Ethnic Variation in the Link between Women’s Relative Employment Positions and Entry into Parenthood in Belgium

Abstract

This study investigates the association between women’s relative employment positions and the transition to parenthood, focusing on women of Maghrebi, Turkish, and Southern European origin in Belgium. Whereas gender specialization is associated with higher chances of entering parenthood in the older literature, the economic preconditions to parenthood have shifted and more recent studies indicate that couples where both partners work are more likely to start a family. However, whereas this shift has been extensively studied among majority populations, we lack insight into whether similar patterns can also be found among population subgroups with a migration background. This paper uses Belgian census data from 2011–2015 to explore how women’s relative employment positions are linked to the likelihood of entering parenthood and whether this association varies by women’s age, generation, and origin of the male partner. The results indicate that couples in which both partners are employed are more likely to enter parenthood regardless of migration background. However, the results for women of Maghrebi or Turkish background suggest that single-earner couples and couples where both partners are unemployed delay entry into parenthood to a similar extent.

1. Introduction

The association between women’s relative labour market positions and the transition to parenthood has shifted substantially over the past decades. During the 1950s and 1960s, specialization in (male) paid work and (female) unpaid work was positively associated with entry into parenthood in most Western European countries [1,2]. Recent studies indicate that this association has gradually shifted toward dual-earner couples being most likely to enter into parenthood [3,4]. This change is often explained by increasing levels of female labour market participation, decreasing real wages, rising economic insecurity [5,6], changing gender roles [7,8], and the increasing availability of work-family reconciliation policies [9,10]. Whereas the link between couples’ division of employment and fertility is well-documented for the majority population, the association has not been studied to the same extent among population subgroups with a migration background. This is remarkable given that the ever-growing population with a migration background is often found to differ from non-migrant groups in terms of fertility patterns [11,12] as well as gendered positions and prospects in the labour market [13,14,15].

This study adds to the existing literature by examining the association between women’s relative employment positions and entry into parenthood among women of Maghrebi, Turkish, and Southern European origin in Belgium. Most European contexts, with Belgium as a prime example, are characterized by substantial migrant-native employment and income gaps and particularly large gender employment gaps among non-European origin groups [13,14,15]. In addition, and possibly as a result of this variation in employment opportunities, more gendered divisions of labour within the household are not challenged to the same extent among migrant populations of non-European origin [16,17,18]. Limited access to and use of policies that facilitate the combination of work and family life may further discourage (stable) labour market participation [19,20]. As a result, the association between women’s (relative) employment positions and entry into parenthood may vary between population subgroups depending on their migration background. To date, this variation has remained largely unexplored.

Using Belgian census data on childless women living with a partner in 2011 and prospective information on first births for the period 2012–2015 from the population registers, we study the link between women’s relative employment positions and first births specifically among women of Belgian, Turkish, Maghrebi (Morocco, Tunisia and Algeria), and Southern European (Italy, Spain, Greece and Portugal) origin. This paper contributes to the literature in three ways. First, we uncover population heterogeneity in the association between women’s relative employment positions and entry into parenthood. While a wide range of literature on the employment-fertility link has primarily focused on majority populations, we examine whether the economic preconditions to parenthood vary between migrant populations characterized by distinct opportunity structures and sociocultural settings. Second, we go beyond the study of women’s employment positions as an individual-level indicator by focusing on women’s relative employment positions within the couple. Several recent studies have uncovered differences by migration background in employment trajectories around motherhood [21] and individual-level associations between employment and having a first child [22,23,24]. The absence of information on the partner’s employment positions is often indicated as a limitation given that lower levels of labour force participation of one partner can be offset by better labour market prospects of the other partner [25]. Including both partner’s employment positions is particularly relevant when studying couples with a migration background. The economic positions of couples are expected to vary substantially as a result of persistent native–migrant and gender gaps in the Belgian labour market [26] and increasing variation in partner choices [27,28,29]. Previous studies have also shown a high level of selectivity in partner choices concerning socioeconomic characteristics [28,30] and gender role expectations [31,32]. Third, this study includes women of Southern European origin alongside Maghrebi and Turkish origin women. Compared to non-European origin groups, the socioeconomic and sociocultural context of European origin groups is often less well-documented. We argue that including Southern European origin groups is of particular relevance in the Belgian context, given similar initial migration patterns yet widely varying opportunity structures in the current Belgian labour market [33,34,35,36].

2. Migrant Populations in Belgium

2.1. Migration History

To revitalize the Belgian economy after the Second World War, the Belgian government recruited large groups of Southern European, Turkish, and Maghrebi—predominantly Moroccan—labour migrants throughout the 1960s and the early 1970s. After labour migration was limited in 1974 and 1976, Turkish and Moroccan migrants often settled permanently in Belgium. This first wave of labour migration was predominantly male and focused on labour market participation. The settlement of Turkish and Moroccan immigrants was regionally concentrated and resulted in local communities with strong ties to their regions of origin [33,37]. After the stop of labour migration, immigration for Maghrebi and Turkish-origin groups was predominantly limited to the reunification of families already residing in Belgium or family-forming migration as unmarried migrant workers married a partner from their home country. This second migration wave was predominantly female and family-focused. As the children of the settled labour migrants—often called the “second generation”—have started entering young adulthood, family-forming migration has persisted through the marriage of Turkish and Maghrebi second-generation men and women to partners living in their countries of origin [30,38]. Whereas the number of marriages between a Turkish and Moroccan young adult of the second generation and a partner from their origin country has been gradually declining, the majority still choose a partner from the same origin group [27,28,29,30]. As a result, family-related migration was still the main migration motive for men and women of Moroccan (65%) and Turkish (46%) origin in 2020. A much smaller percentage migrated to Belgium in the context of education (9% of Moroccan and 10% of Turkish immigrants) or labour (4% of Moroccan and 15% of Turkish immigrants) [39].

