Marriage migration and women’s entry into the German labour market

Objective: We analyse the employment patterns of childless first-generation migrants to Germany. In particular, we focus on the behaviour of female “marriage migrants”. Marriage migrants are defined as individuals who married after their spouse had moved to Germany. Background: Demographic studies have illustrated that marriage migrants have particularly high childbirth rates upon arrival. There is, however, little empirical evidence on how the childbearing behaviour of migrant women is related to their employment behaviour. Method: We use event history techniques to study women’s labour market entry after migration in relation to their childbearing behaviour. We draw on data from the German Socio-Economic Panel (GSOEP). The analytical sample is restricted to immigrant women who moved while childless to Germany between 1990 and 2016 (n=981). Results: Compared to other groups, marriage migrants have very low chances of entering the labour market. Only 32 per cent of the migrants in our sample had ever participated in the labour market in the five-year period after their arrival in Germany. A large share of the differences between these migrants and other migrants can be attributed to the socio-demographic composition of these women, and to their tendency to transition to parenthood soon after their arrival. Conclusion: We argue that the low employment rates of female marriage migrants must also be viewed in the context of Germany’s migration policies, which do not provide many routes for female third-country nationals to move to Germany. One of the few available channels is that of marriage migration. We conclude by discussing the social policy implications of these findings at a time when Germany is gradually becoming a dual-earner society.


Introduction
There is a large body of literature on the employment behaviour of migrants.While many of the studies on this topic have focused on male migrants (see, e.g., Uhlendorff & Zimmermann 2014;Fertig & Schurer 2007;Kogan 2011Kogan , 2007Kogan , 2004;;Velling 1995) or have compared women and men (e.g., Salikutluk et al. 2020;Krieger 2019), relatively few of these studies have focused on women.The literature that has analysed migrant women's behaviour consists primarily of demographic studies that mainly investigated female migrants' family behaviour, in particular their childbearing and marriage patterns around migration (e.g., Wolf 2016;Lievens 1999).This paper seeks to bridge these two strands of the literature by examining how childbearing, marriage, and migration relate to immigrant women's post-migration employment behaviour. 1 The theoretical starting point of our investigation is the classical model of family migration, proposed by Mincer (1978).This classical framework -which distinguishes between tied movers and tied stayers -has regularly been employed to study the employment behaviour of internal and international migrants (e.g., Cooke 2001;Lersch 2016).We critically discuss this framework, and propose a typology that better reflects the patterns of female migration to European countries, including to Germany.We distinguish between three groups of female migrants: a) single migrants, b) marriage migrants, and c) spousal migrants 2 .Single migrants are women who were single at the time of migration.Marriage migrants are women who were married at the time of migration, and are in a marriage that was contracted after the male partner had migrated to Germany (see also , Mohn 2019;Lievens 1999;Wolf 2016). 3Spousal migrants are women who were married at the time of migration.Different from marriage migrants, the marriage was contracted before either of the partners had migrated.In drawing a distinction between spousal and marriage migrants, we seek to understand how the "sequencing" of marriage and migration affects women's employment behaviour.
The data for our analysis come from the German Socio-Economic Panel (GSOEP).The analytical sample includes women who were childless at the time of migration and who moved to Germany between 1990 and 2016.In a first step of the analysis, we examine the socio-economic composition of the three abovementioned groups of migrants (by region of origin, age at migration, level of education, prior work experience).In a second step, we employ event history techniques to study the length of time it takes migrant 1 Migration and marriage are taken as "given" in our framework.One could model migration, marriage, and childbearing as related processes.We refrain from this strategy to avoid having to use overly complex multiprocess modelling techniques. 2 Scholars have also labelled this group "family reunion migrants" (Wolf 2016).3 There are no official statistics on the share of marriage migrants.The visa statistics of the foreign office contain information on the share of visas that were issued on the grounds of spousal reunion.However, this information is regarded as incomplete, as the statistics only include visas that have been issued at foreign embassies.Furthermore, not all of the issued visas result in a migration (BAMF 2020).The German Central Alien Register documents the legal grounds for migration, including "family reunion with a spouse".The drawback of these data is that they cover third-country nationals, but they do not fully cover migrants from other regions.Data from the German Central Alien Register from 2015 suggest that the large majority (more than 80 per cent) of third-country nationals who migrated on the legal grounds of "spousal reunion" to Germany are female (BAMF 2017).
women to enter the German labour market.In the final part of the analysis, we look at the role that childbearing plays in the labour market entry rates of migrant women, and how the patterns of single migrants, spousal and marriage migrants differ.This paper contributes to the literature in several ways.It provides novel evidence on the employment and family behaviour of first-generation migrant women in Germany.Furthermore, it is, to the best of our knowledge, one of the few papers that has used a dynamic framework to examine the employment behaviour of female migrants and how it relates to their fertility behaviour (e.g., Wood & Neels 2017;Kil et al. 2018).However, our investigation has several limitations.First, we do not employ a rigorous causal approach.Instead, the analysis is largely descriptive, and compares the employment and childbearing patterns of key groups of female migrants (single migrants, spousal migrants, and marriage migrants).In addition, our analysis is restricted to women who were at risk of entering parenthood at migration, and, thus, to women who were childless before migration.As a result, we cannot generalise our findings to all migrant women.Although marriage migration is far more prevalent among female than among male migrants, it should also be emphasised that men can be marriage migrants as well.The question of how male marriage migrants perform on the labour market is underresearched, and is not addressed in this investigation.

