1 Introduction

This review addresses the role of institutional contexts in family-related processes and outcomes, taking a cross-national comparative perspective with a focus on “Western”—that is, demographically advanced—societies. We will concentrate on institutions manifested in family policies or family laws, which are embedded in more general configurations of policies, ideologies, and institutions, often referred to as family regimes (Cooke and Baxter 2010, p. 516).

Family policies are shaped by social norms and expectations (e. g., regarding gender roles and responsibilities in the family), but they usually do not directly regulate family life. Rather, family policies support specific types of families or partnerships (e. g., marriage), whilst placing others at a disadvantage (e. g., unmarried cohabitation, which is not illegal but—in the German case, for instance—does not benefit from income tax splitting). This sets incentives for certain behaviors but does not actually prescribe them. Family law, on the other hand, is a more direct expression of norms, consisting of “enforceable […] rules that draw [for example] the boundaries between licit and illicit sex, lay down the grounds for the establishment of maternity and/or paternity and for the membership of kin groups, and define the socially-sanctioned obligations and legitimate expectations of household members and kin” (Willekens 2003, p. 73).

In this article, we will focus on outcomes in the pivotal domains of partnership dynamics (Sect. 2), fertility (Sect. 3), children’s wellbeing (Sect. 4), and intergenerational relations (Sect. 5), as well as on variations therein by institutional context. Institutional effects on the gendered division of labor will not be considered here in their own rightFootnote 1, but only insofar as they are linked to the four family-related domains along which we organize our review.

2 Partnership Dynamics

Both union formation and union dissolution have been highly institutionalized throughout human history, mostly by regulations concerning marriage and divorce (e. g., Goody 1983; Rosenbaum 2014). These regulations reflected economic benefits and constraints as well as social and religious norms affecting, for example, mate selection (“Who marries who?”), age at marriage, as well as individuals’ chances of marrying at all. An important and longstanding geographical pattern that emerged from variations in such regulations was described by Hajnal (1965), who observed that late and non-universal marriage had prevailed in Northwestern Europe for centuries, whereas marriage had remained early and near universal in South and Eastern European countries. Only marriage legitimized a heterosexual couple’s intimate relationship, whereas unmarried couples remained outside of legal jurisdiction. The extent to which the cultural and demographic divide along the so-called “Hajnal line”—ranging from Trieste to St. Petersburg—has continued to exist in the late 20th and early 21st century is subject to an ongoing debate (see Steinbach et al. 2016 for a recent contribution).

In the economically prosperous and politically conservative period following World War II, a pattern of early and almost universal marriage initially gained dominance in Western Europe and North America. When this “golden age of marriage” came to an end in the late 1960s and early 1970s, the Second Demographic Transition (e. g., Lesthaeghe 2010), and its underlying economic, social, and ideational shifts, brought about significant behavioral changes in many parts of Europe and America which have often been described in terms of a “deinstitutionalization” of marriage (e. g., Cherlin 2004; Lauer and Youdanis 2010): Age at first marriage started to increase steadily and substantially, whereas marriage rates decreased (in tandem with increasing divorce rates), to then stabilize at low levels. These developments occurred first in the Scandinavian countries, whereas the Mediterranean countries and—to some extent—the US (e. g., Raley 2001) clearly lagged behind. Women’s total first marriage rate peaked in 1964 in Germany, declined by about half until 1991 (from 111 to 57 per 100 women), and has shown only minor fluctuations since. Women’s age at first marriage increased from 23 in 1964 to almost 26 in 1991, and to over 30 in 2015 (Federal Institute for Population Research 2017).

These changes—including the rise in divorce—have been paralleled by an increase in singlehood (accumulated over the individual‘s life course and in the population; e. g., Bellani et al. 2017) and, importantly, by a rise in the prevalence of non-marital cohabiting unionsFootnote 2, particularly in Western (European) societies. In many cases, cohabitation has become more than just a precursor to marriage but has rather evolved as a long-term alternative to marriage (e. g., Hiekel et al. 2014). Within Europe, Noack et al. (2014, p. 21) identify three distinct geographical clusters in the population aged 18–55 years: The first group, mainly consisting of South-Eastern European countries, exhibits a traditional pattern characterized by about 60% of married people, with only 5% or less of the total population cohabiting. The second cluster of predominantly Western and Central European countries constitutes a middle group, with around 50% married and about 10% of cohabiters. The third group, comprising the Nordic countries and France, is characterized by a high proportion of roughly 20% cohabiters in the population, whereas not more than about 40% are married.

