WHY DO COUPLES RELOCATE?: Considering migration from East to West Germany

One large challenge for couples is that of combining their preferences, wishes, and careers and determining whose wishes regarding migration receive priority. Recent research shows that decisions regarding family migration are typically more influenced by the male partner's career and characteristics (Shihadeh 1991; Bielby and Bielby 1992; Jürges 1998, 2006; Boyle et al. 2001; Smits et al.2003; Nivalainen 2004; Compton and Pollak 2007; Tenn 2010; for a review, see: Cooke 2008b). However, the female partner's employment and income also have an impact, mainly by reducing the likelihood of relocation (Mincer 1978; Jürges 1998; Smits et al. 2003; Nivalainen 2004). More important than the female partner's actual employment are, however, the family's gender norms (Shihadeh 1991), which determine whether the wife's employment is understood as a secondary income with a limited influence on migration (as in traditional couples) or whether the woman is seen as a co-provider whose career receives equal priority (as in egalitarian couples) (Bielby and Bielby 1992; Boyle et al. 2001; Jürges 2006; Cooke 2008a). Gender norms can explain why egalitarian couples not only are less likely than traditional couples to migrate for the sake of the male partner's career (Bielby and Bielby 1992) but are also more likely to migrate to improve the female partner's employment prospects (Boyle et al. 2001).

Although it is well understood how egalitarian or traditional couples decide whether to migrate, it remains unknown if individuals who spent their childhood and adolescence in a socialist country, such as the German Democratic Republic (GDR), whose social system challenged patriarchal gender roles, tend to be more egalitarian when making migration decisions. This article contributes to the research on family migration from East Germany (the former GDR) by examining the 20 years after German reunification to determine whether the migration patterns of East German couples retained the characteristics expected of individuals who grew up in a post-socialist country with an egalitarian ideology, or whether if these characteristics have largely disappeared.

Notably, the high income differences between East and West Germany, combined with low transportation costs, a lack of legal constraints, and the absence of language barriers, provide substantial incentives for couples to migrate. These incentives are not only unique within Germany but are also unusual for any European country.

The following research questions are answered in this paper: (1) How does the migration behavior of married and cohabitating men and women differ from the migration behavior of individuals who live alone? (2) What factors influence family migration? (3) Are there gender-specific differences in the factors that motivate migration?

Hypotheses are developed based not only on gender role theory but also on theories of household economics and bargaining. Data from 1992 to 2007 provided by the Socio-Economic Panel Study (SOEP), which include extensive information pertaining to individuals, their partners, and entire households, are investigated using hierarchical regressions.

Since reunification, more people have left East for West Germany than vice versa. East Germany had lost 7.9% of its pre-reunification population by 1995 and 10.7% by 2000.¹ By 2008, the East German population had declined by 11.7%, or 1.7 million.²

Today, more than 20 years after Germany's reunification, people who leave East for West continue to represent 16% of the migration flows within Germany,³ whereas only 20% of the German population lives in East Germany (West–East migration accounts for 12%).⁴ The characteristics of the individuals who leave East Germany are well documented. These individuals are younger, are better educated, and earn higher wages prior to migration than the individuals who remain in East Germany (Hunt 2006; Brücker and Trübswetter 2007; Windzio 2007; Melzer 2013). Married and cohabitating individuals are less likely to migrate than singles (Hunt 2006; Melzer 2013). Surprisingly, women are more likely to migrate (Hunt 2006; Windzio 2007; Melzer 2013) and are more positively self-selected with respect to education than men are (Melzer 2013). However, less is known about the migration of families that leave East Germany and about how socialization in the former GDR influences migration. The existing research treats male and female migrants as ‘autonomous’ actors, and the characteristics of an individual's partner are not considered to influence migration. Alternatively, the existing research analyzes the migration of male heads of household, and the influence of women on migration is treated as residual (cf. Schwarze and Wagner 1992; Wagner 1992). Because of the economic changes, extensive layoffs, and early retirement associated with reunification, which increased the fraction of women who were the main wage-earners in their households (Diewald et al. 2006), the focus on male heads of household is particularly problematic for East–West migration in Germany (cf. also Acker 1973). Such a framework not only ignores migration by single women but also provides biased results when women's labor market position is higher than that of men.

