Ethnic Differences in Leaving Home: Timing and Pathways

Abstract

Introduction

Extensive research has identified the determinants of leaving the parental home in Western countries. Much of this research deals with the routes out of home (such as marriage, education, and labor market participation), and their trigger roles in determining the decision to leave home (see, e.g., Bernhardt et al. 2005; Goldscheider et al. 1993). Other studies focus on the opportunities and constraints within the parental home and in the labor and housing markets (e.g., Ermisch 1999; Mulder and Hooimeijer 2002; Nilsson and Strandh 1999; Whittington and Peters 1996). Still other work has focused on differences in the patterns and timing of leaving home between generations, among regions within countries, and across countries according to the degree of traditionalism, individualization, and organization of the welfare state (Aassve et al. 2002; Aquilino 1991; Buck and Scott 1993; Goldscheider et al. 1993; Giuliano 2007). The ample research attention paid to leaving the parental home is not surprising because it marks a profound change in the life of young adults. For them and their parents, leaving home has major implications for the housing situation and family relationships. For the young adult, leaving home opens up opportunities to enroll in higher education, enter the labor market, and form a family. The timing of leaving home has implications for housing and labor markets. It has also been suggested that patterns of later versus earlier home-leaving in southern versus northern European countries are a major factor in the very low versus higher fertility in these regions (Dalla Zuanna 2001).

Another reason for studying leaving the parental home is its connection with the profound demographic changes in Western societies since the 1960s that have been termed “the second demographic transition,” observed first in Europe (Van de Kaa 1987) and later also in the United States (McLanahan 2004; Raley 2001). These changes include two that are connected with patterns of leaving home: a decrease in marriage not completely offset by a rise in unmarried cohabitation, and a rise in living alone before first union formation.

Although several North American studies have addressed racial and ethnic differences in leaving home (e.g., Goldscheider and Goldscheider 1997, 1999; Mitchell et al. 2004), few studies have paid attention to the varying patterns of leaving home among youth from migrant families (for Sweden, see Bernhardt et al. 2007; and Nilsson and Strandh 1999; for the Netherlands, see Bolt 2002; and De Valk and Billari 2007; for the United States, see Glick and Van Hook 2002). This lack of attention likely owes to the limited availability of suitable data. Most studies have used sample surveys that offer restricted opportunities to deal with the particular position of migrants.

An increasing share of the population in many Western countries consists of immigrants and their descendants. Differences in the timing and patterns of leaving home between migrant groups can have implications for their investment in human capital, their socioeconomic prospects, and for society at large. It is therefore important to gain more insight into the timing and patterns of leaving home among young adults from migrant families. Furthermore, it is interesting to see whether the changes connected with the second demographic transition are found only among the dominant or native population of a country in which this transition has taken place, or also among people from migrant families.

In this article, we examine the timing of leaving the parental home in the Netherlands among young adults from migrant and Dutch families. For convenience, young adults from migrant families are denoted as “migrant youth” or “migrants,” even though only some of them actually migrated with their parents. A greater share were born in the Netherlands from at least one foreign-born parent. We use unique individual administrative panel data for 1999–2004 from the Social Statistical Database (SSD) on young adults born in 1977 and 1983, who were therefore aged 22 and 16 in 1999. The analyses were performed using discrete-time hazard models with competing risks corresponding with various pathways out of the parental home: to form a union (distinguished further into marriage and cohabitation in an additional analysis), to live alone in an independent residence, and to live in a shared residence with others.

Background

Migrants in the Netherlands

The ethnic composition of the Netherlands population has changed significantly owing to immigration flows following the World War II. Migrants in the Netherlands can be grouped into six categories of origin countries and ranked according to their population size: Turks, Moroccans, Surinamese, Antilleans, other non-Western, and Western migrants. The first Turks and Moroccans came to the Netherlands as guest workers in the 1960s, while immigration flows from Surinam and the Netherlands Antilles have been derived from colonial relations. Immigration from Western countries has been related to economic conditions. The category “other non-Western” covers a variety of more recent immigrants from developing countries who frequently entered as asylum seekers or family migrants.

This historical background reflects the socioeconomic position of these groups and their cultural distance from the host society. Surinamese and Antillean (Caribbean) migrants often speak Dutch and adopt cultural norms similar to those of the Dutch through their colonial relations. Their labor-market position is somewhat less favorable than that of the native Dutch. In contrast, the predominantly Muslim Turkish and Moroccan (Mediterranean) migrants are frequently less well educated, hardly ever speak Dutch prior to immigration, and have a greater cultural distance from the Dutch. There is some empirical evidence that these migrants face significant difficulties in the Dutch education system, labor market, and housing market. They exhibit a high drop-out rate in education, are frequently unemployed, and are concentrated at the bottom of the occupational distribution (Van Beek et al. 1997; Zorlu and Hartog 2008). In addition, some evidence suggests that mortgage banks are reluctant to grant mortgages to ethnic minorities (Aalbers 2007). The category “other non-Western” is a heterogeneous group whose position tends to be similar to that of Turkish and Moroccan migrants. Western migrants are in many respects similar to the Dutch, and their labor market position is favorable (Zorlu and Hartog 2008).