Variations in gender equality have been suggested to be a main driver of cross-national differences in the proportions of married, cohabiting, and single individuals. With regard to lifelong singlehood, for example, the multilevel analysis by Bellani et al. (2017) provides evidence that permanently living without a partner is concentrated within countries where traditional gender values have waned, but gender egalitarianism remains poorly diffused. Cooke and Baxter (2010, p. 524) note that there is a macro-correlation between men’s and women’s aggregate economic equality and union type in the sense that “marriage is more prevalent in male breadwinner family regimes such as Italy, whereas cohabitation is more prevalent in regimes supporting greater gender equality such as Sweden […].” Within more equal gender settings, however, we observe a micro-correlation suggesting that women with greater individual resources tend to opt for marriage rather than for cohabitation.

It is interesting to note that the potential role of policies and legislations has so far often been neglected in investigations of cross-national variations in the prevalence of cohabiting unions (Perelli-Harris and Sánchez Gassen 2012). One reason for this might be that it is difficult to establish the extent to which changes in policies and laws are the cause or the consequence of the demographic phenomena to which they refer (e. g., Bradley 2001; Eekelaar 2010). The far-reaching legal recognition of cohabitation in contemporary Western societies has clearly lifted much of the social and economic pressure to marry that previous generations of couples had borne. However, the legal situation of cohabiters still varies widely throughout Europe (for a comprehensive overview see Perelli-Harris and Sánchez Gassen 2012): Even though Norway and Sweden have not formalized cohabiting unions as registered partnerships (unlike France and the Netherlands), they are nonetheless among the most advanced countries in terms of the legal harmonization of cohabitation and marriage. GermanyFootnote 3 and Switzerland represent the other end of the continuum, as they “have been the most reluctant to equalize cohabitation and marriage, or even to recognize cohabitation” (Perelli-Harris and Sánchez Gassen 2012, p. 463; also see Bradley 2001). Differences pertain to rights and responsibilities both during the union (e. g., the right to co-insure a partner in the public health insurance system, or the obligation to support each other financially) and after union dissolution (e. g., regarding the division of property, the obligation to pay alimony, or—after the death of a partner—entitlements to inheritance). However, even in countries with high levels of recognition and actual cohabitation (such as France or Sweden), attitudes towards cohabitation are not unambiguously positive, and the value attached to marriage remains high (e. g., Noack et al. 2014; Treas et al. 2014).

Marriage is one precondition for divorce; the other is the recognition of divorce as a legal act. By 1950, most European countries permitted spouses to divorce (with Ireland being a noteworthy exception, legalizing divorce as late as 1997), but restrictive divorce requirements and procedures still often made it difficult or costly for married couples to legally separate. This was alleviated by the introduction of “no-fault” grounds for divorce (established in most countries by the middle or in the second half of the 20th century) and, subsequently, by a shift from laws requiring mutual consent to those permitting unilateral divorce (occurring mainly in the 1970s and 1980s; for an overview of legal reforms in a variety of countries, see Perelli-Harris et al. 2017, Appendix). Moreover, legal practice—that is, the de facto divorce regime—has been shown to exhibit a significant influence on divorce rates (e. g., Eekelaar 2010; Kneip and Bauer 2009).

Some legal changes had direct effects on divorce rates.Footnote 4 Prior to the introduction of divorce as a legal opportunity to exit marriage, official divorce rates were obviously zero (which does not of course mean that marital breakdown did not take place). Another example is the prescription of a one-year separation period before divorce, which was introduced in West Germany during the late 1970s and resulted in a substantial short-term decline in divorce (see Federal Institute for Population Research 2017). Caution is however necessary in order to avoid confusing the effects of de jure changes in divorce laws with other underlying trends, such as the increase in cohabitation (see Perelli-Harris et al. 2017). Moreover, and finally, governments might have changed divorce laws because many couples had already separated.

A relatively recent and important development is the emergence of same-sex marriage as a “new social phenomenon” in a number of Western countries (e. g., Chamie and Mirkin 2011; Festy 2006). Whereas Denmark legalized “registered partnerships” as early as 1989, the Netherlands was the first country to allow gay and lesbian couples to actually marry in 2001. Many US Federal States and European countries followed suit in the years that followed. Germany adopted a so called “Lebenspartnerschaftsgesetz’”(Life Partnership Act) in 2001, which enabled couples to obtain legal recognition for their union through a registration procedure that was distinct from marriage, but still provided them with benefits very similar to those received by married opposite-sex couples. This law also regulated child-related issues in same-sex partnerships, particularly custody and adoption rights (see Rupp and Haag 2016). Germany eventually legalized same-sex marriages in 2017.

In summary, marriage, cohabitation, and divorce continue to be subject to strong legal regulation (determining, for example, at which age or under which conditions the transition into a specific state is possible). However, there is little evidence to suggest a direct impact of family policies or family law on changing partnership dynamics in “Western” societies. Marriage and divorce obviously have to be legal opportunities: Gay marriage, for example, was not possible in Germany before 2017, and divorce was not possible in Ireland before 1997. But de jure changes in family laws might be a consequence rather than the cause of changes in legal practices and the demographic phenomena to which they refer. Moreover, whether a couple chooses to live in a marital or non-marital union appears to be influenced (at least) as much by a country’s level of gender equality as by the extent to which marriage and cohabitation are legally harmonized.