Neither type of research provides information on the influence of women's characteristics, such as education or seniority, on family migration. To provide this information, the present article analyzes first the influence of both partners' characteristics on household migration for couples who relocate from East to West Germany. Second, the migration patterns of single individuals are compared with those of individuals in partnerships, taking the individuals' partners' characteristics into account.

According to neoclassical migration theory, the decision to migrate can be understood as a risky investment in the future (Sjaastad 1962). Individuals compare their current incomes to the incomes that they are likely to receive after migrating, discounted of the costs of migration. Individuals relocate when they can obtain positive net returns to utility.

The decision-making process for couples as suggested by the household economic theory is similar to the process for single individuals. The only difference is that, for couples, the benefits are maximized across entire households (Mincer 1978). Thus, the costs and benefits of migration for all of the family members must be considered, and migration occurs independently of gains and losses of the individual members, when benefits at the household level can be obtained. Because couples incur greater migration costs, they are less likely to migrate than singles, and dual-earner couples are less likely to migrate than single-earner couples.

The household economic theory ignores potentially divergent individual interests, conflicts regarding the allocation of resources, and internal power struggles (Lundberg and Pollak 2001). In contrast, bargaining theory addresses these shortcomings and models migration decisions based on individual utility. Partnerships are understood as exchange relationships (Ott 1992). Partners trade goods such as love, tenderness, time, and financial resources and continually negotiate the ‘rate of exchange’. The partners invest in the relationship as long as the utility within the partnership is higher than the ‘outside’ or ‘exit’ options available, i.e., higher than the utility associated with being single or in a different partnership. The partner with better ‘outside’ options, the one who would lose less from a breakup and depends less on the partnership, compromises less and invests less in the partnership. This partner also derives more profit from the partnership. The distribution of power between partners depends on financial resources, including jobs, education, and work experience. Employment is important because participation in the labor market is linked to financial independence, better ‘outside’ options, and increased power. In contrast, engagement in housework and care-giving perpetuates financial dependence. For couples with an unequal power distribution (for example, couples with only one employed partner), migration occurs when the partner who holds more bargaining power is willing to relocate.

Power is distributed more equally in a couple when, for example, both partners are employed. Generally, a simple net gain at the household level is insufficient for the migration to take place, as this was the case under the household economic of migration. Migration occurs only when both partners profit individually from relocation: for example, when both partners find better paid jobs at the destination, which is as Mincer (1978) and Kalter (1998) showed statistically very unlikely. Otherwise, the gains at the household level must be high enough for the ‘tied mover’, i.e., the partner who follows, to generate actual utility gains. When the gains at the household level are very high it is actually possible for the tied mover to generate gains from migration despite the relative decline of his or her share associated with a shift of power in favor of the person who initiates migration.

The household economic and bargaining theories are gender neutral. These theories ignore gender-specific differences in the distribution of power and the influence of gender roles on partners’ decisions (Bielby and Bielby 1992). Thus, different influence of men's and women's characteristics on family migration, found empirically (Bielby and Bielby 1992; Jürges 1998, 2006; Boyle et al. 2001; Smits et al. 2003; Nivalainen 2004; Rabe 2011) is explained by women's lower education and employment rates (Markham et al. 1983; Shihadeh 1991). Gender role theory provides a competing explanation and suggests that partners behave according to what they have learned through the socialization process (West and Zimmerman 1987). This theory distinguishes between couples with traditional and egalitarian gender norms. The gender role theory provides similar explanations for the migration decisions of egalitarian couples as already discussed in bargaining and household economic theories. The careers and the employment of both partners are equally valued, and the partners’ roles are interchangeable (Potchek 1997). Relocations take place in order to support both partners’ careers. In contrast, traditional couples assign men the role of the primary wage earner and women the roles of the primary care-giver and housekeeper. Women's careers are treated as inferior to those of men, and the costs and benefits of migration are calculated in a gender-specific manner (Bielby and Bielby 1992; Halfacree 1995). Traditional couples migrate to support the male partner's career, even if this means that women must accept financial losses or career disadvantages.