In contrast to the Maghrebi and Turkish origin groups, the return migration of Southern European labour migrants was more prominent after labour migration was halted because of economic growth in their origin countries [36]. Previous studies have shown that Southern European first and second generations are less likely to choose a partner from their country of origin [35]. Whereas a substantial amount of Italian (31%) and Spanish (45%) immigrants still migrate to Belgium because of family-related reasons, work is a more prominent migration motive among Southern European immigrants (48% of Italian and 40% of Spanish immigrants) compared to the Maghrebi and Turkish origin groups [39].

2.2. Socioeconomic Position

The Belgian labour market is characterized by substantial inequalities in employment and income positions between majority and minority populations [26,40]. Particularly non-European origin groups such as those with a Turkish and Maghrebi background are more often in unemployment, unstable employment trajectories, and short-term employment contracts and are overrepresented in lower-wage categories [15,41]. Non-European minority populations are also more likely to encounter prejudice and discrimination in the labour and housing markets [42,43,44,45]. Differences with the Belgian majority population are smaller among the second generation compared to the first generation but remain substantial [13]. The labour market position of people with a Southern European background is generally better and can be positioned slightly below that of Belgian natives [34]. In addition to a migrant-native gap, the Belgian labour market still shows a gender gap in employment [34,41]. While the gap between men’s and women’s labour positions and wages is relatively limited among those with a Belgian or Southern European background, the gender gap is larger among Turkish and Maghrebi groups. As a result, women of non-European origin are in a particularly precarious position in the Belgian labour market and potentially have more limited prospects for improving their labour market position.

Despite the findings that some migrant groups are more likely to occupy a more precarious position in the Belgian labour market, migrant communities can also play an important role in providing support and facilitating employment. Particularly among established migrant populations characterized by structural migration mechanisms, such as Turkish and Maghrebi communities in Belgium, social networks based on kinship, friendship, and community ties are vital to exchange information (e.g., job or housing opportunities) as well as provide social, emotional, and financial support [46]. Consequently, the main source of social and material support for migrants is often the informal social network [47].

2.3. The Division of Paid Work

Female educational attainment and labour force participation have been rising since the 1970s. Consequently, the division of paid work within couples has shifted and an increasing number of couples adhere to a dual-earner model where both partners are employed [48]. While a detailed examination of the within-couple division of unpaid work is beyond the scope of this paper, it is important to note that the rise in female labour force participation and dual-earner couples is not necessarily matched by a similar shift toward a gender-egalitarian division of unpaid work [49,50]. In addition, the division of paid and unpaid work has been shown to vary across family transitions, particularly after childbearing [51,52,53]. This section focuses specifically on how the division of paid work is expected to vary by migration background among childless couples.

Given the higher levels of unemployment and employment instability, we would expect to find a higher proportion of women of Maghrebi or Turkish origin in couples where one or both partners are unemployed, particularly when both partners are of the same origin. The more disadvantaged position of non-European origin women is expected to reflect on the division of paid labour within households. Women of Maghrebi or Turkish origin in particular are expected to live in male breadwinner households more often compared to the other origin groups. Alongside these socioeconomic elements, the gendered division of paid work is also affected by sociocultural factors such as gender role attitudes and selective migration mechanisms. Given a more persistent gender gap in labour market participation among non-European origin groups, gender role divisions in paid and unpaid work may not have been challenged to the same extent compared to native Belgian and European origin groups. Previous studies indicate a higher preference for a gendered division of tasks among the Turkish and Moroccan first and second generations in Belgium [16,17,18] and other European countries [17,54,55,56,57]. However, gender role expectations are generally found to vary by gender and partner choice. Several studies have shown that the choice of higher educated second-generation Turkish and Moroccan women to marry a partner from their country of origin can be motivated by more gender egalitarian goals [31,58]. Given that local second-generation men are regarded as “too traditional”, a higher educated and possibly more progressive partner is preferred. In contrast, the choice of a second-generation Turkish or Maghrebi man to marry a first-generation partner of the same origin often confirms the male breadwinner and female caretaker roles given the more limited employment opportunities of first-generation non-European women [16,58]. Whereas these sociocultural elements are broadly studied among Maghrebi and Turkish origin groups throughout Europe, similar studies on the gender role attitudes of Southern Europeans are currently lacking in the literature.

These differences point toward a higher likelihood of gendered specialization in paid and unpaid work among Maghrebi and Turkish origin groups in Belgium, particularly among couples involving a female first-generation migrant. Previous research by Huschek, de Valk, and Liefbroer [16] in Germany, Austria, France, the Netherlands, Switzerland, and Sweden finds that second-generation Turks with a native partner have the most gender-egalitarian division of work. Couples consisting of a second-generation Turkish man and a first-generation Turkish woman have the most gendered division of work.