Migration to Germany
Germany has one of the largest immigration flows in the world (Diehl 2016;OECD 2018).In the 1960s and 1970s, migration to Western and Northern European countries, including to Germany, was heavily dominated by labour migration (Van Mol & de Valk 2016).At that time, most of the migrants were adult males.While the West German government also recruited couples and women to work in Germany, pregnancy and childbirth around the time of migration were not tolerated (Mattes 2005). 4After 1973, when the active recruitment of migrants by the West German government ended, migration flows to (West) Germany changed radically, and the share of migrants who moved on the grounds of family reunion increased in tandem with an increase in the share of migrants who were female (Gonzáles-Ferrer 2007, see also Sprengholz et al. in this volume).At the beginning of the 1990s, migration flows to Germany were dominated by ethnic German migrants from Central and Eastern European countries, apart from the large influx of refugees to Germany from the Yugoslavian wars (Bade & Oltmer 1999).European citizens who have the right to live and work in other European Union countries have also contributed substantially to the flow of migrants to Germany, especially since the onset of the global financial crisis in 2008, and since the Eastern enlargement of the 4 The former East Germany recruited foreign laborers from other socialist counties.Compared to West Germany, the East German government recruited laborers on a smaller scale, and more rigorously prohibited family formation among migrant women (Bade 2004).
European Union (Van Mol & de Valk 2016;Geis 2017).The growing influx of migrants from the EU, and from Central and Eastern Europe in particular, has also contributed to an increase in the share of skilled female migrants (Kogan 2010).
For third-country nationals, the routes for entering Germany have always been much more limited, although the migration options for highly qualified migrants have gradually changed.In 2000, a "green card" initiative for migrant workers in the fields of information and communication technology was introduced.In 2012, the EU Blue Card Directive for highly skilled immigrants was implemented, with Germany being at the forefront of this development (Cerna & Chou 2014).Because the focus of the directive is on recruiting workers for occupations in male-dominated fields with labour shortages, such as IT and engineering, only very small fractions of these permanent residence permits for highly skilled workers have been issued to women (Cerna & Czaika 2016;Kofman 2012).However, in recent decades, a large share of the residence permits granted in Germany have been to third-country nationals for the purposes of family reunion, and women have been overrepresented in this group (Henkel et al. 2015;Grote 2017;BAMF 2020).Thus, family migration is one of the main channels through which female third-country nationals have moved to Germany.In a small percentage of family reunion cases, minor children are reunited with their parents, or elderly parents move in with their children.However, until recently, the overwhelming majority of "family migration cases" have been for the purposes of spousal reunion; i.e., an adult migrant is (re)united with her/his spouse (Wälde & Evers 2018;BAMF 2020).In these cases, the marriage may have occurred before one or both of the partners migrated to Germany.To qualify to bring a spouse to Germany, the foreign applicant cannot be dependent on social welfare, and has to guarantee that s/he can provide sufficient living space and means of subsistence for him/herself and his/her family members (German Residence Act, Section 6).Thus, the economic dependence of the spouse on the breadwinner is already implicitly assumed in the legislation.In 2007, a new requirement that third-country nationals have basic German language knowledge was implemented with the aim of facilitating the spouses' integration after arrival.Furthermore, some exceptions exist for spouses with residence permits linked to high earnings (Grote 2017).