The GDR, or the former East Germany, was characterized by low institutional differentiation, low poverty rates, and (as in most socialist countries) a high degree of income equality (Pollack 1990). The GDR was a one-party authoritarian state. Work was considered to be a right and a duty. It provided not only financial means but also access to benefits such as leisure, free childcare, and housing.

Gender equality was promoted in the GDR's political agenda and supported by the legal framework and public services. The substantial political interest in women's employment and the political expectation that women should remain in the labor force even if they had small children (Dorbritz and Schwarz 1996) was linked to the GDR's labor shortage, which made it necessary for the entire available labor force to work (Rosenfeld and Trappe 2002: 235). In this climate, women achieved high levels of equality and exhibited exceptionally high rates of labor market participation (Mayer 2006: 31 ff.).

After German reunification, the circumstances changed for women. Conventions from West Germany, such as joint taxation, were adopted (Trappe and Sørensen 2005), and the support for employed women, including the availability of free childcare, was reduced. Average incomes in East Germany began to increase. However, unemployment also increased. In addition, the labor market became highly volatile.⁵ Because of West German recruiting practices (Sandole-Staroste 2002), which treated women as secondary earners and increased competition by employing men in jobs previously held by women (Rosenfeld and Trappe 2002), many women left the workforce and became primary care-givers and homemakers (Rosenfeld et al. 2004: 120). Although these radical labor market changes are typically associated with disadvantages for women, certain women experienced career advantages as a result of the reunification, particularly women employed in the higher service sector (Diewald et al. 2006). Despite these economic issues, East German women remain highly work-oriented (Lück and Hofäcker 2003) and exhibit high levels of participation in education and the labor market (Mayer 2006: 31 ff.).⁶ Moreover, East German men and women express highly egalitarian gender views (Kreyenfeld and Geisler 2006; Matysiak and Steinmetz 2008), which implies that migration patterns among East German couples should also be egalitarian.

To test whether the migration patterns among East German couples are indeed egalitarian, two sets of hypotheses are used. The first set (E) describes the behavior of egalitarian couples based on the bargaining and household economic theories, whereas the second set (T) describes the behavior of traditional couples with reference to gender role theory. It is suggested that egalitarian couples in which only one partner, either male or female, is employed are more likely to relocate from East to West Germany than are couples in which both partners are (un)employed (Hypothesis 1E). The high mobility of couples with only one partner employed can be explained by the employed partner's power to initiate migration, relatively low migration costs and the couples' good access to financial means. For couples in which both partners are unemployed a new job should provide a large increase in the household's means and increase both partners’ utility. However, migration implies monetary costs and couples with both partners unemployed have only limited resources to invest, which should limit their migration. Finally, when both partners are employed, exceptional income increases are necessary to cover relocation costs because migration implies that one partner must quit his or her job, at least temporarily. Therefore, it can be expected that couples in which both partners are unemployed are more likely to migrate than couples in which both partners are employed (Hypothesis 2E).

Additionally, a couple's migration decisions will be influenced by the partners’ education levels. It is suggested that egalitarian couples in which one partner, either male or female, is more educated than the other partner are more likely to migrate than couples in which the partners have equal levels of education (Hypothesis 3E). This hypothesis is based on the expectation that partners with similar educations are likely to block one another's migration plans and that therefore couples with unequal education levels should be the most likely to migrate. Nevertheless, because individuals with higher education levels are generally more likely to migrate (cf. Sjaastad 1962), couples in which both partners are highly educated are more likely to migrate than are lower-educated couples (Hypothesis 4E).

In general, for egalitarian couples, both partners’ characteristics are expected to influence migration decisions symmetrically. However, if East German couples behave traditionally, the effects are expected to be asymmetrical: only the employment of the male partner should influence the likelihood to migrate, whereas women's employment should show no influence on migration (Hypothesis 1T). Moreover, it is suggested that only the male partner's education level should influence the couples migration decisions. Couples with a highly educated male partner are more likely to migrate than are couples with a less-educated male partner (Hypothesis 2T).

