someone has been unemployed for more or less than 12 months, 18 months and 24 months during the past 2 years respectively, compared with those who have not been unemployed at all during that period. What becomes visible is that the effect becomes stronger and more negative the longer someone has been unemployed during the past 2 years. Also with this measure, the strong negative effects are only visible for men. In addition, comparing the effects of the models displayed in Table 4 reveals that the household’s financial situation does not mediate the relationship between experienced unemployment and a reduction in the frequency of meeting and helping friends and family.

Fig. 4
figure 4

Average marginal effects of unemployed months during the past 2 years on personal relations by gender. (SOEP 1990–2017—own calculation)

Full size image
Table 4 Individual fixed effects of different unemployment durations on personal relations
Full size table

The different conceptualisations of cumulative unemployment experiences plotted in Fig. 4 yield important insights. It is particularly long or multiple unemployment experiences that show sizable effects on close personal relations. Moreover, effects are more pronounced when someone has been unemployed for a longer period over the past 2 years. Interestingly, this is primarily the case for men, as women seem to be more resilient to negative effects of unemployment on their personal relations. This is in line with research by Heyne and Voßemer (2023), who show that women suffer less from unemployment with respect to well-being than men.

Following the processual perspective of accumulation of unemployment over an individual’s working life, Table 9 shows the effects of accumulated unemployment for different age groups (16–30 years, 31–45 years and over 46 years). However, the impact on interactions does not vary significantly with age, suggesting that the accumulation of unemployment can be harmful at any stage during an individual’s life course.

The results displayed in Table 3 and 4 reveal that even the household’s financial situation affects social relations only to a small extent. A decline in monthly household income does not affect social relations. However, perceiving the financial situation as difficult negatively affects personal relations and perceiving an improvement has positive effects. Yet, the mediation analysis shows that, independent of the fact that feeling dissatisfied with the household’s financial situation reduces the frequency of personal relations slightly, the negative effects of unemployment on men cannot be explained by income losses or dissatisfaction with the household’s financial situation. The relationship is also independent of the way cumulative unemployment experiences are measured. These findings do not support the claims made by classical economic and social exclusion arguments laid out in the first hypothesis (H 1).

4.2 Household Composition—How Partnership and Children Impact Unemployment Effects

The results from the previous section show that accumulating unemployment is not negatively associated with social relations for women and that income losses or perceived financial difficulties do not explain the negative effects of unemployment on men. In the following, I analyse explanations concerning social roles and identity with a focus on household composition. The results are displayed in Table 5 and Fig. 5, which illustrate the effects of unemployment by partnership status and the presence of children under the age of 16 in the household. The left-hand plot displays effects for men and women depending on whether or not they have a partner. On the right-hand side, the results indicate effects depending on whether or not the respondent lives with children. Again, the grey dots represent effects on women and black diamonds represent effects on men.

Table 5 Individual fixed effects of cumulative unemployment on personal relations by household composition
Full size table
Fig. 5
figure 5

Average marginal effects of cumulative unemployment on personal relations by gender and household composition. HH household. (SOEP 1990–2017—own calculation)

Full size image

When differentiating the effects by partnership and the presence of children, there are again no significant effects for women. Unemployment does not significantly influence their social interactions. However, even though the effects are not statistically significant, there is a tendency of unemployment to increase the frequency of helping and meeting friends and family. For men, unemployment negatively influences the frequency of social interactions. This hardly changes depending on whether or not they have a partner. Finding a partner reduces the frequency of meeting and helping friends and family significantly (see main effect of having a partner in models 13 and 14 in Table 5). However, the effect of unemployment differs only slightly for men and women when having a partner, whereas the effect is not significantly different for singlesFootnote 8. As the interaction is not statistically significant, the results do not fully confirm the theoretical assumptions expressed in the second hypothesis about partnership status (H 2b). In addition, when differentiating the effects by partnership status, the household’s financial situation does not link unemployment experiences to social relations, contradicting the first hypothesis (H 1).

Literature on gender and family dynamics and the household division of labour claims that it is particularly the presence of children that reinforces traditional gender roles (Grunow 2013). From this perspective, it might be possible that children intensify the gender differences in unemployment effects. The right-hand plot in Fig. 5 inspects this assumption. It becomes visible that, for men, unemployment negatively affects the frequency of meeting and helping friends and family whether they live with children or not, and that this effect increases slightly with the presence of children. For women, there is no significant effect independent of the presence of children in their household. However, the effect of unemployment on women is significantly different from that on men when children are in the household, showing a positive trend, even though the effect of female unemployment remains insignificant.