2.4. Entry into Parenthood

The research on fertility patterns of immigrant populations has predominantly focused on the link between migration and fertility [12]. Among first-generation women, first-birth patterns are found to be strongly shaped by the reason for migration [11]. Particularly among immigrant women of non-European origin, who more often migrate in the context of family formation or reunification, the likelihood of entering parenthood is higher after migration. Among first-generation women of European origin, who migrate mainly within the context of study and work, lower first-birth levels are found immediately after migration. Compared to native Belgians, Kulu et al. [59] find higher first-birth risks among the Moroccan, Turkish, and Italian first generation after controlling for birth cohort and age group. After controlling for education, the differences diminished for Moroccan and Italian first-generation women while remaining substantial for Turkish first-generation women.

Considering the descendants of immigrants, the fertility behaviour of European origin groups mostly resembles that of the native population, whereas more substantial differences remain among non-European second-generation women. In Belgium, several studies indicate that Moroccan- and Turkish-origin women, in particular, enter into parenthood at earlier ages compared to the native Belgian population [12,60]. The study by Kulu, Hannemann, Pailhé, Neels, Krapf, González-Ferrer, and Andersson [59] indicates that first-birth risks are somewhat higher for the Turkish second generation but not for the Moroccan or Italian second generation while controlling for age and cohort differences. However, first birth risks become more similar after controlling for educational level.

The likelihood of entering parenthood is also found to depend on the composition of the couple. Among first-generation Turkish and Moroccan women who married a second-generation partner, the different transitions of marrying, migrating, and entering parenthood at relatively young ages are strongly interrelated [12,60]. First-birth rates are also found to be lower among Turkish and Moroccan second-generation women who marry a native Belgian partner compared to those who choose a same-origin partner [61].

3. Theory and Hypotheses

3.1. Division of Paid Labour and First Births

The second half of the 20th century was characterized by rising levels of female education, labour force participation, and wage potential. In addition to changing women’s societal position, rising levels of female education and employment also lead to the postponement of entry into parenthood and affect within-couple dynamics. Whereas the male breadwinner female carer model had been strongly entrenched in society until the 1970s, this traditional gender role division became increasingly challenged as a result of women’s rising employment opportunities and men’s decreasing real wages. As a result, a wide range of literature has become focused on the gendered link between employment and entry into parenthood.

From a microeconomic point of view, Becker’s [62] New Home Economics assumes that couples in which partners specialize in paid and unpaid work are most likely to enter parenthood. In addition to the financial resources required to bring up a child, the transition to parenthood is also associated with opportunity costs given that partners have to take time out of the labour market and forego earnings to take up childcare. Whereas the opportunity costs of having children are gender neutral in theory, Becker poses that specific aspects of childrearing (e.g., pregnancy and breastfeeding) cannot be outsourced to the male partner. As a result, the likelihood of having children is expected to be lower when women are employed compared to couples where male and female partners specialize in paid and unpaid labour, respectively. The New Home Economics theory applies the mechanism of opportunity costs to explain the trend of declining fertility levels observed since the 1970s. The increasing levels of female education, labour force participation, and wage potential implied that a gradually increasing number of women had to forego earnings and employment opportunities to have children.

More recent research indicates that the stability and certainty of labour market trajectories as well as the opportunity costs of having children are likely to be contextually determined. When the New Home Economics theory was first developed during the 1980s, policies aimed at reconciling work and family life were mostly absent and gender equality in household tasks was limited [48]. In such a context, the opportunity costs of having children for working women are particularly high. However, the changing economic and normative context observed since the 1980s and throughout the 21st century requires a revision of the assumption that specialized couples are most likely to enter into parenthood. As labour market uncertainty increased and men’s real wages decreased, supporting a family on one income has become increasingly difficult and single-earner couples have become particularly vulnerable to job loss or unstable employment of the main earner [5,6,7]. In addition, the opportunity costs of childbearing for working women have declined as caretaking roles are decreasingly gendered within the household [8] and work-family reconciliation policies have become more widely available [9]. Due to the financial stability provided by the dual breadwinner model and its increasing compatibility with parenthood, having two earners is increasingly regarded as an important precondition to entering parenthood [63]. Recent studies largely confirm the hypothesis that labour force participation increases the chances of entering parenthood for both men and women among majority populations across Europe [25,64,65,66,67,68,69]. Several studies throughout Europe find that higher labour force participation and income of both partners are associated with a higher likelihood of becoming first-time parents [67,68,69]. However, studies that take both partners’ employment positions into account simultaneously are rare.

In addition to the role of opportunity costs in microeconomic theory, there is a wide body of literature that emphasizes the importance of sociocultural aspects such as norms and values. According to the Second Demographic Transition Theory, the increasing economic possibilities for both partners induced a shift in values toward gender egalitarianism, individual autonomy, and self-realization [70,71]. These changes initially clashed with the traditional and gendered expectations that were prevalent and promoted by traditional institutions such as the church and the state until the mid-20th century. As a result, for couples, and particularly women who wanted to invest in education and enter the labour market, entry into parenthood was increasingly postponed to later ages. It is important to note that these microeconomic and sociocultural factors are potentially endogenous. On the one hand, investments in education and gaining a stable labour market position before parenthood are likely to depend on the anticipated continuation of this labour market position when having children. Having children has a profound impact on a couple’s division of labour and particularly women have been found to decrease their work intensity after entering parenthood, even when income differences are limited [52,72]. Women who anticipate reducing working hours or leaving the labour market after entering parenthood may not invest to the same extent in a better labour market position through education or other career investments before family formation. On the other hand, attitudes may alter and different roles may become accentuated in light of socioeconomic prospects. When people are faced with substantial uncertainties about their futures in the labour market, family life, and parenthood, in particular, can present as an alternative pathway to reduce this uncertainty [73] and they may find fulfilment and purpose in roles outside of the labour market [74,75]. In this case, couples dealing with ongoing instability might not wait to become parents until they both occupy a stable employment position [25,76].