Migrants who are legal inhabitants of the Netherlands are usually entitled to the same welfare provisions as Dutch citizens. The only exception is formed by a small number of very recent asylum migrants, almost exclusively belonging to the category “other non-Western.” In practice, recent migrants may be disadvantaged with respect to those benefits that are related to the number of years spent in the Dutch labor market (part of the employment and disability benefits). This problem rarely affects the young adults we study, however, because most young adults have no, or just a very short, labor market history, and because most young adults from migrant families are second-generation migrants or arrived as children.

Caribbean and Mediterranean migrants have usually settled in the Netherlands permanently; return migration is infrequent. In contrast, one-fifth of other non-Western and one-third of Western migrants leave the country within four years of their entry (Zorlu and Mulder 2008). Most probably, return migrants are predominantly those without children.

Hypotheses on Ethnic Differences in Leaving Home

We aim to shed light on the extent to which the leaving home of migrant youth is consistent with three different behavioral patterns: (1) leaving home in accordance with the cultural norms of the origin society; (2) leaving home in accordance with the patterns exhibited by the native-born in the host society, possibly more so for second-generation migrants; (3) patterns of leaving home connected with the specific position of migrants between the cultures of the origin society and the host society.

Ethnic differences in cultural norms refer to dominant cultural norms and preferences in a migrant’s country of origin in the timing and routes of departure from the parental home. In the origin countries of non-Western migrants, decisions regarding the timing of transitions into adulthood are often prone to familial and religious concerns. In the secularized and individualized Dutch society, in which values concerning family formation are typical of those societies developing along the lines of the second demographic transition (Lesthaeghe and Surkyn 1988), young adults are much more likely to make autonomous decisions.

Caribbean and particularly Mediterranean migrants are more family-oriented than are the Dutch (Schans 2007). There are also fundamental differences between Caribbean and particularly Mediterranean migrants and the Dutch in preferences regarding the timing and patterns of leaving home, union formation, and childbearing. Turks and Moroccans prefer a much younger age of marriage than the Caribbean and Dutch, but a somewhat older age at leaving home. In contrast, unmarried cohabitation, childbearing outside marriage, the economic independence of women, and single motherhood are more common in the Caribbean tradition than in the traditions of other groups (De Valk and Liefbroer 2007b). Differences in cultural norms might therefore lead to a less important role for independent living and a more important role for marriage among Turkish, Moroccan, and other non-Western migrants. In the Netherlands society, leaving home to live with relatives might be an attractive option for Turkish, Moroccan, and other non-Western youth who need to leave home for education or work.

The above considerations lead to the following hypothesis:

Hypothesis 1:

Compared with Dutch youth and Caribbean migrants, Turkish, Moroccan, and other non-Western migrants are (a) more likely to leave home to form a union, particularly for marriage; (b) less likely to leave the parental home to live alone independently; and (c) more likely to leave home to share a residence with others.

There are, however, reasons for expecting alternative patterns of ethnic differences in leaving home. First, the differences in leaving home between non-Western migrants and native Dutch might be smaller than argued above. There are indications that the cultural norms of non-Western migrants have changed in the direction of those of the native Dutch. From an analysis for the late 1990s of the union-formation preferences of Turkish, Moroccan, and Dutch students in secondary schools, it appeared that Turkish and Moroccan adolescents, and particularly Moroccan boys, were much more in favor of unmarried cohabitation than one might think on the basis of research among the general population (De Valk and Liefbroer 2007a). Even though adolescents might change their opinions after reaching young adulthood, or act according to the norms rather than according to their own opinions, it cannot be ruled out that union-formation patterns may have undergone some change in the direction of the native Dutch—that is, toward more unmarried cohabitation. In fact, De Valk (2008) found some indications from survey data for Turks and Moroccans living in the large cities in the Netherlands that, consistent with preferences, unmarried cohabitation has started to grow among Moroccan men.

Leaving home earlier does not necessarily mean that family ties are neglected and parental care needs are ignored. In the Dutch context, geographical distances are small, even more so for migrants who are concentrated in large cities and tend to remain in the same city if they move. Leaving home at a small distance from the parents enables the young adult to escape daily parental control, but at the same time, the young adult and the parents can support each other, as expected in Mediterranean cultures, albeit in a weaker form.