Labour and family policies
The legal regulations that govern the employment of migrants in Germany have undergone radical changes in recent years.EU migrants and ethnic German migrants have had immediate access to the German labour market, while the options for thirdcountry nationals to work in Germany have been more limited.Before 2005, a thirdcountry national who entered the country as a family migrant and had a non-German spouse was subject to a waiting period during which s/he could not access the labour market, and was thus economically dependent on her/his partner.It was not until 2005 that these family migrants were given the same legal rights to work as their spouse (Peers et al. 2000;Kreienbrink & Rühl 2007).
Moreover, since 2005, major social policy reforms have been enacted that seek to increase women's employment participation by making it easier for mothers to combine family and work responsibilities.In particular, public day care services have been significantly expanded, and an income-related parental leave scheme was introduced in 2007 (Hipp, Bernhardt, & Allmendinger 2015).While these measures apply to all legal residents in Germany, regardless of their nationality or country of origin, migrants may not be able to take full advantage of them.For example, parental leave benefits are income-related, and a woman's employment and earnings one year prior to giving birth determine the amount of the parental leave benefit payments she receives.Migrants are less likely to benefit from these payments because they are generally less likely than natives to be employed before they have children.Moreover, a migrant from a non-EU country is at a disadvantage, as the income she earned in a "third country" is not taken into account in the calculation of the parental leave benefit payments (Office of the German Federal Government 2020).
In terms of their access to publicly provided child care, migrants generally have the same rights as natives.In 2013, the right of parents to enrol their child (aged one year or older) in public day care was implemented at the national level in Germany.While access to child care is now a legal right, the parents need to apply for a day care slot through their local municipality, and the application procedure can often be bureaucratic.This may be a barrier for migrants, particularly if they have just arrived in Germany.Indeed, empirical evidence suggests that the children of recent migrants are less likely than other children in Germany to attend day care (Krapf 2014;Bujard et al. 2020).Another reason why migrant families may be less likely than native families to use child care and parental leave is that female migrants have relatively low labour market participation rates.Overall, it is clear that the main focus of parental leave and child care policies in Germany is supporting dual-earner families.As first-generation migrant families are often organised as single-earner households, they are less likely to benefit from these measures.

Migration patterns and employment
Much of the prior research on the labour market performance of female migrants has been based on Mincer's (1978) model of family migration.This model views migration as being motivated by the gains in income that are associated with relocation.In contrast to classical economics, which focuses on individual gains and losses, Mincer's (1978) model conceptualises migration as a decision that maximises the utility of the household unit.This model provides an explanation for why a person may choose to relocate even "though his (or her) 'private' calculus dictates staying" (ibid.: 751).The model is allegedly gender neutral, but was primarily designed to explain the "lower market earnings, and a diminished migration payoff for the wife" (ibid: 753).
Mincer's model has been criticised for focusing narrowly on the economic benefits of migration (Cooke 2008); for neglecting how gender role attitudes moderate migratory and employment patterns (Shihadeh 1991;Bielby and Bielby 1992); for ignoring sub-optimal outcomes due to non-cooperative behaviour (Lundberg and Pollak 2003); and for promoting a simplified view of what constitutes a household unit (Boyle et al. 2001).
Nevertheless, the model has been used as a reference in numerous empirical studies that have examined gendered migration patterns and the employment behaviour of female international and internal migrants.For example, Cooke (2001) analysed internal migration in the US based on this model, and found that the labour market participation of "trailing wives" dropped after a move.However, he also emphasised that the effect was stronger for women who had children shortly after the relocation.Lersch (2016) focused on the migration patterns and the employment behaviour of married dual-earner couples.Using data from the UK, he showed that the partner's gender roles attitudes were an important factor in why women left their employment in their region of origin to follow their partner.Based on data from the German Socio-Economic Panel (GSOEP), Krieger (2019) examined the behaviour of international migrants who had a residence permit that allowed them to seek employment immediately after arrival.She distinguished between migrants who were "leaders", "equal migrants", and "tied migrants" (operationalised based on a subjective assessment of who played the decisive role in the migration decision).The results of the analysis suggested that leaders of both genders were more likely to enter the German labour market than tied migrants (those who's partner played the decisive role in the migration decision) or equal migrants (who were equally involved in the decision making process).