3 Fertility

Against the background of sustained below-replacement fertility in demographically advanced societies, the role of family policies in childbearing behaviors has received considerable attention (for reviews see Bujard 2016; Gauthier 2007). Welfare state institutions may intentionally affect the timing and quantum of fertility (as a consequence of pronatalist family policies), or they may do so unintentionally (as a consequence of, for example, labor market policies affecting fertility through employment decisions). Even though there is a plethora of fertility-related policy measures, the core “family policy package,” on which we will focus in this section, has been suggested to consist of three main types of policy instruments, namely: financial transfers, paid leave, and childcare services (e. g., Luci-Greulich and Thévenon 2013).

Drawing primarily on economic—or, more generally, rational choice—approaches to fertility (e. g., Werding 2013), it is argued that “[f]amily policies potentially contribute to re-increases in fertility as they can reduce the costs of fertility, either in monetary terms or in terms of opportunity costs.” (Luci-Greulich and Thévenon 2013, p. 390). Direct compensation for the economic costs of children usually comes in the form of cash benefits and/or fiscal transfers to families. An early macro-level time-series analysis covering 22 Organisation for Economic Co-operation and Development (OECD) countries over the period 1970–1996 finds minor positive effects of cash benefits on the total period fertility rate (Gauthier and Hatzius 1997). This result was corroborated more recently by Luci-Greulich and Thévenon (2013), whose study was based on 18 OECD countries in the period 1982–2007. These effects, however, seem more obvious when the timing of births rather than the quantum of fertility is considered.

Research based on microdata generally confirms these findings, but also indicates a varying effect of cash benefits by birth order (e. g., Aassve and Lappegård 2009, for Norway; Laroque and Salanié 2004, for France; Vikat 2004, for Finland). A noteworthy exception is Kalwij (2010), whose cross-nationally comparative analysis of data from the European Social Survey showed no significant impact of more generous family allowance programs on the timing of births or individuals’ completed fertility. Also in the German context, analyses of the role of child benefit (Kindergeld) payments tend to provide no or at most weak evidence of overall effects on fertility (Bujard 2016, p. 627). In 2007, however, the German government introduced the new parental allowance (Elterngeld), which replaced means-tested parental leave benefits targeted at lower-income families with payments related to pre-birth earnings. Analyzing administrative microdata, Raute (2018) indeed identified an increase in fertility following this reform, driven mainly—as intended—by women at the middle and upper end of the education and income distributions (also see Bujard and Passet 2013).

A similar policy was introduced earlier in Sweden, where Andersson et al. (2006) did not find any major educational differentials in the reaction to the reform. The authors’ primary interest, however, lay not in the role played by parental leave benefits, but in the duration of paid parental leave (specifically the eligibility interval during which benefits may be retained). Confirming results of a previous study by Hoem (1993), their analysis of population register data provides evidence that the extension of the eligibility interval set incentives to have another child while still being on parental leave. The Swedish leave policy reform was thus interpreted as a “speed premium” affecting the timing of births. Similar effects are not only found in other Nordic countries (e. g., Rønsen 2004, for Norway and Finland), but also in two Austrian studies (Hoem et al. 2001; Lalive and Zweimüller 2009).

A more recent innovation in parental leave policies is the introduction of “daddy months,” dedicating some share of the total leave duration to fathers. The first countries to establish this policy were the Nordic ones, but others—such as Germany—followed suit (see for example Geisler and Kreyenfeld 2011). Whereas the main aim was to promote gender equality, Duvander et al. (2010) showed—based on an analysis of register data—that fathers’ take-up of parental leave is positively associated with continued childbearing in Sweden and, even more so, in Norway.

Despite these findings, it is important to note that parental leave policies are not designed to influence parents’ fertility behavior directly, but that they particularly aim at enhancing children’s wellbeing (see Sect. 4) and the compatibility of childrearing and female employment (e. g., Ellingsæter 2009). This latter issue is important because Brewster and Rindfuss (2000, p. 271), for example, concluded from their review of the literature that “women’s labor force participation lies at the heart of most explanations of fertility and fertility change,” and that the frequently observed inverse “association between fertility and women’s labor force activity reflects the incompatibility between caring for children and participation in economically productive work that typifies industrialized societies.” Even though access to affordable, high-quality childcare has been proposed as one of the most important structural conditions to solve this compatibility problem, empirical studies employing multilevel data provide inconclusive evidence regarding its effect on fertility.Footnote 5