Although cultural norms are in general important in the timing of life course transitions, these transitions are also related to institutional factors (Aassve et al. 2002). Living arrangements in developing countries are probably not independent of credit constraints, housing and labor markets, and institutional structures. For example, the young age of leaving home in northwestern Europe is often attributed to the advanced welfare state that provides a high level of support for young adults, such as student loans, unemployment and welfare benefits, and rent subsidies (Billari and Liefbroer 2007). Consequently, the leaving-home behavior of non-Western youth in the Netherlands may differ significantly from the patterns in their countries of origin. In fact, delays in leaving home in the countries of origin may be caused by constraints in the housing and labor markets on young adults’ transition to independent residence, rather than by cultural norms. In the Netherlands, young adults can leave the parental home without a substantial loss of living standard, thanks to the generous facilities of the welfare state.

If the above arguments hold, the differences between non-Western and native Dutch youth in leaving home would be limited, and Hypothesis 1 would not be supported or would be only partly supported. It is possible to go even further and put forward an alternative hypothesis to Hypothesis 1b. Previous research reveals that the atmosphere in the parental home is important in the timing of leaving home (De Jong Gierveld et al. 1991), and so is the comfort in the parental home (Murphy and Wang 1998). Given the practical feasibility of independent living arrangements in the Netherlands, the leaving-home decisions of young adults may be quite strongly related to the comfort and atmosphere in the parental home. In this context, individuals who have a smaller private space in the parental home, who encounter more parental control of their daily lives, or who experience more intergenerational conflict may tend to leave home earlier. It is likely that Mediterranean young adults have to deal with these sorts of discomforts. They grow up in the individualistic Dutch society and internalize mainstream cultural norms and values through education and contact with peers, while their parents mostly adhere to the cultural norms of their origin countries. These young adults may experience tension and conflict in their parental families through their awkward position between two distinct cultures that are difficult to reconcile. Young adults from migrant families may therefore leave home early not only for union formation, but also for independence. This leads to the following alternative hypothesis for Hypothesis 1b:

Hypothesis 1b, Alternative:

Turkish, Moroccan, and other non-Western youth are more likely to leave home to live alone independently than are Dutch youth.

As in several other Mediterranean countries, leaving home among men in Turkey frequently follows a fixed sequence: entry into the labor force, marriage, the birth of the first child, and finally leaving home (Koc 2007). Women frequently move into their husband’s parents’ home on marriage. This pattern might also show up in the Netherlands. From a small-scale study by Bolt (2002), it has become clear that, unlike young adults from Dutch families, those from Moroccan and particularly Turkish families are much more likely to live with their parents or with other family members or friends after forming a partnership. Previous research has shown that in Western countries, women leave the parental home at younger ages than men, particularly for union formation (Aquilino 1991; Buck and Scott 1993; Goldscheider et al. 1993; Mulder and Hooimeijer 2002). Over and above this conventional finding, the Mediterranean pattern might show up in greater gender differences among Turkish, Moroccan, and other non-Western migrants than among the Dutch or Caribbean migrants in the timing of leaving home for union formation.

Cultural norms among Mediterranean migrants, such as those promoting strong family solidarity, are stricter for women than for men (De Valk and Liefbroer 2007a; Merz et al. 2009). Mediterranean women might therefore encounter more restrictions than men on leaving home for reasons other than marriage.

Hypothesis 2a:

Women are more likely than men to leave the parental home for union formation. This gender difference is particularly great among Turkish, Moroccan, and other non-Western migrants (and conversely, ethnic differences in leaving home for union formation are greater among women than among men).

Hypothesis 2b:

Turkish, Moroccan, and other non-Western migrant women are particularly unlikely to leave the parental home for independence and shared residence compared with their male counterparts and with Dutch women.

Whereas young adults who migrated to the Netherlands with their parents have spent part of their youth in the origin country and have probably been partially socialized there, this is not the case for children whose parents were migrants but who were themselves born in the Netherlands (second-generation migrants). Their preferences and norms might be shaped more by mainstream cultural norms in the Netherlands than those of first-generation migrants. This influence may hold even more for the children of couples consisting of an immigrant and a non-immigrant parent (mixed second-generation migrants), since the non-immigrant parent will have social norms closer to mainstream norms. A recent study suggests that values regarding intergenerational family solidarity are weaker among second-generation Turks and Moroccans in the Netherlands than among the first generation (Merz et al. 2009).

Hypothesis 3:

The leaving-home behavior of second-generation and particularly mixed second-generation migrants is more similar to that of the Dutch than to that of the first generation.