Marriage migration and the role of migration policies
The "Mincer model" of female migration did not distinguish between residential mobility and international migration.It implicitly assumed that both could be approached with a similar "toolkit".However, residential moves differ greatly from international migration.Most importantly, international moves are strongly influenced by migration policies.The migration policies in a country set the rules and regulations that enable or hinder people from crossing international borders.By defining the pathways for entering the country, these policies indirectly define the composition of the migrant population in terms of both their skills and their family status (Beck-Gernsheim 2011; Borjas & Bronars 1991;Glick & Park 2016;White & Johnson 2016;He & Gerber 2019).Policies such as the BLUE card attract high-skilled migrants and individuals working in shortage occupations."Family reunification" is a legal pathway to entry that does not discriminate based on profession, but that requires the migrant to have established family ties to a person in the country of migration.For female third-country nationals (TCN), family reunification is the major legal channel through which they can migrate to Germany.Given their limited options for legal migration, TCN migrants are more likely to be married at the time of migration than non-TCN (e.g., European citizens) migrants.In addition, TCN migrants are more likely to be in a marriage that was contracted "transnationally"; i.e., the woman married a man who had been living in the country of destination for several years.
Thus, it could be argued that while "followers", "leaders", and "stayers" are meaningful categories when distinguishing between different types of internal migrants, they are less appropriate when applied to international migrants.When examining international migration, a comparison group of "stayers" is often lacking (see, however, e.g., Guveli 2015).Another reason why distinguishing between followers, joint movers, and leaders may not be meaningful in this context is that the overwhelming majority of female international migrants are either joint movers or followers.A more important demarcation line for international migrants is whether a marriage was contracted before or after the (male) partner migrated.The term marriage migrant is often used in demographic research to refer to an international migrant who married an individual who was already living in the destination country (Aybek et al. 2015;Wolf 2016;Kraus 2019;Mohn 2019).However, the term is not yet commonly used in research on the employment behaviour of international female migrants.This is surprising, as female international migrants who can be described as marriage migrants obviously have certain characteristics (in terms of country of origin and human capital endowment) that may shape their labour market outcomes.
This argumentation leads to clear predictions regarding the composition of the migrant population.More, specifically it leads us to make the following assumptions about the characteristics of marriage migrants, spousal migrants, and single migrants: Hypothesis 1a: We assume that marriage migrants, spousal migrants, and single migrants differ greatly in terms their country of origin.As third-country nationals (TCN) have few channels to migrate to Europe other than through marriage migration, we assume that TCNs are heavily over-represented in the group of marriage migrants.

Hypothesis 1b:
We also assume that marriage migrants, spousal migrants, and single migrants differ greatly in terms of their human capital endowments (education and work experience).Single female migrants are positively selected for their labour market skills.Thus, we assume that single migrants enter the country with higher human capital endowments than married female migrants.Among the married migrants, we assume that the marriage migrants are the least likely to have acquired education or work experience, as they are primarily selected based on their marital ties to a person in the country of destination.

Childbearing after migration and subsequent employment
As well as defining "marriage migrants", demographic research has paid particular attention to the childbearing behaviour of migrants, and how it evolves in relation to the length of time migrants have spent in the destination country (Anderson 2004;Toulemon 2004;Kulu 2005;Milewski 2007;2010;Kulu & Milewski 2007;Ortensi 2015).These studies have uncovered a pattern of the "interrelation of events".For most female migrants, and particularly for female marriage migrants, the likelihood of having a first birth increases in the period immediately following a move (e.g., Wolf 2016; Kraus 2019).Scholars have argued that this pattern may "reflect poor employment prospects of marriage migrants" (Kulu & González-Ferrer 2014, pg. 422).From an objective perspective, having children may be regarded as irrational if an individual's labour options seem bleak, and if childbearing would further limit the woman's future employment chances.From an individual perspective, childbearing in such a situation may still be rational, because motherhood may be regarded as a "biographical alternative" that structures an otherwise uncertain life course (Friedman, Hechter & Kanazawa 1994).Scholars have also argued that the acceleration of childbearing after migration may be related to traditional family values, which are particularly prevalent among marriage migrants (Kontos 2007;2009;Merali 2008;Brettell 2017;Baykara-Krumme 2009).Still others have argued that this pattern can be viewed as a logical chain of events, with childbearing representing a natural next step after marriage (Wolf 2016).Regardless of which of these arguments are correct, they all suggest that having children shortly after migration is a common behaviour, particularly among female marriage migrants, and that it lowers a woman's chances of entering employment after migration.
Hypothesis 2: In line with prior empirical demographic evidence, we assume that marriage migrants, spousal migrants, and single migrants differ in their childbearing behaviour.We assume that marriage migrants are especially likely to have children around the time of migration, while these patterns are less pronounced among spousal migrants, and do not apply to single migrants.
We further argue that these patterns of behaviour, together with the different socioeconomic characteristics of marriage migrants, spousal migrants, and single migrants, will have direct effects on the likelihood of entering employment for the three groups: Hypothesis 3a: Differences in the transition rates to first employment between marriage migrants, spousal migrants, and single migrants can be partially attributed to differences in terms of country of origin, skills (education and prior work experience), and childbearing behaviour.
Hypothesis 3b: Based on the assumption that having children before entering employment can be particularly detrimental for marriage migrants, who are the least likely to be prepared to enter the labour market upon arrival, we assume that there will be interaction effects between the migration group and parental status.