In Southern European lowest-low fertility, familialistic welfare state contexts, Del Boca (2002; for Italy) and Baizán (2009; for Spain) found that more comprehensive availability of formal childcare had a positive effect on fertility. Rindfuss et al. (2010) report similar findings for a somewhat different demographic and welfare state context, namely Norway, where greater childcare availability increases transition rates at every parity, and thus also completed fertility. However, such an effect was neither found in earlier Norwegian research (Kravdal 1996; also see Rønsen 2004), nor in Andersson et al.’s (2004) study of continued childbearing in Sweden. For Germany, Hank et al. (2004) found that the availability of public childcare had a positive impact on Eastern German women’s transition to the first child, whereas this was not the case for their Western German counterparts. However, this analysis based on Socio-Economic Panel (SOEP) data covered a rather short window of observation around the turn of the millennium, and was thus based on a relatively small number of events. More recently, Bauernschuster et al. (2016) exploited the temporal and spatial variation in childcare coverage induced by a significant expansion of childcare slots for young children in the mid-2000s. Matching information from birth registration records with county-level data on childcare coverage, their analysis suggests that a ten-percentage-point increase in childcare coverage leads to an increase in birth rates of almost three percent. The authors not only claim that their findings actually reflect a quantum effect, but also that investments in public childcare are more efficient with regard to raising fertility than expansions in child benefit expenditures (Bauernschuster et al. 2016, p. 1002).

This latter finding is consistent with Kalwij (2010, p. 517), whose findings from 16 Western European countries indicate “that increased expenditure on family policy programs aimed at empowering women through opportunities to combine family and employment—thereby reducing the opportunity costs of children—generate positive fertility responses. More specifically, extending maternity and parental leave as well as childcare provision causes women to have children earlier in life, and to have more children.” It therefore seems important to acknowledge that it is a combination of policy instruments that is most likely to facilitate the choice to have children, but that not all measures have the same weight (see also Harknett et al.’s (2014) analysis of the role of countries’ broader “family support environments” in individuals’ childbearing plans and actual childbearing behaviors).

Finally, alongside the abovementioned set of family policy instruments there are important legal regulations potentially affecting the number of children parents may have, especially if abortions, adoptions, and the use of assisted reproductive technologies (ART) are considered: Abortion has been discussed as a possible substitute to modern contraception in less developed countries, and its legalization has thus been suggested to potentially impact fertility (see Gutierrez Vasquez and Parrado 2016; Miller and Valente 2016 for recent investigations). Considerable variation in both legal restrictions and rates of termination of pregnancies continues to exist in Europe (Gissler et al. 2012; see David 1992 for a historical account). However, countries with unrestricted access to early termination of pregnancy do not exhibit higher rates than countries with more restricted access. Germany, for example, which allows early terminations of pregnancies without legal indication upon women’s requests, reported only 6 terminations per 1000 women aged 15–49 in 2008 (compared to an EU average of 10/1000; see Table 1 in Gissler et al. 2012).

The prevalence of adoptions varies substantially across countries, being relatively high in the US and comparatively low in Germany, where the number of adoptions has continuously declined—to a total of 3812 in 2015—since the 1980s (Bovenschen et al. 2017). Whereas some of this decline seems to be attributable to more generous state support for families, advances in birth control and reproductive medicine, as well as more liberal abortion laws, higher adoption rates in other countries also suggest an important role played by a lower level of social acceptance and more complicated legal regulations on adoptions in Germany (for a review of the latter see Reinhardt 2017). The number of live births following ART treatment in Germany is substantially higher than the number of adoptions, with a peak of more than 18,000 in 2003, followed by a sharp decline in 2004 and a subsequent recovery to roughly 14,000 in 2012. The decline in the number of women treated, treatment cycles, and—consequently—in live births, was not due to changes in the overall legal framework for ART, but resulted from a significant reduction in the reimbursement of the costs of treatment by statutory health insurance (for a detailed overview see Trappe 2017). Variations in reimbursement levels—rather than legal regulations—have also been suggested to be the main driver of cross-national differences in the use of ART across Europe. Usage is particularly high in Denmark, Slovenia, and Spain, where the cost of treatment is completely covered by national health plans (Präg and Mills 2017). Even though the numbers of both adoptions and successful ART treatments are moderate in absolute terms (compared to, for example, a total of more than 730,000 births in Germany in 2015), they are likely to become increasingly relevant phenomena against the background of further medical advances, a sustained delay in childbearing, and the liberalization of same-sex parenthood (e. g. Waaldijk 2009).

In summary, whereas there is some evidence to suggest an impact of specific policy instruments on the timing (financial transfers, paid leave) and quantum (public childcare services) of childbearing, combinations of such instruments aiming to empower women appear to be most effective with regard to the aim of raising fertility. Moreover, legal regulations are important to shape the conditions under which, for example, induced abortions or the use of assisted reproductive technologies may take place, but they do not seem to have a major quantitative impact on the fertility outcomes that are derived from such practices.