Differences between migrants and the Dutch in leaving home might be caused partly by differences in population composition with regard to socioeconomic resources, family structure, or the residential context. Compared with the Dutch, non-Western migrants tend to have fewer resources, live in housing of lower quality, have larger families, and be highly concentrated in particular neighborhoods of large cities. Support for the three hypotheses would suggest that population composition is not the whole story and that cultural norms are important—at least if we control adequately for other major factors influencing leaving home. We acknowledge that we might not have succeeded fully in performing these controls. At the same time, we have no indications of particular behavior with respect to labor migration or circular migration to the country of origin among migrant youth or their parents that would lead us to expect migrant youth to show particular leaving-home behavior.

Other Factors Influencing Leaving Home

Individual Resources

Empirical evidence suggests that young people with increasing economic resources are more likely to leave home (Aassve et al. 2002; Avery et al. 1992; Haurin et al. 1997; Whittington and Peters 1996). We use the young adult’s employment position and fiscal income to capture the effects of individual economic resources, and whether the young adult receives a benefit to indicate the use of the Dutch welfare system. Unfortunately, we do not have information about completed level of education, but we do have information about enrollment in higher vocational training and university. Enrollment in higher education is frequently associated with a step toward residential independence, either to live alone or to share with roommates (Bernhardt et al. 2005; Mulder and Hooimeijer 2002). Furthermore, students are unlikely to form partnerships (Blossfeld and Huinink 1991).

Parental Economic Resources

The impact of parental resources (mainly income) is considerable, but differs with age and with the pathway of leaving home. Parents mainly use their resources to prevent early marriage and to encourage leaving home among older children (Avery et al. 1992). Evidence for the Netherlands suggests that parental resources have a positive influence on leaving home to live without a partner (Mulder and Hooimeijer 2002).

The Quality of the Parental Home

The quality of the parental home is strongly associated with the propensity to leave home (Ermisch 1999). The scarcity of physical space in the parental home and a corresponding lack of spatial privacy can accelerate the process of leaving home, whereas young adults may prefer to stay in a high-quality home (Murphy and Wang 1998)—a phenomenon known as the feathered-nest effect (Goldscheider and Goldscheider 1999).

Family Structure

Children from nonstandard families are more likely to leave home at a given point in time than are children from two-parent families (Aquilino 1991; Bernhardt et al. 2005; Buck and Scott 1993; Haurin et al. 1997). Therefore, we considered the marital status of the mother in detail. Additionally, we constructed control variables indicating whether siblings are coresiding with their parents. These variables can capture the intensity of competition for parental resources, possibly leading to a delay in leaving home. Alternatively, the presence of siblings leads to less space and privacy within the parental household, which might accelerate the leaving-home process (cf. Murphy and Wang 1998). In contrast with previous research, we also accounted for the number of siblings outside the parental home. These siblings may encourage those still at home to leave by setting an example and by helping to find accommodations based on their experience or new network. The children of older parents might have more intergenerational conflicts than do the children of younger parents, but older parents will also have had more time to accumulate assets. We therefore also took the age difference with the parents into account.

The Residential Context

Young adults might feel pushed from less attractive neighborhoods. At the same time, those who live in large cities may have a lower necessity to leave for work or education (cf. Mulder and Clark 2000). We therefore accounted for neighborhood attributes (the share of non-Western migrants and the value of homes) and for whether the parental home is located in one of the Netherlands’ four large cities.

Data

We used a rich individual administrative panel database that covers the entire population of the Netherlands: the Social Statistical Database (SSD) housed by Statistics Netherlands. The SSD contains a variety of variables measuring individual sociodemographic and socioeconomic position and geographic mobility. The measurement moment is the last Friday of September of each year. Data were available for the period 1999–2004. The data have been derived from the population register and thus contain information about every legal inhabitant of the Netherlands. The follow-up through the years is based on a unique registration number and is 100% complete as long as people remain registered in the Netherlands.

We selected those young people who lived with at least one of their parents and were 16 or 22 years old on the last Friday of September 1999. Most of these people were born in 1983 or 1977, but some were born in 1984 or 1978. For simplicity, we refer to these two groups as (birth) cohorts 1983 and 1977. These people were followed until 2004, or until they left the parental home if that happened before 2004. Cohort 1983 is observed at ages 16–21, cohort 1977 at ages 22–27. We thus capture the most dynamic period of leaving home, from age 16 to 27.

Almost all individuals from birth cohort 1983 were living in the parental home in 1999, when they were 16 years old. This fact ensures a nonselective population. For birth cohort 1977, the population of those living at home is selective with regard to not yet having left. This cohort experienced the process of leaving home at age 16–21 in 1994–1999, before the period of observation. If we can assume that leaving-home behavior did not change fundamentally between 1994–1999 and 1999–2004—and we see no reason why we cannot—we can treat the observation of cohort 1977 at age 22–27 as a follow-up of the observation of age 16–21 for cohort 1983. Another possible complication is that the population of cohort 1977 also includes individuals who left home earlier but returned. The share of these returners is likely to be too small to influence our analysis. The validity of this supposition is indicated by the smoothness of the curves in Fig. 1, which depicts the percentage living in the parental home by age estimated from the data for the two separate cohorts. The figure shows the well-known picture that women leave the parental home earlier than men do.