Data and methods
The data for this analysis come from the German Socio-Economic Panel (Release 35). 5  Our analytical sample consists of female respondents who were childless and between 18 and 49 years old in the year of their last migration to Germany.This age range was chosen to ensure that these women had migrated after completing their basic schooling, and were not beyond of childbearing age.We also chose to observe the behaviour of childless women, as we were interested in comparing marriage migrants -most of whom entered the country after marrying -with other women who were at a similar stage of their life course.Furthermore, we restricted the analysis to women who migrated between 1990 and 2016.Finally, we omitted respondents with missing childbirth and employment histories; women who moved for humanitarian reasons; and women who were in tertiary education in the year after migration, as they may have migrated for reasons of education.We also eliminated respondents who moved to live with a German citizen, as well as the small group of women who were married at the time of migration, but who moved without their husband.Finally, there were several cases for which we were unable to reconstruct the marital history or the migration date of the partner, as well as a few cases in which the woman moved after her marital union had dissolved.We excluded these cases from the analysis as well.This left us with 981 respondents, of whom 651 were single and 330 were married at the time of migration.Among the migrants who were married, 108 were spousal migrants and 222 were marriage migrants. 6

Dependent variable and method
The dependent variable is the employment rate.The process starts at the time of migration and ends at the first labour market entry.Each case is censored 10 years after migration or at the date of the last interview, whichever came first.The data are available to the accuracy of a year.We consider each respondent's employment status after the first year following migration to ensure that the status refers to the woman's entry into employment in Germany, and not to a previous employment spell in the country of origin.
If the respondent reported engaging in several activities in one year, we prioritise employment over the other activities.As a descriptive method, we estimate survival functions using the life table method.In the regression analysis, we employ a complementary-log-log link binomial regression (Allison 1982).

Independent variables
The main variable of interest is the woman's migration pattern.We distinguish between women who were and were not married in the year of migration.We further differentiate between married migrants based on their marriage-related migration patterns.We use three years as a cut-off point to differentiate between groups, which is in line with the operational definitions used in previous studies (Lievens 1999;Wolf 2016).Thus, we distinguish between: (1) Women who were single at the time of migration. 6 The original sample contained female 3448 migrants who moved to Germany between 1990-2016, were aged 18-49 at migration, and did not have refugee status at the time of migration.This sample was further reduced when we limited it to women with valid birth histories (3437 left).We then restricted the original sample even further by only including women who were childless at the time of migration, leaving us with 1814 cases.From these cases, we have deleted respondents with invalid information in their marriage and employment histories, leaving us with 1511 cases.Next, we omitted those respondents who were divorced or widowed at migration (1505 left).We then excluded women who were in education in the first year after migration (1266 left), women with very short durations of less than one year (1251 left), cases in which the partner's migration background or immigration year was invalid (1216 left), women who moved to join a German partner (1019 left), and cases in which a man was following his wife (993 left).Finally, we omitted cases in which the woman was married at migration but had changed partners after migration, resulting in a final sample size of 981 respondents.
(2) Women who were married at the time of migration, and moved at the same time or up to two years after their (foreign-born) husband had moved to Germany.We refer to this group as "spousal migrants".
(3) Women who married and then migrated to join their (foreign-born) husband who had been living in Germany for at least three years before the marriage and the migration of the wife.We refer to this group as "marriage migrants".
To construct the variable, we rely on the retrospectively collected migration and marital histories.To match each respondent with her partner, we use information on the partner identifier that is included in each wave in the German Socio-Economic Panel (GSOEP) after migration.The few cases in which the respondent changed partners over time in Germany were excluded from the sample.
A key variable of interest is the origin of the women.We distinguish between two origin groups: free movement and TCNs.These groups correspond to the different pathways for entering Germany that were available to the women at the time they migrated.The free movement group is composed of women from the EU and other privileged nations who were able to migrate based on the freedom to move for employment.Most of these women were from Western or Southern Europe, while smaller proportions were Central and Eastern European migrants who moved after 2011 or 2014 (depending on the country) following the enlargement of the European Union, or who were of ethnic German origin.The group of TCN migrants is further divided into regions of origins: Central and Eastern Europe, Africa and the Middle East, and the "rest of the world".These women had more limited options for employment-specific migration.Many migrated on the legal grounds of family reunion.
We control for standard human capital endowment covariates.Education is measured as the respondent's maximum registered level of education at the time of the interview.We distinguish between those individuals who had low or no education, medium education (vocational level), and high education (university level).Because we excluded the respondents who were enrolled in education immediately after arriving in Germany, we can assume that the educational levels of the women in our sample did not change much after they migrated.We also control for years of labour market experience before migration by adding up the years reported by the individuals in their retrospective employment histories.
Furthermore, to account for the changes in migration policies that have occurred over time, we use a categorical migration period covariate that distinguishes between migrants who moved in the 1990-2000, 2001-2006, and 2007-2016 periods.We have selected these cut points because in 1990, a new Foreigners Act, which signalled further openness to foreigners residing Germany, went into effect.In 2000, the first green card system for TCN workers was implemented.The passage of the new Immigration Act in 2005 made it easier for the spouses of foreign workers to gain access to the labour market.In 2007, the German language knowledge requirements were increased for individuals migrating on the grounds of family reunification.Age at migration is entered as a time-constant and continuous covariate.We furthermore control for childbirth as a time-varying covariate.