4 Children’s Wellbeing

Whereas families constitute the most important context for children and their development, they are affected both directly and indirectly by institutional contexts shaping the circumstances under which they grow up. The relevant policies and laws here are often the same ones affecting parents’ decision to have children, as well as the consequences resulting from this decision (especially in terms of labor force participation; see Sect. 3). A major concern is the role of such welfare state institutions in children’s wellbeing—health, educational opportunities, poverty risks—and how they might buffer, for example, adverse effects of family disruption (for a comprehensive analysis see Engster and Stensöta 2011).

A central question is who cares for children (and under what conditions). Parental leave regulations provide opportunities and set incentives for parents—primarily mothers, but increasingly for fathers as well (e. g., Boll et al. 2014; Bünning 2015)—to stay away from the labor market for some time and provide full-time care for their children. Longer leave entitlementsFootnote 6 may potentially affect a variety of child outcomes. To begin with, there might be health effects resulting, for example, from reduced maternal stress or prolonged breastfeeding.Footnote 7 Macro-level evidence from a number of OECD countries (e. g., Patton et al. 2017; Tanaka 2005) suggests that longer job-protected, paid parental leave substantially decreases mortality among infants born to eligible mothers (with additional smaller positive effects on birth weight). Whereas Tanaka (2005) did not identify any significant effects if leave was provided without job protection or adequate paymentFootnote 8, Rossin (2011) found that even the introduction of 12 weeks of unpaid maternity leave mandated by the 1993 Family and Medical Leave Act in the US led to small increases in birth weight and a significant decline in infant mortality. Studies assessing other specific health outcomes (such as infections, chronic conditions, or hospital admissions) using microdata did not systematically find causal effects of the length of parental leave on younger children’s wellbeing (e. g., Baker and Milligan 2008 for Canada; Beuchert et al. 2016 for Denmark), but recent evidence from Australia indicates that paid leave entitlements might reduce disadvantaged children’s probability of having multiple ongoing health conditions (Broadway et al. 2017).

The more general institutional setting in which a leave policy is enacted obviously matters: “a reform expanding paid leave from twelve to fifteen months in a setting with subsidized child care and universal health insurance […] is dramatically different from one that provides six weeks of paid leave for the first time in a setting where neither child care nor health insurance is guaranteed” (Rossin-Slater 2018, p. 14). This might also, and particularly, be the case, if children’s educational outcomes are considered, given that countries’ educational systems (including the arrangements that they make for preschool public childcare) vary widely. However, recent micro-level evidence from institutional contexts as diverse as, for example, Norway (Dahl et al. 2016) or Austria (Danzer and Lavy 2018), does not suggest any significant effect of parental leave extensions on schooling outcomes (such as test scores or high school dropout ratesFootnote 9). In a comprehensive study of several parental leave reforms in Germany, Dustmann and Schönberg (2012) showed: (a) that the expansion in paid leave from 2 to 6 months in 1979 did not increase children’s average years of schooling, (b) that the expansion from 6 to 10 months in 1986 did not substantially raise the probability of completing a high-track school (i. e., Gymnasium, a grammar school equivalent), and (c) that the expansion in unpaid leave from 18 to 36 months in 1992 even seems to have lowered children’s educational attainment. Finally, in an analysis of macrodata from 20 OECD countries, Engster and Stensöta (2011, p. 84) found “little long-term effect of family policy regimes on educational achievement (test score), but a significant correlation between family policy generosity and higher educational attainment (remaining in school longer).”

Importantly, some studies also point to differential effects caused by, for example, parental education: Liu and Skans (2010) identified a positive effect of prolonged parental leave for children of well-educated mothers in Sweden, and Cools et al. (2015) report that Norwegian children’s school performance improved if their fathers took paternal leave, especially when they had attained a higher level of education than the mother had. Another important distinction is made by Rossin-Slater (2018, p. 15; italics not in the original), who concludes from her review of the literature that “extensions in existing paid leave policies have had little impact on children’s well-being, [while] the evidence suggests that the introduction of short paid and unpaid leave programs can improve children’s short- and long-term outcomes.”

Whereas leave programs foster parental childcare at home, many countries have also expanded the provision of public daycare for children, and a growing number of studies investigate the effects of center-based early childhood education and care programs with regard to children’s school achievements as well as their cognitive and socio-emotional development (for reviews see Anders 2013; Burger 2010). Cross-national comparative studies covering a broad range of economically developed societies point to a generally positive micro-level correlation between attendance of pre-school institutions and subsequent PIRLS or PISA test scores (Cebolla-Boado et al. 2017; Schütz 2009). The strength of this association seems to vary by country, depending on the “structural” quality of preschool education: It tends to be strongest in contexts with higher spending on pre-primary education per pupil, larger shares of children attending privately managed pre-primary institutions, as well as higher relative pay and higher levels of training for pre-primary teachers (Schütz 2009). Evidence from Anglo-Saxon countries suggests that early childcare is positively associated with test scores at school entry (e. g., Hansen and Hawkes 2009; Magnuson et al. 2007), but that this effect tends to dissipate later on (which is consistent with Spieß et al. 2003, who show that there is no significant relationship between kindergarten attendance and children’s later school placement in the German tracking system). However, even though long-term effects of early educational interventions may be smaller than initial effects, they can still be substantial—especially for children from disadvantaged social backgrounds (e. g., Cebolla-Boado et al. 2017)—if designed properly (e. g., Barnett 2011).