The dependent variable measures the competing risks for the pathways out of the parental home. Coresiding with one or both parents is the reference category. The pathways were constructed sequentially as follows:

• Event = Union formation: if individual i lived in the parental home in year t but left home and was married or cohabited in year t + 1. In an additional analysis, the event of union formation was further distinguished into “married” and “cohabiting” based on the legal status of the union.

• Event = Independent residence: if individual i lived in the parental home in year t but left home to live as a single-person household in year t + 1.

• Event = Shared residence: if individual i lived in the home in year t and was not classified in the category union but left home to share a residence with other people in year t + 1.

Whether someone lived in the parental home was determined through the record linkage of parents and children and a check to see whether they were registered at the same address. For the age category under study, the record linkage between parents and children and between persons and addresses is 100% complete, as long as people fulfill their legal obligation to register their changes of residence. Those registered as married and living at the same address as their spouse, and those who were classified as cohabiting unmarried according to the standard decision rules used by Statistics Netherlands were assumed to have formed unions. Independent residence was detected from a check to see whether the individual was the only person registered at the particular address. Finally, those who were registered at the same address with people other than parents or partners were assumed to share their residence.

Although register data provide highly reliable information on the socioeconomic and demographic characteristics of individuals, there might be some measurement error in the dependent variable. The reported residential address might not always reflect the true residential location for students because student grants are higher for those living away from their parents. This rule can generate incentives to manipulate the registered address. At the same time, however, such manipulations are counter-affected by the legal necessity to report the true residential address for many other administrative purposes in the highly organized Dutch society. Another possible source of some measurement error is that some of those categorized as sharing a residence may actually be cohabiting, or vice versa. The decision rules of Statistics Netherlands to determine unmarried cohabitation lead to credible estimates at the population level that are in line with estimates from large-scale surveys. It cannot be ruled out, however, that some individuals are misclassified and that these misclassifications are concentrated in particular small population subgroups.

As Fig. 2 shows, the importance of the various pathways of leaving home changes with age. The share of home-leavers for union formation is very small at younger ages and sharply increases after age 21. Leaving for independence gains in importance up to age 21 and then appears to stabilize. The fraction of those leaving for a shared residence among nest-leavers remains quite stable over time.

The variables used are listed in Table 1, together with the mean values according to the three pathways at the time of the last observed person-year or at censoring. All individual covariates can vary with time.

Method

We modeled individual leaving-home behavior using a discrete-time duration model with competing risks. The process time for coresiding with the parents is assumed to start at the age of 16 or 22 in 1999, and to end when an individual leaves the parental home for the first time. It is censored for an individual still coresiding with the parent(s) in the last observation year, 2004. Repeated spells are not considered.

The duration of stay in the parental home, T, is assumed to be a discrete random variable that takes on positive integer values only. The population at risk is in the parental home in the year 1999, which is the starting point, with time T = 0. Each observation continues until time t, at which point an event occurs or the observation is censored in 2004. The parental home spell can end, T = t, in any of j states. In each period, a person can stay at the parental home (reference) or can leave home for union formation (j = 1), for independent residence (j = 2), or for shared residence (j = 3). In an additional analysis in which marriage and cohabitation were distinguished from each other, there were four rather than three end states. By modeling the pathways of leaving home as competing risks, we allow variations in the baseline hazard and in the coefficients for the covariates for all destination states.

For the ith person, the hazard rate of leaving home in spell k into state j in period t, h_ij(t), is the conditional probability of a transition into state j in this interval t, given the survival in the parental home until ⁠. Assuming that the error terms of the utility functions for the competing risks are independent, the hazard rate of leaving home is given by ⁠. We suppose the discrete interval-specific hazard to be multinomial logistic, as suggested by Allison (1982),¹

(1)

where x_i(t) is the observed vector of explanatory variables, β_j is the vector of corresponding parameters to be estimated, and α_j(t) describes the variation in the baseline hazard, which captures the destination-specific duration dependence. We specify the baseline hazard as piecewise constant, the most flexible form, by specifying annual dummy variables α_j(t). In the separate models for people from different ethnic origins, however, we chose a linear baseline hazard specification when we estimated competing risk models by migrant origin (see Table 5). This is because a piecewise-constant specification yielded unrealistic parameters owing to the small number of events in the first year for specific pathways and specific ethnic origins.