This study uses data for East Germany from 1992 to 2007 from the SOEP.⁷ The SOEP is a representative longitudinal survey of private households in Germany and has covered East Germany since 1990 (Wagner et al. 2007).

The first estimates are based on data that include singles and individuals in partnerships (27,599 person-years for 4227 men and 28,635 person-years for 4247 women). The second set of estimates focuses on couples who live in joint households, whether they are married or cohabitating. Couples who live in separate households, whether because they are divorced or because they are simply not living together, are excluded from the analyses.⁸ After singles, couples who live apart and couples in which at least one partner is older than 64 years of age or younger than 18 years of age⁹ are excluded, the sample includes 2898 couples (19,314 person-years). The migration rate for all of the households that met the selection criteria is 3.2%. Because the analysis is based on a small number of migration events, the missing values for the independent variables are imputed using single equations.

To ensure the comparability of migrants and non-migrants, I measure all of the features of the couples (or individuals) who moved to West Germany based on the last year that the couple (or individual) lived in East Germany. All of the periods that follow the relocation are omitted to ensure that all of the observed changes occurred prior to the migrations and that the causes and the consequences of migration are not confused.

The data suffer from one shortcoming: despite the high incentives for East–West migration in Germany, household relocation is rare, and the total number of instances of relocation in the sample is small (98). However, these data are the only available data that offer longitudinal information on migration by East German households.

Individuals are classified as migrants when they move from East to West Germany. The interviews are performed at the new residence. The couples were defined as migrants only when the partners in the couple relocated simultaneously.¹⁰

The operationalization of the independent variables (Table 1) differs partially for the individual and household levels. At the household level, it was important to include the characteristics of both partners in the analysis. To avoid multicollinearity, the combined characteristics of both partners were used.

In longitudinal surveys, such as the SOEP, individuals are interviewed on multiple occasions, which creates a hierarchical structure of data in which observations from several time points are ‘nested’ within households. Time-variant and time-stable information can be obtained from longitudinal data. Hierarchical logistic regressions, which are used here, make it possible to consider both types of information. They control for the fact that households are more similar to themselves over time than they are similar to other households; that is, for the dependence of the likelihood of migration over time.

In addition, hierarchical logistic regressions do not require the same number of observations for all of the households included in the analysis. Thus, using this approach makes it possible to analyze unbalanced data and include households in the analyses that exist for different periods of time. Because the duration of partnerships is highly dependent on the age of the partners and because younger couples are more likely to migrate, a method that employs balanced data would lead to the underestimation of the number of couples at ‘risk’ of migration and thus of an underestimation of the likelihood of migration.

In particular, I use random-effects hierarchical logistic regression because it relies on between-subject and within-subject variance and is more suitable than fixed-effects hierarchical regression for analyses of data for small groups or a small number of events (Snijders and Bosker 1999: 43 ff.) In addition, the Hausman test indicates that the random-effects estimates are consistent and can be used. Using random-effects regressions, it is possible to analyze the impact of changes within households on migration over time. For example, the influence of one partner's becoming unemployed can be estimated. In addition, the variation in time-stable characteristics across households: for example, whether couples with unequal education degrees are more likely to migrate than couples with equal education degrees.

In this study, households are allowed to vary in terms of their intercepts. A separate intercept is estimated for each household that is included in the data, and these intercepts are allowed to deviate from the overall mean.

Table 2 provides an overview of the characteristics of migrating and non-migrating individuals, regardless of whether they have a partner, and for couples. Compared with non-migrants, migrants are younger, are more likely to be single, and spent less time at their last job. Individuals who migrate with their partners tend to be younger than their non-migrating counterparts. Men who move to West Germany with their partners are better educated and earn more than non-migrant men in partnerships, whereas women who migrate with their partners spent less time at their last job and are more likely to be unemployed than non-migranting women.

In Table 3, the migration behavior of individuals who relocate alone and thus have no constraints is compared with the migration behavior of individuals in partnerships, who must consider the regional preferences of their partners.

For both men and women, living with a partner reduces the likelihood of migration regardless of whether a couple is married or cohabitating. Although the effect seems to be slightly stronger for cohabitation, the differences are not statistically significant. Thus, the two groups can be combined in further analyses. The effect of living with a partner is weaker for men than for women. Relationship status is the most influential factor for migration among women. For men, having children or owning a house decreases the likelihood of migration even more.