The Belgian context is characterized by high overall levels of male and female labour force participation and wide availability of work-family reconciliation policies. The native Belgian population, in particular, has good prospects to attain a stable labour market position regardless of gender, which would, in turn, facilitate access to work-family reconciliation policies, lower the opportunity costs of entering parenthood, and reduce the gains of gender specialization in the household. The general theoretical framework on employment and entry into parenthood has not yet taken population heterogeneity in terms of migration background into account, which will be elaborated on more broadly in the next section. We develop the following hypothesis:

Hypothesis 1.

We expect that women of Belgian origin in a dual-earner couple will be most likely to enter parenthood, followed by women in single-earner couples or in couples where neither partner is employed.

3.2. Variation by Migration Background

In general, a situation where both partners participate in the labour force increases the probability that couples have the necessary financial resources to start a family regardless of migration background. However, the degree to which couples have good prospects to find these stable and well-paying labour market positions is expected to play an important role in the couple’s division of labour. Within the Belgian context, Southern European origin groups are found to occupy a socioeconomic position that closely resembles that of the native Belgian population [26]. In addition, the gender gap in employment is relatively limited among population subgroups with a European background. Hence, despite the findings that unemployment levels are somewhat higher among European-origin groups, entering the labour market and attaining a dual-earner model are realistic goals for women of Southern European origin. For women with a Southern European background, it is a realistic option to view the dual-earner model as a precondition to parenthood and delay having children until this precondition is fulfilled. As a result, we propose the following hypothesis:

Hypothesis 2a.

Among Southern European origin women we expect the dual-earner model to be associated with the highest likelihood of entering parenthood while women in couples with one or no earners are expected to have lower first birth risks.

However, other population subgroups within the Belgian context, such as Maghrebi and Turkish origin groups, are characterized by distinct opportunity structures and sociocultural settings. Consequently, there are several reasons to expect a differential link between the division of paid work and the transition to parenthood among women of Maghrebi or Turkish origin. Concerning differing opportunity structures, the higher levels of unemployment and employment instability found among Turkish and Maghrebi origin groups limit their economic agency to a higher extent compared to Southern European origin groups and Belgian natives [15,26,40,41]. Due to a more substantial gender gap [34,41], women of Turkish and Maghrebi origin have more uncertain employment prospects relative to their male partners. Difficulties in finding stable employment may also lead to more limited access to Belgian family-reconciliation policies such as parental leave where eligibility typically depends on having a stable labour market position [19,77]. Rather than delaying entry into parenthood until both partners have attained economic security, motherhood may present itself as an alternative pathway toward certainty and fulfilment [73,74] among women with a Maghrebi or Turkish background who face more persistent economic uncertainty compared to Southern European and Belgian women. This leads us to the following hypothesis:

Hypothesis 2b.

We expect that the differences in the likelihood of entering parenthood between different breadwinner models are more limited among women of Maghrebi and Turkish background compared to women with a Southern European background and Belgian women.

In addition to differences between non-European and European origin groups, labour market prospects and preferences may also differ substantially by generation and partner choice, especially within Maghrebi and Turkish groups. Attaining a dual-earner model may be particularly difficult for couples involving a Maghrebi or Turkish first-generation female partner, at least partly due to selective marriage migration mechanisms. In contrast to Southern European immigrants, immigration from Maghreb countries and Turkey is predominantly related to the family domain [78] and prospects of entering the labour force are often limited. A study by Timmerman and Wets [79] indicates high levels of unemployment and a weak labour market attachment among Turkish marriage migrants. In addition, the choice of second-generation Maghrebi and Turkish men to marry a female migrant is potentially aimed at confirming the male breadwinner role [16]. Combined with the particularly distinct gender gap in employment of Maghrebi and Turkish groups, first-generation women of Maghrebi and Turkish background may be particularly unlikely to postpone entry into parenthood until both they and their partners have obtained a stable labour market position. In general, second-generation Maghrebi and Turkish women occupy a better socioeconomic position in terms of education and employment levels. However, they may have been socialized within households with a more gendered division of paid work given that the migration patterns of their parents strongly linked male immigration to labour force participation and female immigration to family formation. In case similar expectations remain prevalent among the second generation, women of Maghrebi and Turkish origin may anticipate leaving the workforce after entering parenthood and may not invest in labour force participation to the same extent as Belgian couples. As a result, we hypothesize:

Hypothesis 3a.

The link between the division of paid labour and first-birth risks will be weaker among first-generation compared to second-generation women.

Regarding differences in the partner choices of women, couples consisting of two Maghrebi and Turkish partners are expected to experience more economic vulnerability and less favourable labour market prospects compared to couples where a Maghrebi or Turkish woman has partnered with a man of Belgian origin. Hence, we expect that labour market preconditions to parenthood will be eased to a greater degree among endogamous couples compared to mixed couples with a Belgian male partner. In addition, previous studies indicate that the choice of a Belgian partner is highly selective in terms of socioeconomic characteristics such as higher educational levels which would also be associated with the labour market prospects of these women [28,30]. We develop the following hypothesis:

Hypothesis 3b.

We expect the link between the division of paid labour and first-birth risks to be weaker among women who have partnered with a same-origin partner compared to women with a Belgian-origin partner.