Sample statistics by migration group
Table 1 displays the composition of the sample by migration group.As expected (see hypothesis 1a), we find large differences by origin group.Most single women and most spousal migrants migrated from European countries.At the other extreme, almost none of the marriage migrants came from a country of origin with freedom of movement: i.e., although some came from a non-EU European country, most were from Africa or the Middle East.Marriage migrants also differed in their skill sets (hypothesis 1b).They had lower human capital endowments than the other groups: half of the women in this group had no degree, while almost half of the women in the other groups had a university degree.Moreover, compared to the other groups, the marriage migrants had, on average, lower levels of labour market capital and fewer years of work experience before migration.This may be in part because these migrants tended to be young at the time of migration.The table also shows that since 2007, single migrants and spousal migrants have become much more common, while marriage migrants have become less common, probably due to the tightening of regulations; i.e., the requirement that migrants who seek to enter on family reunion grounds show proof of basic German language skills (Liebig 2007).As expected, we find that single migrants entered the labour market rapidly, while spousal migrants entered later.Nonetheless, after five years, 70 per cent of the spousal migrants had entered employment.In contrast, marriage migrants did not enter employment until much later: i.e., five years after migration, only around 30 per cent of these women had participated in the labour market.

Childbearing behaviour by migration group
In order to shed light on the childbearing patterns by migration group, Table 2 displays the person-years at risk of labour market entry, broken down by whether the respondent had a child or not.In the year of migration, all respondents were childless. 7However, large discrepancies emerged between the groups as the duration of their stay in the country increased.Among the migrants who were single at the time of migration, the share of those who had children increased only gradually.While this is in part because many of these women were single (meaning they had no co-residential partner) before they migrated 8 , it is also because many of the women who were single at the time of migration entered the labour market before they had a child. 9When we focus on the spousal migrants, we see that after the third year of their stay in the country, only 30 per cent of their "exposure time" was spent childless, compared 16 per cent for the marriage migrants.Thus, the table clearly reveals the stark differences in the childbearing strategies of the groups, including in how they timed childbearing around migration.These findings support hypothesis 2, and, in turn, previous research that documented the strong acceleration of first birth risks among marriage migrants.Figure 2 approaches the topic from another perspective and conceptualises childbirth and labour market entry as competing events.The figure shows that only 25 per cent of the marriage migrants first entered the labour market before they had a child, while more than 70 per cent had a child before they entered the labour market.The remainder 7 Note that the year of migration does not enter the event history model.We displayed this year in this table for illustrative purposes to show that all migrants enter the sample as childless individuals.8 A separate investigation in which we included marriage as a time-varying covariate showed that women who were single at the time of migration and who married thereafter were a small group, and were as likely to enter the labour market as those who remained single for the observed time.9 Once they enter the labour market, we do not observe them anymore.
(around five per cent) had neither entered the labour market nor had a child in the fiveyear period following their arrival in Germany.