For women, the effect of a partnership remains stable or increases when the partner's income (Model 3) and education (Model 4) are controlled for. The partner's income has no effect on the migration of women in partnerships, and the partner's education has only a weak positive effect. The differences in the migration patterns of single and partnered women cannot be explained by women's partners’ characteristics. It appears that for women, the mere presence of a partner is decisive, and women's migration patterns change as soon as they start a partnership. Consistent with gender role theory, women seem not to migrate to benefit their own careers when they have partners.

For men, the partnership dummy variable loses power and significance when partner's characteristics and especially income are considered. The lack of significance of the partnership dummy variable in Model 3 indicates that after the partner's income is controlled for, the migration behavior of single and partnered men no longer differs significantly. The differences in single and partnered men's migration behavior can thus be traced to their partners’ characteristics. Additionally, the strong negative impact of the partner's income indicates that women in partnerships use their power to prevent migration. Every additional €100 earned by a female partner decreases the likelihood of migration by 12%. These findings are consistent with those of recent research (cf. Mincer 1978; Jürges 1998; Smits et al. 2003; Nivalainen 2004) and with the household economic and bargaining theories. The higher the woman's income, the higher the migration costs and the financial obstacles to relocation, the higher is also woman's power to prevent migration.

Research at the household level (Table 4) shows that couples in which the woman is unemployed are more likely to migrate, as is assumed under bargaining theory and hypothesis 1E (Couples with one employed and one unemployed partner, regardless of the gender of the unemployed partner, are the most likely to migrate). However, this conclusion is not true for couples with unemployed men, and thus hypothesis, 1E must be rejected. In circumstances in which the woman should have the power to determine migration, the bargaining mechanisms appeared to be overruled.

Because couples in which both partners are unemployed are not more willing to migrate than couples in which both partners are employed, hypothesis 2E must also be rejected. In fact, the investigation of the employment constellations within households reveals no egalitarian patterns. However, the results do not support gender role theory either and contradict hypothesis 1T, which suggested that only the man's employment circumstances influence the likelihood of migration because the results show that the employment status of female partners is relevant.

The education of the male partner had the strongest effect on couples’ migration. Couples in which the male partner is highly educated (i.e., has completed more than 11 years of school or training) and the female partner has only obtained a modest education are twice as likely to migrate than are couples in which both partners have obtained a limited level of education (Model 4). Couples in which both partners are highly educated are also more likely to migrate than are couples in which both partners are modestly educated. Couples in which the female partner has a higher level of education than the male partner do not differ from couples with both partners modestly educated. These results contradict hypothesis 3E, derived from bargaining theory, which suggested that both partners’ education levels would be similarly influential. The female partner's education appears to be of secondary importance because couples in which the male partner is highly educated are the most mobile regardless of the female partner's education level. This pattern is exactly the pattern suggested for traditional couples in hypothesis 2T, which was derived from gender role theory, and mirrors the results of earlier research (cf. Nivalainen 2004; Compton and Pollak 2007; Tenn 2010). However because couples in which both partners are highly educated are more likely to migrate than couples in which both partners are less educated, hypothesis 4E is also supported.

A result that might seem unusual at first is that the presence of children in a household does not affect the migration decision for couples in East Germany. One reason for this phenomenon could be the high selectivity of the sample; only couples who live together are considered, and this group is most likely to have children. In addition, the results of other studies on this issue have also been mixed. For example, Smits et al. (2003) find negative effects of children on migration regardless of the children's ages. Rabe (2011) and Nivaleinen (2004) find no effects for younger children, and Jürges (1998) finds no effects for household size, which indicates the number of children in a household, in most of his models.

The purpose of this study was to investigate the migration decisions of couples who were socialized in a post-socialist country. The study focused on the influence of both partners’ education and employment status on migration from East to West Germany, controlling for the characteristics of the household and the sending regions. Hypotheses were derived from the gender role, household economics, and bargaining theories. The investigation started at the individual level, with the identification of gender-specific differences in the migration patterns of men and women. Next, the influence of household constellations on the migration decisions of couples was analyzed. SOEP data from 1992 to 2007 and hierarchic logistic regression models were used.