Reducing child poverty, which has been shown to exert substantial adverse short- and long-term effects on a variety of life domains (e. g., Duncan et al. 2012), is another major policy concern. Whereas relative child poverty is as low as 5% in Norway, it exceeds 20% in the US—and is higher than overall poverty in most countries (Smeeding and Thévenot 2016: Fig. 1). Household composition and parents’ labor market participation have been suggested to play a crucial role among childhood poverty drivers. Particularly single mothers and their children almost universally experience elevated risks of poverty. These are highest in the US and substantially lower in welfare state contexts providing strong public cash support as well as work support to increase mothers’ labor earnings (e. g., Smeeding and Thévenot 2016; see also Brady and Burroway 2012). Moreover, when studying child poverty by family structure in a set of five liberal welfare states during the 2008 recession, Rothwell and McEwen (2017) found that children in cohabiting families were less well protected against market instability than those whose parents were married. The authors also show that family benefits in the form of income transfers substantially contribute to reducing poverty among non-married—often fragile—families, whose risk of being poor is again highest in the US. Finally, Engster and Stensöta (2011, p. 84) conclude from their study of OECD countries that “dual earner regimes, combining high levels of support for paid parenting leaves and public child care, are significantly associated with low levels of child poverty.”

A plethora of studies have shown that separation or divorce are associated with a variety of adverse outcomes for children: Alongside increased poverty risks and educational disadvantages, there is also evidence of greater psychological and behavioral problems, as well as a greater propensity to get divorced themselves in adulthood (for reviews see Amato 2000; Härkönen et al. 2017). Whereas such relationships between family disruption and child outcomes are found almost universally, many studies suggest cross-national variation in the strength of the associations observed. Detrimental effects on children’s school achievements, for example, seem to be slighter in family policy contexts that balance out resources between single- and two-parent families (e. g., Hampden-Thompson 2013; Pong et al. 2003).

Moreover, and importantly, child support and custody laws are likely to affect children’s wellbeing after their parents’ separation or divorce (e. g., Del Boca 2003). Child support consists of a regular income transfer from the father to the mother that is often ordered—and legally enforced—because of income disparities between the parents (e. g., Huang et al. 2003; Stirling and Aldrich 2008). With regard to child custody, one needs to distinguish between legal (regulating parents’ decision-making) and physical (regulating parenting time). Sole physical custody usually results in a situation where the child lives with one parent only (most often the mother), thus substantially losing financial and emotional support that was previously provided by the other parent (most often the father). Even though non-resident fathers may still contribute to children’s wellbeing (King and Sobolewski 2006), custody agreements and living arrangements have been shown to have a major impact on fathers’ involvement in their children’s lives (e. g., Swiss and Le Bourdais 2009). Several Western countries have thus revised their custody laws in the past decade, thereby strengthening joint physical custody arrangements that support shared parenting after separation or divorce. Whereas the consequences of such an arrangement (in which the child is supposed to live 35% or more of the time with each parent) are not yet fully investigated, previous research suggests that the wellbeing of children in joint physical custody is at least as high as in sole physical custody (for recent reviews see Baude et al. 2016; Steinbach 2018).

In summary, whereas some studies identify (direct) policy effects on children’s wellbeing, such effects are far from universal. Whether parental leave, for example, affects children’s health or education very much depends on which specific dimension of the outcome is considered (e. g., infant mortality vs. chronic conditions; educational achievement vs. attainment). With regard to education, children from disadvantaged social backgrounds appear to be the ones benefiting the most from early educational interventions, whereas the children of more highly educated parents seem to be the main beneficiaries of parental leave extensions. Moreover, introducing a parental leave program may have a larger impact than extending the eligibility interval, and the long-term effect of early educational interventions, for example, may be smaller than their initial effect. Overall, laws and policies fostering mothers’ and fathers’ active involvement in both parenting and paid work appear to contribute the most to improving children’s wellbeing.