Separate models were estimated for the two birth cohorts. First, we estimated models in which migrant origin is an independent variable. These models reflect the overall relative risks of the migrant groups compared with the Dutch, given the background variables. It is quite likely that these variables impact differently on the risks of leaving home for particular migrant groups. To see these differences, we also estimated the competing risks model given by Eq. 1 separately for those ethnic origins and ages for which the differences in the timing and pathways of leaving home are strongest: Dutch, Moroccan, and Turkish youth aged 16–21. These separate models also permit a further test of Hypotheses 2 and 3.

Even though the data were not derived from a sample, but from a complete population, we still pay attention to significance levels. We believe this makes sense because the data can be regarded as a one-moment sample from a theoretical population of many time points.

Parameter Estimates for Ages 16–21 and 22–27

The parameter estimates of the competing risk models by gender are presented in Tables 2 and 3 for ages 16–21 (cohort 1983) and Table 4 for ages 22–27 (cohort 1977). For ages 16–21 in Table 2, in the first four years the hazard of leaving for union formation increases sharply. The hazards for the pathways “independence” and “shared residence” increase less spectacularly and decrease for women for age 21. The changes in hazards for ages 22–27 in Table 4 are less strong than for ages 16–21. The risk of leaving home for union formation shows a marked decrease for women in this second age category.

With respect to ethnic differences for ages 16–21, the hazards of leaving home are particularly great for the young people of Turkish or Moroccan origin compared with the Dutch. It is striking that the differences are more pronounced for the pathways “independent” and “shared” than for “union formation”; for women, the difference in leaving for union formation is not even significant. This difference is also made clear in Fig. 3, in which the predicted pathway-specific hazard rates for the sample means, based on the model presented in Table 2, are depicted for Turks, Moroccans, Surinamese, and the Dutch. Surinamese and other non-Western women have a significantly lower probability of leaving home for union formation than Dutch women. The high likelihood of leaving home for shared residence among non-Western migrants is as expected. Migrants of all origins, but particularly Turks and Moroccans, leave home more frequently for a shared residence than the Dutch do. Unexpectedly, however, Turkish and Moroccan migrants in particular are also more likely to leave home for independent residence than are the Dutch. This holds for both men and women originating from Turkey and Morocco.

Selected multinomial odds ratios from the models for ages 16–21 with four pathways instead of three (marriage, cohabitation, independence, and shared residence) are presented in Table 3. The probability of leaving home for marriage is spectacularly greater for Turkish men and women, and also considerably greater for Moroccan women, than for their Dutch counterparts. Apparently, Turkish migrants in particular marry frequently at these young ages, which would be considered very young in Dutch society. The estimates further show that the probability of leaving home to cohabit is significantly lower for Moroccan, Turkish and Other non-Western women than for Dutch women, which is in line with Hypothesis 1a. However, at these ages, Moroccan men are surprisingly even more likely to leave home to cohabit than Dutch men. This result should be considered with caution because the numbers of home-leavers are small at these ages, but it is in line with the findings of De Valk and Liefbroer (2007a) and De Valk (2008) from other data sets concerning the preferences and behavior of young Moroccan men.

Although they are less spectacular, ethnic differences in leaving home for union formation tend to turn in the opposite direction for ages 22–27 (Table 4). At these ages, leaving home for union formation occurs significantly less among most migrant groups than among the Dutch. Women are especially less likely to leave home for union formation. Differences in leaving home for independence show a mixed pattern. Ethnic minority women are less likely to leave home for independent residence, whereas the differences for men are often not statistically significant. Leaving home for shared residence continues to be more popular among most migrant groups than among the Dutch.

In all, these results are only partly in line with Hypothesis 1. Apparently, the impact of cultural norms on leaving home is at least partly counter-affected or complemented by other factors, as predicted by the alternative hypothesis to Hypothesis 1b. The facilities of the welfare state might play some part in the great likelihood among migrants to leave home for independence. Because a considerable part of welfare benefits in the Netherlands depends on the household situation and young adults thus become eligible for these only after leaving home, it is not possible to account fully for the use of these benefits in the analyses. But, although the characteristics of the welfare state might explain why Turkish and Moroccan youth do not leave home later for independence than the Dutch do, they do not explain why they might leave earlier. As suggested in the argumentation leading to the alternative to Hypothesis 1b, there might be a connection with the difficult position of migrant youth placed between two cultures. Migrant youth possibly have to deal with greater intergenerational conflict because they have internalized mainstream cultural norms and values that differ from those of their parents who immigrated to the Netherlands at adult ages.