4. Data and Methods

4.1. Data

This study uses a prospective research design, linking data on origin, household characteristics, and employment from the Belgian 2011 census to birth histories recorded for the years 2012–2015 in the population registers. Given that we focus on the link between within-couple employment dynamics and the transition into parenthood, the sample for 2011 is limited to childless women between 18 and 50 years old who are in a co-residential partnership and are not enrolled in education. The focus on how paid work is divided between men and women also implies a selection of heterosexual couples. The LIPRO-typology1 is used to identify married and unmarried cohabiting couples [80].

To compare women’s relative employment position and the link to first births by migration background, we select women of native Belgian origin and women of Southern European (Greece, Italy, Portugal, Spain), Maghrebi (Morocco, Tunisia, Algeria), and Turkish origin. Origin and generation are determined based on the woman’s first known nationality and birthplace as well as both parents’ first known nationality. If an individual’s and both parent’s first known nationality is Belgian, then they are assigned to the Belgian origin group. An individual is of the first generation if their first known nationality is non-Belgian and they were born outside of Belgium. Their first known nationality defines which origin group they belong to. The second generation is defined as having at least one parent with a non-Belgian first-known nationality but being born in Belgium. The mothers’ first known nationality is used to determine the origin group of second-generation migrants. If the mother’s first known nationality is Belgian or missing, the father’s first known nationality is used instead.

After selecting couples based on the criteria mentioned above, we are left with a sample of 187,731 childless women in 2011. These women were followed up between 2012 and 2015 to see whether they entered into parenthood during this period. Women are excluded in case of emigration or death of themselves or their partner, relationship dissolution, or at the end of the observation in 2015. Relationship dissolution is defined as no longer living in the same household. At the end of the observation, the selected women had contributed 697,022 person-years to the analysis.

4.2. Methods

First, we provide a descriptive analysis of differences in women’s relative employment positions by origin (Belgian, Southern European, Maghrebi, or Turkish) and generation of the woman and her partner’s origin (Belgian or same origin). The woman’s relative employment position, measured on 1 January 2011, distinguishes between women in couples where both partners are employed (dual breadwinner), only the man is employed (male breadwinner), only the female is employed (female breadwinner), or neither partner is employed (no breadwinner). The Belgian Census of 2011 does not contain information to determine whether partners are full-time or part-time employed.

Next, we investigate the link between women’s relative employment position and the transition to parenthood through a set of multivariate event-history models of first birth hazards between 2012 and 2015 using a logit link function. The baseline in these models is the women’s age. First births during 2011 are excluded to ensure at least 12 months between the measurement of employment positions and entry into parenthood to avoid reverse causality. Four models are run separately for each of the origin groups. Model 1 investigates the association between women’s relative employment position and first births. Model 2 includes the interaction between women’s relative employment position and the baseline (woman’s age) in a quadratic specification to test whether first birth schedules differ significantly depending on the relative employment positions. In Model 3, an interaction between women’s relative employment position and generation is included to investigate variation in the link between relative employment positions and first birth risks by generation of women. This model is only estimated for women of Southern European, Maghrebi, and Turkish origin. Model 4 examines the interaction between women’s relative employment position and the origin of the partner. For women of Belgian origin, we distinguish between a Belgian partner or a partner of non-Belgian origin. For women of Southern European, Maghrebi, and Turkish origin, we distinguish between a Belgian-origin partner or a partner from the same origin group. For Models 2, 3, and 4, the deviance (−2 Log likelihood) is compared to the deviance of Model 1 to examine whether the inclusion of the interactions significantly improves the model fit.

The baseline (women’s age in a quadratic specification), generation, and origin of the partners are included in each of the models. In addition, we include two additional control variables. Marital status is included to control for varying fertility levels between married and unmarried couples. This is particularly important given that unmarried cohabitation has become the main (first) union formation type among native Belgian and European origin groups while (direct) marriage remains the dominant union type among Maghrebi and Turkish origin groups [81]. Region is included to control for varying fertility patterns between Flanders, Wallonia, and the Brussels Capital Region. The spatial distribution of the migrant population also strongly varies by region within Belgium [26]. An overview of all the variables can be found in

Table 1. Distribution of covariates by women’s migration background (in % of N couples).

	Belgian	Southern European	Maghrebi	Turkish
N persons	167,584	11,500	8930	2944
N person-years	624,092	41,387	31,829	10,862
N first births	44,501	2769	1178	3270
Mean age	33.41	32.45	31.97	29.53
Mean age at first birth	28.72	29.47	28.70	27.00
Relative employment position
Dual breadwinner	82.60	66.57	31.62	66.57
Male breadwinner	8.01	15.13	29.12	15.13
Female breadwinner	6.46	9.43	12.32	9.43
No breadwinner	2.93	8.87	26.94	8.87
Generation
1st generation		28.08	66.81	51.29
2nd generation		71.92	33.19	48.71
Origin partner
Same origin		59.39	20.62	10.07
Belgian	89.02	40.61	79.38	89.93
Other origin	10.98
Married	37.70	40.37	83.90	83.19
Region
Flanders	70.45	22.43	31.43	54.35
Brussels	4.60	20.70	43.95	22.18
Wallonia	24.95	56.87	24.61	23.47

Source: Belgian 2011 Census and Register data, calculations by authors.

.