Regression results
Table 3 reports the results from the event history models on the entry into first employment after migration.We have estimated a stepwise model and displayed the relative risk (hazard ratios).Model 1 only includes the process time (time since migration), the period of migration, the age at migration, and the group membership (single at migration, spousal migrants, marriage migrant).Model 2 also includes the region of origin.In Model 3, labour market capital endowment variables are included.Model 4 also accounts for having a child before entering employment.The stepwise modelling approach allows us to determine whether the previously described differences in employment rates can be attributed to the women's marital status, or to their country of origin, socio-economic characteristics, and childbearing behaviour.
Model 1 shows that the hazard rates were very high in the year immediately after migration, declined thereafter, and then increased slightly after five years of stay in Germany.Taking the first year of migration as a reference, the employment rates declined by 90 per cent in the second year.In the third through the fifth years, the employment rates were 86 per cent lower than in the first year.After the fifth year, the employment rates increased again, but were still 75 per cent lower than in the first year. 10These findings suggest that the women who did not enter the labour market in the first year after migration had reduced chances of ever entering the labour market.As we expected to find given the changes in migration and employment regulations, the model also indicates that the employment rates of migrant women had increased over time, especially since 2007.Moreover, the model's results corroborate the descriptive findings from Figure 1 by showing that spousal migrants had employment rates that were in-between those of single migrants and marriage migrants.The employment rates of marriage migrants were 53 per cent lower than those of spousal migrants.
In Model 2, we control for the region of origin of the women.The results show that women who enjoyed the right of free movement had higher employment rates than all of the TCN groups.Compared to women in the reference category who came from Central and Eastern Europe (but were not ethnic Germans), women who came from countries with freedom of movement had an employment risk that was, on average, 82 per cent higher.Women in the "TCN: African and Middle Eastern group" had the lowest employment rates.Indeed, their employment rates were 34 per cent lower than those of women in the "TCN: CEE group".Including this covariate explains a large share of the differences between spousal migrants and marriage migrants.
Model 3 accounts for human capital endowment (education and work experience before migration).The results show that having work experience prior to migration had a significant positive effect on employment rates.In the same vein, we see that having a vocational degree or a university degree was associated with increased employment rates.We should note that after we included human capital endowment, the direction of the coefficient of age at migration changed.This means that the positive effect of age at migration was largely due to older women having better education and more work experience, which, in turn, positively affected their employment participation rates.Net of these factors, we find that age had a negative effect on employment rates after migration.Accounting for human capital covariates further narrowed the differences in employment rates between marriage migrants and spousal migrants, but it increased the differences between single migrants and other migrants.Thus, human capital endowment was a "suppressor".More specifically, if we consider that many of the single migrants were fairly young at the time of migration, and therefore had less work experience than the spousal migrants, they were still more likely to enter employment.
In the final Model 4, the differences between the migration groups are further explained by looking at the compositional differences in their childbearing behaviour.Having a child after migration reduced the risk of entering employment by 69 per cent.Furthermore, childbearing behaviour explained a large portion of the differences between single migrants and spousal migrants, as well as between spousal and marriage migrants.As we expected given the elevated birth transitions of marriage migrants upon arrival, the remaining differences between these two groups ceased to be significant.Overall, we find that there is no single and dominant socio-demographic characteristic that explains all of the differences between the three groups of women.Instead, we must conclude that country of origin, skills, and childbearing behaviour all contribute to reducing the differences between the three migrant groups.Thus, hypothesis 3a is supported, but without giving priority to a single explanation.

Differences between childless women and women with children
In the final model, we examine in more detail the differences in the employment rates of the different migration groups, and depending on whether the women in these groups had a child.Thus, our aim is to determine to what extent having a child after migration affected the employment rates of the different groups of women.Figure 3 shows the margins (average predicated probabilities) of the interaction model.We find that there are differences between childless women and those who became mothers.A steep gradient and differences between single and married women remain, even when the women were childless.However, for women in all of the groups, having a child had a similar effect.Thus, we have to reject hypothesis 3b, which stated that childbearing is more detrimental for marriage migrants.Childbearing is always consequential for women's employment, at least in the short run.The main difference between the three groups appears to be that the marriage migrants were more likely than the other women to have children before entering the labour market (see Figure 2).This behaviour was detrimental for future employment, regardless of the migration pattern.