Taken together, the results only partly support the hypotheses on the dependence of the effects on the household and partnership situation (H 2b). The existence of a partner does not significantly alter the results for men or women. Nevertheless, whether or not children are living in the household does make a small difference. Living with children slightly intensifies the effect of unemployment for both men and women, but in opposite directions. This might be because caregivers, who are often female, reduce their working hours on purpose, or because children make them less prone to withdrawal from social relations. Furthermore, parents often engage with other parents at day care facilities, schools and leisure activities. Hence, the results can be interpreted as pointing to the male provider identity, as there is a slight increase in the strength of the effects when there are children present, and social relations of men seem to suffer in all cases from experiencing unemployment. At the same time, the results show that the effects of unemployment on personal relations for women who are single and living in households without children approximate those of men, which hints at the mitigating effect of being able to care for others as an acceptable social role available primarily to women.

When comparing the models (M13 to M14 and M15 to M16) of Table 5, it becomes clear that the household’s financial situation does not explain the effects of unemployment on social interactions. Instead, the results support the claims made by the second hypothesis at least partially (H 2b), as they indicate the importance of social roles that differ by gender and the presence of children.

4.3 How Family and Work Values Shape the Effects of Unemployment

The next section investigates social identity through values concerning work and family to further explore the explanation of unemployment being a threat to (particularly male) social identities. Therefore, an index measuring the importance of family, partnership and caring as well as an index including the importance of career and self-realisation are included in the analysis (see Table 6).

Table 6 Individual fixed effects of cumulative unemployment on personal relations—work and family values
Full size table

The left-hand plot of Fig. 6 shows the effects of accumulated unemployment depending on how important family, partnership and caring for others are for the individual. For women (grey line), the effect of unemployment stays insignificant independent of how important family, a happy partnership and caring for others are; unemployment does not alter their frequency of interactions with friends and family. For men (black line), the negative effect intensifies the less they care about family, a happy partnership and caring for others. Although this finding runs counter to the idea that, for men, caring for others is related to providing financially for the family, having strong preferences for a happy marriage, caring for others and having children seems to protect them from experiencing a decline in social relations due to unemployment. The results might seem rather small at first sight. However, to reduce complexity and improve interpretation of the interactions with work and family values, the unemployment variable used in these models is the baseline continuous variable. Here, one unit of unemployment equals 1.2 months. The results show that adding up the effects of 3 years of unemployment for men who give the least importance to children, partnership and caring for others leads to a reduction of 2.7 units of the social interaction index, which ranges from 1 to 7. For those men who give a medium level of importance to family and partnership, adding up the effects of 3 years of accumulated unemployment still reduces the frequency of social interactions by 1.6 units. This amounts to almost half the scale of the social relation index for men who highly value family, care and partnership, and a fourth of the scale for those who attribute medium importance to it, which still represents a substantive change in social interactions.

Fig. 6
figure 6

Average marginal effects of cumulative unemployment on personal relations by gender and family and work values. (SOEP 1990–2017—own calculation)

Full size image

The right-hand plot of Fig. 6 shows the effect of unemployment with respect to the importance of a career and self-realisation. In general, these effects are smaller than those that depend on family values. The degree of importance of career and self-realisation does not significantly alter the effects of unemployment, which again remain only significant for men.

One potential explanation for why work values do not change the effect of unemployment on personal relations is that individuals who experience multiple or longer unemployment episodes adjust the importance of work and career downwards to reduce the conflict between their expectations and achievements. This might be less the case for values around partnership, children and caring for others. To test for this possible reverse causality, lagged variables are added to a separate model (see Table 9). If someone who experiences labour market marginalisation adjusts his or her importance of work and self-realisation in reaction to the experienced difficulties, prior values of importance of work moderate the effects of unemployment in a stronger way. The effects for women do not change in direction or significance. For men, the effects of unemployment change the direction only slightly and the confidence bands mostly overlap. Therefore, there might be some adjustment of the importance of work and self-realisation, but not muchFootnote 9.

Overall, the results partially support the claims made by the second hypothesis (H 2c), as family values impact the effects of unemployment on personal relations differently for men than for women, with high values mitigating the negative unemployment effect for men and family values not significantly altering how women’s unemployment affects their personal relations. Yet, work values play only a subordinate role.

5 Discussion of Results

Summing up, the analysis shows that the relationship between unemployment and personal relations is not straightforward, but that the effects of unemployment depend on the duration and quantity of unemployment experiences as well as on a person’s socio-demographic characteristics. This shows that it is fruitful to look beyond simple average effects, as the specific effects differ in direction and strength. The analyses revealed that particularly long-term or multiple unemployment experiences for men negatively affect the frequency of social relations. With respect to the two mechanisms described in the literature, a clear trend emerged in which financial strain had no explanatory power but there were indications supporting the claims made by social identity theory.