Given that first birth hazards were measured between 2012 and 2015, while the census provides a static measure of the division of paid work and income in 2011, additional sensitivity models were run. The sensitivity analyses cover a shorter period of 2012–2013, closer to the measurement of employment in 2011. The results do not show any noteworthy differences compared to the analyses that use a lower follow-up period. Given that the detailed ethnic composition is prone to yielding low cell counts, a longer follow-up period is used in the main analyses.

5. Results

5.1. Descriptive Results

The descriptive results in Figure 1 indicate substantial heterogeneity in the division of paid work according to the woman’s origin and generation and the origin of her partner. In general, the results indicate that women of Belgian origin are most likely to be in a dual-earner couple, while the male breadwinner model is more common among couples with a migration background. Concerning women of non-Belgian origin, second-generation women are generally more often in a dual-earner model and less often in single-earner couples or couples without a breadwinner compared to women of the first generation. These results are in line with the higher levels of employment among women of the second generation compared to first-generation women [26]. There are two profound differences in the division of paid work within origin and generation depending on whether the woman is partnered with a Belgian or a same-origin male partner.

First, the percentage of women with a migration background who are in a dual-earner couple is remarkably higher when they are in a partnership with a Belgian man. The large differences in employment opportunities between Belgian men and non-Belgian origin women found in the Belgian labour market do not seem to lead to a higher percentage of male breadwinner couples. Instead, these results suggest that choosing a Belgian partner may be selective on socioeconomic characteristics such as educational level [28,30]. In combination with the high employment prospects generally found among Belgian men, this would increase the couple’s possibilities to attain a dual-earner model. Consequently, Turkish and Maghrebi women with a Belgian partner are potentially more likely to delay entry into parenthood due to more positive employment prospects. Among Southern European second-generation women, differences in the percentage of dual-earner couples are limited, indicating less differentiation in the ability to enter the labour market and less selective partner choices.

Second, Figure 1 indicates high percentages of first-generation women of Maghrebi and Turkish origin who have a same-origin partner and are in a male breadwinner couple or have no earner. The high levels of first-generation Turkish and Maghrebi women who do not participate in the labour market are in line with previous findings of high unemployment levels for non-European first-generation women, who often migrate to Belgium within the context of family formation or reunification [26,79]. These results suggest that women of the Turkish or Maghrebi first generation are unlikely to achieve a dual-earner model, either because they are unable to find employment or because they choose to not participate in the labour market, potentially in light of particularly limited employment prospects. As a result, the dual-earner model as a precondition to entering parenthood may not be prevalent among first-generation Turkish and Maghrebi women. In addition, the high percentages of couples without an earner also indicate high levels of economic vulnerability and uncertainty among first-generation women with a non-European background.

5.2. Women’s Relative Employment Position and Entry into Parenthood

Model 1 (

Table 2. Distribution of covariates by women’s migration background (in % of N couples).

	Model 1								Model 2
	Belgian		Southern European		Maghrebi		Turkish		Belgian		Southern European		Maghrebi		Turkish
	OR	Sig.	OR	Sig.	OR	Sig.	OR	Sig.	OR	Sig.	OR	Sig.	OR	Sig.	OR	Sig.
Age	1.78	***	1.47	***	1.15	***	0.97		1.82	***	1.55	***	1.08		1.10
Age2	0.99	***	0.99	***	0.99	***	1.00	*	0.98	***	0.99	***	1.00	***	0.99	*
Relative employment position
DBW (ref.)	1.00		1.00		1.00		1.00		1.00		1.00		1.00		1.00
MBW	0.73	***	0.88	**	0.78	***	0.82	*	1.91		0.54		0.56		15.4
FBW	0.81	***	0.87	*	0.91		0.82	*	7.78	***	18.1		0.17		1.44
NBW	0.56	***	0.72	***	0.74	***	0.80	*	9.29	*	5.50		0.05	**	14.5
Generation
1st gen.			1.08		0.98		0.88	*			1.07		0.98		0.89
2nd gen. (ref.)			1.00		1.00		1.00				1.00		1.00		1.00
Origin partner
Same origin (ref.)	1.00		1.00		1.00		1.00		1.00		1.00		1.00		1.00
Belgian			0.96		0.88	*	1.02				0.96		0.88	*	1.03
Other origin	0.99								0.99
Division of paid work * age
MBW * age									0.94		1.00		1.01		0.81
FBW * age									0.85	***	0.81		1.12		0.93
NBW * age									0.83	**	0.85		1.20	*	0.83
Relative employment position * age2
MBW * age2									1.00		1.00		1.00		1.00
FBW * age2									1.00	***	1.00		1.00		1.00
NBW * age2									1.00	**	1.00		0.99	*	1.00
Df	11		11		11		11		17		17		17		17
−2LL	273,140.0		17,497.4		15,033.0		6204.0		273,118.4		17,484.3		15,025.2		6198.0
ΔDf (model 1)									6		6		6		6
Δ−2LL (model 1)									21.45 ***		13.15 **		7.77		5.95

* Source: Belgian 2011 Census and Register data, calculations by authors; significance levels: p < 0.10 = *, p < 0.05 = **, p < 0.01 = ***. Results are controlled for relationship status (married vs. unmarried) and region (Flanders, Brussels, Wallonia); DBW = dual breadwinner; MBW = male breadwinner; FBW = female breadwinner; NBW = no breadwinner.