5 Intergenerational Relations

In contradistinction to the notion of “less family” that has sometimes been used to describe the main trends in marriage and fertility observed during the second half of the 20th century (see Sect. 2 and 3 of this review), the “family decline” hypothesis (Popenoe 1993) has been widely rejected as far as intergenerational relations within families are concerned. However, despite high levels of solidarity between family members overall across two or more generations throughout Europe and the US, we also observe considerable variations across welfare states with regard to both upward and downward assistance or transfers. In (Western) Europe, for example, there is a continuum marked by relatively “weak” family ties in the Nordic countries and relatively “strong” family ties in the Mediterranean ones (e. g., Hank 2009). This geographical pattern reflects longstanding variations in cultural characteristics, social norms, and preferences, which are, inter alia, manifested in different policies and legal obligations to support parents or children in need. In more general terms, these have sometimes been described along a “familialism/de-familialization” continuum (see Saraceno and Keck 2010; see also Dykstra 2018).

Requirements to contribute financially to the costs of eldercare for parents (upward intergenerational support) are a prominent and obvious example; see Haberkern and Szydlik (2008, 2010) for a detailed discussion. Consistent with the notion of “de-familialization” (that is, reduced family responsibilities and dependencies), there are no such obligations in the Scandinavian countries.Footnote 10 The “familialistic” Mediterranean countries (as well as many conservative welfare states), however, provide publicly funded services only if the person in need or his or her close relatives—children or in some cases siblings—cannot afford to bear the costs of care themselves. Accordingly, eldercare provided by the younger generation of family members is substantially more common in the latter countries than it is in Northern Europe, where professional services are more readily available and their use is widely accepted. Unfortunately, most countries so far offer only very limited (financial) support for informal carers, and policies to assess their needs are still at an early stage, especially in familialistic settings (see Courtin et al. 2014). Moreover, there are important gender differences in the provision of informal care to elderly parents, which also vary by welfare state context. Daughters are universally more likely to provide care to the older generation, but this gender inequality has been shown to be highest in countries with a high level of intergenerational care, high public spending on old-age cash benefits, low provision of professional care services, high family obligation norms, and a high level of division of labor across gender lines (Haberkern et al. 2015).

Cross-national differences in the provision of childcare by grandparents (downward intergenerational support) have been suggested to result from the interplay between female employment and family policies, specifically the provision of public daycare for children (e. g., Bordone et al. 2017; Hank and Buber 2009). Whereas Scandinavian grandparents are more likely than their Southern European counterparts to provide grandchild care, the latter are more likely than the former to provide intensive (that is, regular) childcare. One explanation for this (seemingly counterintuitive) pattern is that the high level of regularly provided public childcare in Northern European countries creates an opportunity structure that fosters maternal employment, but also requires that grandparents occasionally complement institutional care (e. g., if the grandchild’s mother needs to work extra hours). In Mediterranean countries, on the other hand, the lack of public daycare for children inhibits maternal employment, and there is only limited demand for grandparents to step in because mothers tend to be full-time carers. If, however, a Mediterranean mother seeks gainful employment, she has to rely on grandparents’ support on a regular basis (Hank and Buber 2009; also see Di Gessa et al. 2016).

Clearly, cross-national differences in the use of close kin as providers of child- or eldercare are not driven by legal and structural conditions alone, but also by cultural factors, especially variations in preferences, attitudes, and norms regarding the use of formal care services (e. g., Haberkern and Szydlik 2010; Jaapens and Van Bavel 2012). Moreover, a simple dichotomy distinguishing societies that are characterized by strong (weak) families and weak (strong) welfare state institutions does not provide an adequate concept to explain the more complex empirical patterns that have been observed in recent studies (e. g., Saraceno and Keck 2010). Models postulating a joint responsibility of welfare states and families in the production of social services appear as a powerful alternative to previous simplifications. They allow researchers to transcend (partly ideological) questions such as whether welfare states crowd out families, asking instead how existing needs can be met in the most efficient way and in line with people’s own preferences. Motel-Klingebiel et al. (2005, p. 864) thus argue that in a “situation of ‘mixed responsibilities’, it is possible for formal and informal support systems to be complementary and to take on specialised roles.” Along these lines, Igel et al. (2009, p. 220) showed for example that, in more generous European welfare states, “[p]rofessional providers take over the more challenging, demanding and essential care of the elderly, whereas children tend to give voluntary, less intensive, and less onerous help.”

The interplay between welfare state institutions and families becomes even more complex if the growing shares of non-intact families and non-biological parent–child relationships are taken into account. Laws regulating child custody or alimony payments, for example, have been shown to have long-term implications for intergenerational relations in adulthood: Custody arrangements affect children’s living arrangements (Cancian et al. 2014) and non-resident fathers’ involvement with children (Seltzer 1998), whereas the generosity of alimony payments influences the level of economic distress in non-intact families (Kreyenfeld and Martin 2011). Specifically, Arránz Becker et al. (2013, p. 1133) suggest that more generously provided welfare state support for children “benefits the generally disadvantaged stepchildren especially, and […] may make the socioeconomic situation of stepchildren less conditional on their relationship with the stepparent.” Such institutional effects may ultimately have long-term direct and indirect implications for a variety of interrelated dimensions of (step-)parent–child relationships (see, for example, Steinbach and Hank 2016).