The second hypothesis can be only partially evaluated with the results reported in Tables 2, 3 and 4. Greater ethnic differences in leaving home for union formation were expected for women than for men, but instead a greater difference is found for Turkish and Moroccan men compared with Dutch men. In line with Hypothesis 2a, however, ethnic differences are greater for women than for men for unmarried cohabitation (Table 3): Turkish, Surinamese, and other non-Western migrant youth are particularly unlikely to cohabit. In line with Hypothesis 2b, the greater likelihood for non-Western youth of leaving home for independent residence is less pronounced for women than for men.

With regard to the estimates for ages 16–21, second-generation migrants have, in general, a lower risk of leaving home, particularly for independence and shared residence. This finding is in line with the third hypothesis, which predicts that the leaving-home behavior of second-generation migrants is more similar to Dutch traditions than to traditions of the origin society. Some difference is observed between the mixed second generation and the second generation with two foreign-born parents. Among the second generation, those with only one foreign-born parent have a lower risk of leaving home for shared residence. Women from the mixed second generation are more likely to cohabit, which partly offsets the ethnic differences in cohabitation among women (see Table 3). These smaller differences from the Dutch are as expected. The great likelihood of leaving home for marriage among men from the mixed second generation is striking. Leaving home for union formation becomes an important pathway for the mixed generation for the age range 22–27.

Having a job leads to a significantly higher risk of leaving home for any destination (though not always significantly), while a higher income induces a greater likelihood of leaving home for union formation but a lower departure rate for independent and shared residence at younger ages. This latter result is difficult to interpret because we controlled for enrollment in higher education. Receiving a benefit increases the likelihood of leaving for independence but not for other pathways. Enrollment in higher education has a great positive effect on leaving home for independence and shared residence, as expected. This effect is strongest at young ages. Also as expected, the probability of leaving home for union formation is lower for women who are enrolled in higher education, but surprisingly the opposite holds for young men.

The employment of the parents has a clear negative impact on the likelihood of leaving home, particularly for independence and shared residence at younger ages; this effect is less clear for older ages. In line with previous evidence for the Netherlands, a higher parental income is associated with a greater risk of leaving home for any pathway for both younger and older ages. The impact of the father’s income is often more pronounced than that of the mother’s income. A higher quality of the parental home, indicated by a higher value of the home, is associated with a significantly greater likelihood of leaving home for independence and shared residence at younger ages but a smaller likelihood of leaving for union formation. Possibly, the feathered-nest effect prevails for union formation, whereas the resource effect of housing value prevails for leaving home to live without a partner.

Children of never-married and divorced mothers are more likely to leave home for independence and shared residence at young ages; those with divorced mothers are also more likely to leave for independence and shared residence at older ages. This tendency does not, surprisingly, hold for leaving home for union formation. Children of widowed mothers are less likely to move along all pathways, but particularly for independence and union formation, at both young and older ages. An age differing more from that of the mother, and to a lesser extent the father, leads to a lower risk of leaving home.

The effects of siblings living in the parental home are mostly insignificant for the younger ages, but these effects are positive for leaving home for shared residence and negative for union formation and independent residence for the older ages. This finding suggests that competition for parental resources leads young adults from larger families to postpone the departure from the parental home to live alone or with a partner but encourages sharing with others. The number of siblings living out of the parental home has, in general, a positive effect on the risk of departure along any pathway.

There are also some associations between leaving home and contextual variables. Young people from neighborhoods with higher housing values are less likely to leave home than those who live in neighborhoods with a low mean value of dwellings. The smaller likelihood of departure among those whose parental home is in Amsterdam (for women at young ages and for both men and women at older ages) is noteworthy, whereas a higher departure rate for independence is observed for those whose parental home is in The Hague.

Estimates by Ethnic Origin

The results of the separate models for Moroccans, Turks, and Dutch are presented in Table 5. It is striking that the baseline hazard of leaving home for union formation increases more strongly with age for Dutch than for Moroccan or Turkish youth. This difference is completely in line with a cultural tradition of early marriage among Turks and Moroccans.

As expected, the risk of leaving home for union formation is substantially greater for women than for men. We expected this difference to be greater for Mediterranean youth than for the Dutch (Hypothesis 2a), but this expectation is not confirmed. Women leave home for independence and shared residence at a faster rate than men. In accordance with Hypothesis 2b, this difference is smaller for Turks and Moroccans than for the native Dutch.

With respect to differences between first- and second-generation Turks and Moroccans, the risk of departure tends to be lower for the second generation with one foreign-born parent, although the estimates are statistically significant only for independence. No significant differences are observed between the first generation and the second generation with two foreign-born parents. This outcome suggests that leaving home by the second generation with a mixed background is closer to that of the Dutch than first-generation and second-generation migrants, which implies partial support for the third hypothesis.