) shows the odds of entering parenthood by women’s relative employment position among women of Belgian, Southern European, Maghrebi, and Turkish origin. The odds ratios are displayed in Figure 2, using the dual breadwinner model as a reference category and controlling for age, generation, origin of the partner, relationship status, and region. The results indicate that women in a dual-earner couple have the highest odds of entering into parenthood in all origin groups. However, we find some variation in the odds of single-earner and no-earner couples depending on the women’s origin. First, among Southern European and Turkish women, being in a male or female breadwinner model lowers the odds of entering parenthood to a similar extent. Among Belgian women, both single-earner models are negatively associated with entering parenthood, yet the female earner model lowers first birth risks to a lesser extent compared to the male earner model. These results suggest that Belgian women in female-earner couples are less likely to delay entry into parenthood until the male partner has found a job. This finding also recurs among Maghrebi women in a female earner couple, who do not have significantly lower odds of entering parenthood compared to Maghrebi women in a dual breadwinner couple. Second, the odds of not having a breadwinner is examined for each origin group. For women of Belgian, Southern, and Maghrebi origin, being in a couple without an earner is associated with the lowest odds of entering parenthood. Among women of Turkish origin, the absence of an earner does not appear to lower the odds of entering parenthood more strongly compared to being in a single-earner couple of Turkish origin.

Figure 3 shows the predicted probabilities of having a first child by age (using a quadratic specification). It is important to note that these are first-birth probabilities for a selection of women who are in a co-residential partnership with a male partner (Belgian or same origin) and cannot be generalized to all women belonging to these origin groups. These results have to be interpreted in tandem with the results for Model 2, which tests whether the association between women’s relative employment position and entry into parenthood by age varies across origin groups. The results indicate that the fit of the model is significantly improved by including the interaction between age (quadratic specification) and women’s relative employment position among Belgian and Southern European women. Among Belgian women, differences between relative employment positions are particularly large for women in their twenties and smaller at older ages. Among women of Southern European origin, the predicted probabilities of women in a female earner model in particular are shown to be higher at younger ages compared to the other breadwinner couples. Whereas the results indicate some variation in the impact of having no earner by age for Maghrebi women, Model 1 is not significantly improved by allowing the association between women’s relative employment positions and first birth odds to vary by age among Maghrebi and Turkish-origin women. The results in Figure 2 predominantly indicate an earlier pattern of entering parenthood among the sample of partnered women of Maghrebi and Turkish women compared to the other origin groups.

The results for Model 3 (

Table 3. Odds ratios for logit Models 3 and 4 of first birth, Belgium 2011–2015.

	Model 3						Model 4
	Southern European		Maghrebi		Turkish		Belgian		Southern European		Maghrebi		Turkish
	OR	Sig.	OR	Sig.	OR	Sig.	OR	Sig.	OR	Sig.	OR	Sig.	OR	Sig.
Age	1.47	***	1.15	***	0.97		1.78	***	1.47	***	1.15	***	0.97
Age2	0.99	***	0.99	***	1.00	*	0.99	***	0.99	***	0.99	***	1.00	*
Division of paid work
DBW (ref.)	1.00		1.00		1.00		1.00		1.00		1.00		1.00
MBW	0.85	**	0.73	***	0.81	*	0.73	***	0.87		0.76	***	0.83	*
FBW	0.84	*	0.92		0.80	*	0.79	***	0.93		0.93		0.85
NBW	0.70	***	0.72	***	0.81	*	0.58	***	0.74	**	0.76	***	0.81	*
Generation
1st gen.	1.04		0.94		0.86				1.08		0.99		0.88	*
2nd gen. (ref.)	1.00		1.00		1.00				1.00		1.00		1.00
Origin partner
Same origin (ref.)	1.00		1.00		1.00		1.00		1.00		1.00		1.00
Belgian	0.96		0.88	*	1.01				0.97		0.88		1.10
Other origin							0.99
Relative employment position * generation
MBW * 1st gen.	1.13		1.13		1.05
FBW * 1st gen.	1.16		0.99		1.13
NBW * 1st gen.	1.12		1.06		0.98
Relative employment position * origin partner
MBW * Belgian							1.01		1.02		1.20		0.93
FBW * Belgian							1.09		0.87		0.80		-
NBW * Belgian							0.87		0.94		0.68	*	0.59
Df	14		14		14		14		14		14		13
−2LL	17,496.1		15,031.7		6203.6		273,135.8		17,496.6		15,026.3		6197.7
ΔDf (model 1)	3		3		3		3		3		3		2
Δ−2LL (model 1)	1.33		1.30		0.36		4.13		0.80		0.08		0.57

* Source: Belgian 2011 Census and Register data, calculations by authors; significance levels: p < 0.10 = *, p < 0.05 = **, p < 0.01 = ***. Note: Results are controlled for relationship status (married vs. unmarried) and region (Flanders, Brussels, Wallonia); DBW = dual breadwinner; MBW = male breadwinner; FBW = female breadwinner; NBW = no breadwinner.

) indicate that the association between women’s relative employment position and the odds of having a first child do not vary significantly by generation of the women. None of the interaction terms were significant, and the model did not significantly improve. With regards to the interaction between women’s relative employment position and the origin of the partner, Model 4 (Table 3) indicates no significant interaction for any of the origin groups. While the overall fit of the model does not significantly improve when including the interaction with the partner’s origin for any of the origin groups, Maghrebi women in couples without an earner have a significantly lower likelihood of entering parenthood when the partner is Belgian compared to women with a same-origin partner.