Consistent with previous findings for East and West Germany, the results show that men and women in partnerships are less likely to migrate (e.g., Jürges 1998; Kalter 1998; Hunt 2006; Melzer 2013). However, the mechanisms that generate this outcome are gender specific. The differences in single and partnered men's migration patterns can be explained by the characteristics of the men's partners. It appears that for men, rational calculations based on costs and gains determine whether migration occurs; this pattern persists even when men have a partner. Thus, for relocation to occur, the gains must be much higher for partnered men than for single men because the partnered men must compensate for both partners’ migration costs; otherwise, their female partners can prevent migration. Women, in turn, appear to abstain from migration for the sake of their own careers when they are in a partnership. Their partners’ characteristics are of secondary importance, as the involvement in a partnership is the deciding factor. Thus, the behavior of women is consistent with the assumptions of gender role theory. The findings at the household level support this view. The influence of men's partners’ characteristics (particularly their education level) on couples’ migration decisions is larger than the influence of women's characteristics. Of the female partners’ characteristics, only unemployment appears to be important for migration decisions; unemployment makes relocation more likely. This phenomenon generates an environment in which men can realize at least some of their migration goals if they have good qualifications and the migration gains are high enough to compensate for the migration costs of both partners. However, under similar circumstances, women are not likely to initiate their migration.

Although the migration patterns found in this study tend to support the gender role theory, it is impossible to reject the bargaining or household economic theories because elements of these theories also find support. Moreover, it appears that all three theories provide insight and that couples base their decisions on traditional gender views and rational costs–benefits calculations.

Additional factors that appear to make it easier for non-single men to realize their migration goals are occupational segregation, which is strong in East Germany (Rosenfeld and Trappe 2002), and the ubiquity of women's jobs, which make women less geographically restricted (Morrison and Lichter 1988). These factors facilitate the employment search for women at the destination, decrease women's migration costs, and enable men to compensate for their partners’ migration-related losses more easily. However, these factors reinforce not only the position of women as tied movers but also their position as secondary earners, weakening their future bargaining position in the household as well as their future positions at the labor market (cf. Halfacree 1995).

Although socialization within the GDR tended to support gender equality, couples do not behave in an egalitarian manner when making migration decisions; indeed, our results are similar to those for traditional couples in West Germany (cf. Jürges 2006). Although women in East Germany are highly career-oriented (Lück and Hofäcker 2003), their career orientation appears not to influence their migration decisions. The lack of egalitarian behavior when East German couples make migration decisions might indicate that the high labor-market participation of East German women is not driven mainly by egalitarian views but instead by the necessity for both partners to work. Thus, East German women in partnerships should reduce their labor-market participation once they arrive in West Germany. In addition to exploring changes in employment patterns following migration for women in partnerships with quantitative research methods, for future research it might be promising to explore with qualitative methods why post-migration employment patterns change, and how the immigrants' understanding of gender norms changes after the migration. Finally, it would be interesting to analyze how integration into a new society contributes to those changes.

Migration is based on compromises and women in East Germany compromise and accept a suboptimal position as tied movers despite their high qualifications and labor-market attachment (cf. also Shihadeh 1991; Boyle et al. 2001 for other countries). It appears that the need to find a compromise reproduces the traditional patterns of gender inequality in households and the labor market, as suggested by Halfacree (1995) and the socialization in a socialistic system has not enough influence on people's decision finding to counterbalance this mechanisms of gender inequality.

I would like to thank Hans-Peter Blossfeld and Herbert Brücker for their advice and support.

Silvia Maja Melzer is a Postdoctoral Fellow at the Collaborative Research Center ‘From Heterogeneities to Social Inequalities’ of the University of Bielefeld in Germany. Her research interests are mainly in the area of migration and happiness. In her research she examines the causes and consequences of the migration from East to Western Germany after the reunification in 1989. Currently she is expanding her work to include work organizations and is participating in a project on ‘Interactions between Capabilities in Work and Private Life: A Study of Employees in Different Work Organizations’.