The estimates for many other variables are similar for the three categories. For some other variables, there are noteworthy differences. Several enhancing effects of individual and parental resources on leaving home for independence and shared residence are only found for the Dutch (individual employment, parental income, value of the parental home). The positive impact of having a divorced mother on leaving home for any pathway is only found for the Dutch; for Turks and Moroccans this impact is negative on the risk of departure for independence. These findings suggest that the most important factors usually put forward to explain leaving home in Western countries—individual and parental resources, family structure, and the quality of the parental home—do not necessarily work in the same way for immigrants and their descendants as they do for natives. At the same time, the differences in effects are limited and are not capable of explaining the earlier timing and higher incidence of leaving home among Moroccans and Turks than among the Dutch.

Conclusions

This study examines differences between youth from migrant families and Dutch youth in the timing and pathways out of the parental home, applying discrete-time survival models with three competing risks standing for three pathways of leaving home: union formation (distinguished further into marriage and cohabitation in an additional analysis), independence, and shared residence. The administrative panel data used were derived from the Social Statistical Database. Two entire birth cohorts, aged 16 and 22 in 1999, were followed from 1999 until 2004. These population data ensured that our analysis did not suffer from such major methodological problems as small numbers, selectivity, and unobserved heterogeneity. These administrative data reflect the formally registered dynamics of the population. In general, the distance between registered and true residential addresses is probably not great in the highly organized Dutch society. However, we cannot rule out that precisely at young adult ages some measurement error occurs owing to less accurate registration.

We aimed to gain insight into the extent to which migrant youths’ leaving home is consistent with cultural norms in the home country, adaptation to the host society, or the specific position of migrants in the host society. From theoretical arguments based on differences in cultural norms and patterns of leaving home in the origin countries, we derived three hypotheses: (1) on ethnic differences in the timing of leaving home along the various pathways; (2) on gender differences; and (3) on differences between first-generation, second-generation, and mixed second-generation migrants and the Dutch. Our findings confirm the popularity of early marriage as a route out of the home among non-Western migrants, just like in the origin countries. But at the same time, we present evidence of a strongly deviant timing of leaving home to live alone for migrant groups compared with their counterparts in origin countries.

Non-Western migrant young men and women leave the parental home at younger ages than their Dutch counterparts. The most deviant pattern is observed for Mediterranean (Turkish and Moroccan) migrants. In line with our first hypothesis, Mediterranean migrants leave home for union formation (particularly marriage) and shared residence significantly earlier than the Dutch. But, in contrast with that hypothesis and in line with the alternative to it, these migrants also leave home significantly earlier to live alone independently. This finding contradicts the findings of earlier Dutch studies based on small surveys (Bolt 2002; De Valk and Billari 2007). Our findings are also at odds with the suggestion that migrant youth tend to coreside with parents because they provide the most of household income (Glick and Van Hook 2002).

Clearly, the differences in leaving home between Turkish and Moroccan youth compared with the Dutch cannot be fully explained from their more traditional cultural norms. In particular, the greater likelihood of leaving home to live alone independently among Turks and Moroccans than among the Dutch is surprising. This finding can only partly be ascribed to effects of the opportunity structure enriched by the Dutch welfare state, which counteract part of the effects of differences in cultural norms. There are no indications that migrant families have particular strategies with regard to work or geographical mobility that would lead their youth to leave home early. It is therefore likely that the difficult position of young adults from migrant families related to discomfort in the parental home and inter-generational tensions leads many to leave home early.

In contrast with the second hypothesis, gender differences in leaving home for union formation are less pronounced for Turks and Moroccans compared with the native Dutch. But we do see more traditional behavior among Turkish and Moroccan women, in the sense of more marriages directly from the parental home as opposed to less cohabitation. In contrast with women, Moroccan men seem to have a relatively high likelihood of leaving home for unmarried cohabitation, which might be an indication of unconventional behavior among these men. The common finding that women leave home earlier than men for independence and shared residence is less pronounced for Turks and Moroccans than for native Dutch, which is in accordance with stricter cultural norms for women than men among Mediterranean migrants. There are also some indications that the leaving-home behavior of second-generation migrants, particularly those with one foreign and one Dutch parent, is more similar to that of Dutch youth than to that of first-generation migrants.

Our findings seem to suggest that despite differences in marriage patterns between youth from migrant families and Dutch youth, part of the changes associated with the second demographic transition are not exclusive to native Dutch but also affect migrants. The early departure of migrant youth can also be regarded as a strong indication for the adaptation of migrants into the Dutch society.

Acknowledgments

This research is supported by the Netherlands Organisation for Scientific Research (NWO), VICI Grant No. 453-04-001. The article is part of a joint project with Statistics Netherlands; we are grateful to the Department of Total Statistics for allowing us the use of the Social Statistical Database. We also thank to the anonymous reviewers for comments on earlier versions of this article.

Note

References