In summary, the provision of care is an important phenomenon at the intersection between families and welfare states. Eldercare (that is, upward intergenerational support) is clearly more directly affected by legal regulations and policies than, for example, the provision of grandchild care (that is, downward intergenerational support). In both cases, however, we observe a complementary relationship of specialized roles that families and welfare states take on in the production of care. Maintaining this balance will be a challenge in a situation characterized by population aging, (partial) welfare state retrenchment, changing gender roles, and increasing family complexities.

6 Conclusion

The aim of this article was to provide an overview of families and their institutional contexts in Western societies, focusing on the role played by family policies and legal regulations in union dynamics, fertility, children’s wellbeing, and intergenerational relations. This makes the topic of our review a moving target with closely interrelated parts: Family dynamics are driven by changing institutional opportunities and constraints, whereas welfare state institutions constantly need to adapt to the changing needs of “new” family forms (e. g., Vaskovics and Huinink 2016).

The studies covered here provide ample evidence of manifold direct and indirect institutional effects on family-related behaviors and outcomes in a variety of domains. A general conclusion that we can draw from this research is that family policy regimes supporting greater gender equality are those under which favorable outcomes—such as higher fertility or greater child wellbeing—are most likely to occur. Importantly though, the effects of specific policies are not always as large, sustainable, or robust as might have been intended or expected beforehand. Evaluating the effectiveness and efficiency of family policy measures and legal regulations thus appears to be an important task for future research (e. g., Bonin et al. 2013; Fichtl et al. 2017).

Whether a pronatalist family policy, for example, has been successful can often not be properly assessed by simply comparing a population’s total fertility before and after the introduction of that policy, even when trying to hold other factors fixed. It may be difficult to actually disentangle the intended or unintended impact of a specific reform from direct or indirect effects of a country’s general institutional (“family regime”) set-up and possible parallel changes therein. Moreover, the same policy might affect individuals’ fertility in different ways, depending on whether we consider its timing or quantum, first- or higher-order births, marital or non-marital childbearing. In addition, there might be cross-level interactions between policies and individual characteristics, such as education, resulting in differential effects for various subpopulations. And, eventually, it may be difficult to establish whether certain legal regulations (e. g., the introduction of “daddy months”) are primarily a cause or a consequence of changing sociodemographic behaviors (such as fathers’ greater involvement in childrearing).

These methodological challenges, amongst others, call for great caution to be applied when interpreting the results reported in empirical studies as causal effects. Some studies investigate, for example, consequences of individuals’ use of parental leave (and/or the uptake of benefits; e. g., Aassve and Lappegård 2009), whereas other—econometrically more rigorous—studies account for exogenous changes in individuals’ eligibility to take paid leave (e. g., Dahl et al. 2016; Dustmann and Schönberg 2012; Lalive and Zweimüller 2009), thereby avoiding potential selectivity issues. Many of the latter (quasi-experimental) studies apply a regression discontinuity or differences-in-differences design, exploiting within-country institutional variation over time rather than between-country institutional variation, which is a common identification strategy in multilevel research.Footnote 11 Even though multilevel modeling has nowadays become a standard tool in cross-national comparative research, there is also an increasing awareness of its limitations, resulting from the necessity of a sufficiently large number of aggregate-level observations in order to obtain reliable estimates of parameters summarizing country effects (e. g., Bryan and Jenkins 2016; Schmidt-Catran et al. 2019). Because multilevel analysis is thus not a panacea, it seems important to further explore the potentials of alternate research designs for “small n” cross-national studies. “Most similar/most different systems” designs, for example, are well established in political science (see Anckar 2008), but have so far rarely been employed in family research (for an application see Berninger 2013).

Inevitably, our review has several limitations: First, we did not consider any “non-Western” societies (see the contributions in Hill and Kopp, 2015, Section I, for overviews of families in African, Asian, and Latin American contexts). Another “geographical” restriction is that we did not systematically account for potentially relevant social contexts at sub-national levels of spatial aggregation (e. g., Hank and Huinink 2015). Second, we exclusively considered institutions manifested in family policies or family laws. The educational system, however, is an important example of other kinds of institutional contexts that might also play an important role in individuals’ demographic behaviors, especially partner choice and family formation (e. g., Blossfeld and Huinink 1991; Blossfeld and Timm 1997). Third, and finally, it was beyond the scope of this review to thoroughly incorporate the recent discussion about the diffusion of gender-egalitarian norms, the ongoing “gender revolution,” and their interaction with welfare state institutions in shaping changing family behaviors (see Esping-Andersen and Billari 2015; Goldscheider et al. 2015). This latter issue in particular deserves adequate attention in future investigations.