With regard to financial strain, the lack of explanatory power of the household’s financial situation stood out. In all models, income loss and the perception of difficulties with the household’s financial situation do not account for any of the effects. However, there may be counteracting forces in place: financial resources received through social policy programs might enhance shame and stigma while at the same time reducing actual decline in disposable household income, which positively affects the household’s economic situation and reduces stigma (Gurr and Lang 2018). Empirically, such effects may cancel each other out and produce the observed pattern. Moreover, meeting and helping friends and family in itself does not need to be costly, so it is reasonable that the cause of a reduction in social interactions is not financial but social.

The findings point to the relevance of social roles and identity in several ways. First, one common thread running through the analysis is that there are almost no significant effects of unemployment on social relations for women. As the basic levels of meeting and helping friends and relatives is almost identical for men and women, it is improbable that gender differences in the data are caused by different baseline levels of social interactions. Instead, contrary to men, women do not reduce their frequency of meeting and helping friends and family when experiencing unemployment. This could well be because their identities are not primarily bound to paid work in a way that would prevent them from reducing their social contacts. Furthermore, women are often responsible within the household for tasks related to interacting with other people, such as childcare and care for grandparents (Luppi and Nazio 2019; Pailhé et al. 2019), tasks that persist independently of their employment situation. Yet, when unemployed, women do not significantly increase their social interactions, as the extra available time and potential need for support could suggest. Instead, the results support findings from other studies, which have found that women allocate extra time owing to reduced working hours to care work and to alleviate multitasking during the week (Pailhé et al. 2019) and that unemployed wives do more housework than unemployed husbands (Gough and Killewald 2011). Conversely, men experiencing unemployment significantly reduce the frequency of meeting and helping friends and family under most circumstances. This is in accordance with results indicating that unemployed men suffer more from lower self-esteem and depression than women do (Álvaro et al. 2019). The same is true with respect to well-being (Heyne and Voßemer 2023).

These findings relate to the “dark side” of social integration described in the introductory article of this special issue. On the one hand, women belong to the groups who benefit less from labour market integration while carrying a higher burden of unpaid work to maintain the current status quo of social integration in society. On the other hand, the findings show that, given women’s marginalized position in the labour market and their additional unpaid societal roles, they experience fewer negative consequences of unemployment than men. This might be a reason why women do not rebel more forcefully against the gender status quo of social integration: they subjectively feel socially integrated, even if this implies facing disadvantages in the labour market (Grunow et al., this issue).

One’s household composition has a comparatively small influence on shaping the effects of unemployment. Having a partner or not does not significantly change how unemployment affects the frequency of meeting friends and family for men or women. However, children intensify the negative effects of unemployment for men, whereas for women we see a positive tendency that is not significant. Yet, for singles and those without children the effects of unemployment do not differ by gender. The effects may be stronger when men live in a household context in which they feel responsible for but cannot fulfil the social role of providing for the family. However, this relationship could not be fully captured by the analyses. Theoretically, there are aspects that could confound results depending on partnership status. Analysing whether consequences of unemployment on social relations depend upon the presence of one close relationship is difficult because the partner’s presence in the household is a social contact in itself that covers at least some part of the need for social interaction. Furthermore, being in a partnership and having children by itself does not necessarily signal a traditional “male breadwinner, female homemaker” model, which strongly depends on norms and values (Grunow and Baur 2014). However, even the models including further information on the partner’s labour market position did not alter the results.

The analysis of individual values regarding family and work attempted to further grasp the explanatory power of social identity. Overall, the effects of unemployment on men’s social relations depended far less on individual work values than on family values. However, the analyses showed that unemployment particularly affects social relations of those men who give a low level of importance to partnership, children and caring for others. Contrary to that, for women the effects of unemployment did not depend on the importance of family and caring. It could be that this insignificant interaction results from different effects for mothers and women without children. In addition to this, the index of family values is not as straightforward as one would wish. Caring for others can mean different things for different social groups and individuals. For some, the importance of partnership, children and caring for others can mean preparing food, washing and doing emotional work, but for others it may mean providing financially and enabling a comfortable lifestyle (Lamont 2000). Nonetheless, the results show that not only social roles related to gender and parenthood, but also individual values around work and family life shape the way in which individuals’ unemployment experiences have the capacity to alter their personal relationships with family and friends.