6. Discussion

Whereas the link between employment and fertility has been thoroughly studied among majority populations, variation in this link among population subgroups with a migration background has not been examined to the same extent. This is remarkable given that the differential opportunity structures and sociocultural elements among migrant populations may impact the economic preconditions of fertility decisions. This paper focused specifically on how women’s relative employment position affects entry into parenthood among women with a Belgian, Southern European, Maghrebi, or Turkish background who are in a co-residential partnership. We used data from the 2011 Belgian Census combined with longitudinal information from the population registers on first births for the period 2012–2015.

First, the descriptive results for this study indicate substantial variation in women’s relative employment positions by migration background. Compared to women of Belgian origin, women of Maghrebi and Turkish origin are more often in couples with only a male earner or no earner at all while no differences are found for women of Southern European origin. These results are in line with previous findings on higher levels of unemployment and employment instability and the larger gender gap in employment of Maghrebi and Turkish-origin groups in the Belgian labour market [15,26,40]. In addition, more limited employment prospects may discourage investments in education and labour force participation prior to entering parenthood [52,72]. Women of Maghrebi and Turkish origin have also been found to be more likely to enter into a co-residential union from more vulnerable economic positions such as inactivity and unemployment [81], further contributing to the differential relative employment positions found in this paper. These results can also be linked to selective partnering patterns. The choice to partner with a Belgian man has been associated with higher educational levels and thus better employment prospects [28,30]. In general, the descriptive findings regarding women’s relative employment positions indicate substantial disparities in the prevalence of couples with a single earner or no earner at all. It is important to note that these breadwinner models are more often found among women of Maghrebi and Turkish origin and particularly among those of the first generation. As a result, these groups may also experience higher levels of economic uncertainty and vulnerability.

Next, the results of the event-history analysis indicated that women in a dual-earner couple were most likely to enter into parenthood regardless of migration background. These results are in line with previous studies and confirm that the dual-earner model provides the most desirable position to enter parenthood. Aside from providing the financial resources to support a family, in the Belgian context, some work-family reconciliation that lower opportunity costs of parenthood for working women in particular (e.g. parental leave) are strongly tied to stable employment [19]. These findings are in line with our expectations for women of Belgian (Hypothesis 1) and Southern European (Hypothesis 2a) origin. Among Maghrebi and Turkish-origin women, we expected differences in first-birth risks between relative employment positions to be more limited (Hypothesis 2b). This hypothesis can only be confirmed partly. In contrast to our expectations (Hypothesis 3a,b), the link between women’s relative employment was not found to differ by generation or origin of the partner. The results indicate that relying on a single earner is associated with lower first-birth risks among all origin groups. However, whereas the absence of a breadwinner is linked to an even lower likelihood of entering parenthood among Belgian women, this negative link is weaker among women with a migration background. Among Maghrebi and Turkish women in particular, the association between having no earner or having a single earner is largely the same, with the exception of Maghrebi women in a female earner couple. These results suggest that women of Maghrebi or Turkish background are not necessarily delaying parenthood when experiencing economic uncertainty or vulnerability. In the context of particularly limited employment prospects, women may seek to find certainty or fulfilment in motherhood [73,74]. Hence, in addition to the findings that women of migrant origin are more likely to be in a vulnerable socioeconomic position, the economic vulnerability associated with the absence of an earner does not limit family formation more compared to having one earner. This information is crucial for policymakers given that higher levels of economic vulnerability in the household are found to have adverse effects throughout the life course of children [82]. In addition to the awareness of the higher percentage of single-earner and no-earner households among Maghrebi and Turkish origin groups, labour market policies aimed at maximizing employment prospects for all groups and for non-European origin groups and women, in particular, would increase income positions and the well-being within these households. Ensuring good employment prospects encourages investments in education and labour force participation before entering parenthood, thereby increasing women’s labour force positions after family formation and access to work-family reconciliation policies.

Given that this study is only a first step in examining variation in the economic mechanisms underlying family formation, we encountered several limitations that open avenues for future research. First, the main limitation of this study is the lack of a time-varying indicator for women’s relative employment position. Sensitivity analyses have shown that results using a shorter follow-up period (limiting it to one or two years) give results that are similar to the findings in this study. We have opted to include all years in the study to maximize the number of person-years used, particularly among the smaller migrant groups. Including a yearly measure of women’s employment positions instead of a fixed measure for 2011 would allow us to model the link to first birth risks in a more detailed way but may also lead to increased problems of endogeneity. Second, taking a more detailed operationalization of women’s and men’s origin and generation into account would be a relevant avenue for future research. Specifically among the migrant population, a detailed distinction for the origin and generation of both partners often results in small cell counts. For this reason, this paper focused specifically on the origin and generation of the women, taking only the origin of the male partners into account to a limited extent. Ideally, studies would include detailed information on the composition of couples based on both partners’ origin group and generation, which would allow a more developed couple perspective in the investigation of variation in the economic preconditions of parenthood. Third, this paper focused specifically on entry into parenthood given that it is a pivotal life-course transition and has been elaborately linked to employment positions in previous studies [66,83]. Future research can extend upon this research by investigating similar variations in the economic mechanisms underlying higher-order births. Fourth, the census data do not allow us to include other potentially relevant factors such as sociocultural factors (e.g., norms and values regarding gender roles and social expectations with respect to starting a family) or the division of unpaid work (e.g., childrearing and household tasks). Previous studies suggest that these factors may also vary substantially by migration background [16,56,57]. Whereas the quantitative findings of this study can indicate varying economic preconditions to parenthood, future qualitative studies are better suited to discover whether and why the underlying decision-making process and within-couple discourses with respect to dividing work and having children also vary by migration background.