Discussion and conclusions
In this paper, we have examined migrant women's transitions to employment after their arrival in Germany.We identified three groups of migrant: those who were single at migration, marriage migrants (women who married after the male partner had migrated), and spousal migrants (couples who married before either of the partners had migrated).We argued that the three groups of women differ greatly in terms of their country of origin and socio-economic characteristics.These differences are closely related to the migration channels that are available to them.Migration policies in European countries (and in other industrialised countries) make it difficult for third-country nationals (TCNs) to enter.As marriage is one of the few routes to migrate to Europe for TCNs, marriage migrants often come from "third countries".The findings of our empirical analysis, which was based on data from the GSOEP, supported this view.Among the single and the spousal migrants, the shares of women from third countries were relatively small.Conversely, a large fraction of the marriage migrants were from third countries.In particular, women from Africa and the Middle East, but also from Central and Eastern Europe (before the EU accession), were heavily represented in this pattern of migration.
The further empirical analysis uncovered large differences in the employment entrance rates of the migration groups.More than 75 per cent of the women who were single at migration and 55 per cent of the spousal migrants entered employment in the first year after migration.The group who stood out as having exceptionally low employment rates were the marriage migrants: i.e., five years after migration, less than 35 per cent of these women had ever entered the labour market; and even after 10 years in Germany, more than 50 per cent of them had never held paid employment.Some of the differences between these migrant women can be explained by their countries of origin and the limited access they had to the labour market.For example, many marriage migrants came from third countries, and arrived during periods when the employment opportunities for third-country nationals were very limited.Levels of education and work experience before migration also explain some of the differences in employment between the migration groups.Beyond the differences in human capital endowments, having children shortly after migration, which was a particularly common behaviour for the marriage migrants, was a major factor that strongly delayed labour market entry.These women often delayed seeking employment or did not look for it at all, as the "prime mover" was also the male breadwinner who was already well established in Germany when they arrived.
Our finding that employment rates were low among married migrant women is not novel.Giving birth after migration may be seen as a logical step in the family formation process when marriage and migration are strongly interrelated.However, having a child shortly after migration -and, thus, before getting established in the host country's labour market -will have repercussions for a woman's chances of ever entering the labour market.While this relationship may be obvious, these findings are rarely discussed together, even though they have clear social policy relevance.Having a child before entering the German labour market means that the mother has less access to the incomerelated parental leave schemes, and, in turn, that migrant families with young children will have lower incomes than dual-earner families.As the dual-earner family is gradually becoming more common in Germany, single-earner families face increasing economic risks.Thus, low female labour market participation rates can have a range of long-term consequences for the economic and social well-being of female migrants and their families.Finally, non-working migrants are less able to integrate into German society via paid employment, and may also be prone to social isolation (Kontos 2009;Kontos & Schinozaki 2007).There is, moreover, an interplay between women's employment and child care usage.A number of studies have shown that the children of non-working mothers are less likely to start day care at an early age (Krapf 2014).Thus, the failure to integrate migrant women into the labour market will have implications for the integration of immigrant children into German society.
Our analysis provided social policy-relevant findings on a group of female immigrants to Germany that is still relatively large.However, our investigation had to leave a number of questions unresolved.Most importantly, we were unable to include micro-level information on women's work orientation.We controlled for work experience and educational level before migration, but using a more direct measure of gender role attitudes or work orientation would have been helpful to capture the values and attitudes that guide women's employment choices.The great strength of the GSOEP is that it provides the full migration and marriage histories of both partners.However, among the GSOEP's limitations are that it does not provide fine-grained information on the evolution of partnerships (for example, if female marriage migrants were already partnered with the future husband when the man had migrated).Moreover, we had no information on the bargaining process behind the migration decisions.Qualitative studies are better able to address such issues, as they can elucidate the conditions under which marriage migration occurs (see, e.g., Aybek et al. 2015).Furthermore, although we controlled for compositional differences in observable characteristics and behaviour after migration, there was further unobserved heterogeneity that we were unable to account for.This particularly pertains to access to networks that may enable or hinder entry into the labour market.Many third-country nationals who arrive as low-skilled migrants also belong to large ethnic networks, which may further delay their labour market integration by, for example, delaying the need to acquire language skills or to create social ties to members of the "majority population"; i.e., ties that could be helpful in finding a job (Lancee & Hartung 2012).Finally, we have studied only the transition to first employment after migration.The next step would be to take a more holistic approach by studying complete employment trajectories after migration, and differentiating between different types of employment, such as part-time, full-time, unemployment, and non-employment.Ergebnisse: Heiratsmigrantinnen weisen im Vergleich zu den anderen Gruppen extrem niedrige Übergangsraten in den Arbeitsmarkt auf.Nur 32% haben nach fünf Jahren den Übergang in eine Beschäftigung erfahren.Unterschiede zwischen Heiratsmigrantinnen und anderen erklären sich zu einem großen Teil durch die großen soziodemographischen Unterschiede, die zwischen den Gruppen existieren, wie auch durch den schnellen Übergang zum ersten Kind nach Migration.

Figure 1
Figure 1 displays the survival curves for the entry into first employment after migration.As expected, we find that single migrants entered the labour market rapidly, while spousal migrants entered later.Nonetheless, after five years, 70 per cent of the spousal migrants had entered employment.In contrast, marriage migrants did not enter employment until much later: i.e., five years after migration, only around 30 per cent of these women had participated in the labour market.

Figure 1 :
Figure 1:Probability that the respondents had not yet entered the labour market (life table estimates)

Table 1 :
Descriptive statistics of time-constant covariates by migration group, column %

Table 2 :
Person-year distribution by migration group and childless/not childless if they have not entered employment, column %

Table 3 :
Results from complementary logit models: Relative risks of entry into first employment after migration, all childless women