Introduction

The increase of single motherhood and parental divorce has become of the most important social transformations experienced by Western societies in the last half-century. This change has not been even across these societies; it has started later and moved slower in some places (Härkönen Reference Härkönen2017). Hence, there are substantial cross-national differences in the percentage of nontraditional living arrangements (Pong, Dronkers, and Hampden-Thompson Reference Pong, Dronkers and Hampden-Thompson2003). It has been demonstrated that parental divorce and growing up in a single-mother family have negative effects on children’s well-being (McLanahan, Tach, and Schneider Reference McLanahan, Tach and Schneider2013), and several studies have tested to what extent these effects diverge between countries and over time (see Bernardi et al. Reference Bernardi, Härkönen, Boertien, Rydell, Bastaits and Mortelmans2013 for a review). It was expected that these negative associations would be lower in countries and time periods where nontraditional family forms are more common, where there is a greater acceptance of new family forms, and where there are generous policies for single-mother families (Gähler and Garriga Reference Gähler and Garriga2013). Surprisingly, most studies that address the variation across countries and over time show that the effects of parental divorce and family structure on children’s well-being have been relatively constant (see Bernardi et al. Reference Bernardi, Härkönen, Boertien, Rydell, Bastaits and Mortelmans2013). Some studies have even found that the impact of parental divorce has increased over time, contradicting most expectations that a reduction in stigma and an increase in father involvement might mitigate the effects (Bernardi et al. Reference Bernardi, Härkönen, Boertien, Rydell, Bastaits and Mortelmans2013).

A possible explanation for why the consequences associated with single parenthood have not decreased is that over time, the prevalence of single motherhood has increased faster among those with lower levels of education (Gähler and Garriga Reference Gähler and Garriga2013). Research documenting this has mainly focused on the United States and has not considered whether or not the increasing polarization of family structure by educational level diverges between countries in different time periods (Garriga, Sarasa, and Berta Reference Garriga, Sarasa and Berta2015; McLanahan and Jacobsen Reference McLanahan, Jacobsen, Amato, Booth, McHale and Van Hook2015). McLanahan (Reference McLanahan2004) showed that in Canada, Finland, Germany, the Netherlands, Sweden, and the United Kingdom, less-educated women were more likely to be single mothers, while in Italy it was more educated women who were more likely to be single mothers. However, to our knowledge, only four studies have focused on the changes in trends in the educational differences of single motherhood from multiple causes in European countries. Kennedy and Thomson (Reference Kennedy and Thomson2010) show that the probability that a Swedish child spent time in a single-mother family during her childhood increased between the 1970s and 1990s. Garriga and Cortina (Reference Garriga and Cortina2017) showed that between 1991 and 2011, the educational gradient of single motherhood reversed from positive to negative in Spain. Garriga, Sarasa, and Berta (Reference Garriga, Sarasa and Berta2015) have also found that in Italy the relationship between mother’s education and single motherhood was positive in 2005 and became insignificant by 2011. Härkönen (Reference Härkönen2017) is the only study that has observed the educational gradient of single motherhood in different time periods in multiple countries. Using data from the Luxembourg Income Study (LIS) Database, Härkönen showed that “diverging destinies” are not confined only to the United States, but there are nonetheless major cross-national variations. The main limitation of this study, however, is that the educational gradient of single motherhood is not adjusted for mother’s immigration status. Taking this variable into account might substantially affect the results since the percentage of foreign born mothers has increased in most Western countries and, on average, they have a lower educational level than native born mothers (Garriga and Cortina, Reference Garriga and Cortina2017; OECD, 2012).

Several researchers have argued that marked increases in the prevalence of single motherhood among the low-educated together with the well-documented negative effects of parental divorce and growing up in a single-mother family on child outcomes have exacerbated the inequality between children from different socioeconomic backgrounds and different family structures (Augustine Reference Augustine2014; Cherlin Reference Cherlin2005; Härkönen Reference Härkönen2017; Härkönen Reference Juho, Nieuwenhuis and Maldonado2018; McLanahan and Percheski Reference McLanahan and Percheski2008). However, Bernardi and Boertien (Reference Bernardi and Boertien2016) and Bernardi, Boertien, and Popova (Reference Bernardi, Boertien and Popova2014) have argued that this conclusion is only true if a third premise is also true; namely that the consequences of parental divorce and family structure are greater among children of lower socioeconomic background, or that the consequences are the same regardless of socioeconomic background. If instead growing up in a nonintact family entails more negative consequences for children from higher socioeconomic backgrounds, they have claimed that this might actually counterbalance the increase of nonintact families among children from disadvantaged backgrounds. In other words, the increase of parental divorce and single-mother families may reduce inequality in children’s outcomes and life chances between children from different socioeconomic backgrounds if these single-motherhood costs relatively advantaged children more (Leopold and Leopold Reference Leopold and Leopold2016).

Despite the importance of the issue of varying costs of divorce and family structure by family socioeconomic background, it has not received much attention until recently. To date, the research has obtained mixed findings. Some studies have found that higher socioeconomic background can compensate for the negative effects of family structure and parental divorce, but other studies have found that larger negative effects at higher socioeconomic status. Alongside methodological reasons, two other possible explanations for why these studies may not produce consistent results are that they focus on different children’s outcomes and on different countries: The conditioning role of family socioeconomic background may depend on the outcome and country studied.

Overall, this chapter aims to address these gaps in the literature by using data from twenty-one Western countries from the Programme for International Student Assessment (PISA) of 2012. First, we explore to what extent there is a general pattern in Western countries of single motherhood being common among women with less education. Second, we analyze the effects on children of being in a single-mother family on three school outcomes: Standardized math test scores, grade repetition, and truancy. Most cross-national studies on the effect of family structure on school outcomes have only focused on achievement tests, despite evidence of stronger effects of family structure on educational attainment and school behavior outcomes than on test scores (McNeal Reference McNeal1999). In addition, truancy or repeating a grade has negative consequences for children’s educational attainment, plus both are also strongly associated with labor market and socio-emotional outcomes and risk behaviors such as drug abuse or crime (Garry Reference Garry1996; Jones, Lovrich, and Lovrich Reference Jones, Lovrich and Lovrich2011; Range, Yonke, and Young Reference Range, Yonke and Young2011). Third, we look at the heterogeneity of family structure effects by focusing on a specific dimension of family socioeconomic background: Mother’s education.

We use this analytical approach based on the study of different outcomes and countries to address the question of whether the growing number of single mothers in Western countries generally increases or decreases inequality in children’s outcomes and life chances between those from different socioeconomic backgrounds. Answering this question requires knowing: (1) whether single motherhood is generally concentrated among women of lower education in most Western countries; (2) if the effects of single motherhood matter across a range of children’s important educational outcomes; and (3) whether the impact associated with single motherhood depends upon the mother’s education. We argue that even if children of lower socioeconomic status are generally more likely to be in single-mother homes, the retreat from traditional family structures would increase children’s inequality only if there were a consistent pattern across countries and outcomes of single motherhood having consistent negative effects on children’s outcomes and life chances regardless of mothers’ education, or if children with less-educated mothers have greater disadvantages associated with single motherhood. In contrast, if living with a single mother were associated with deeper disadvantage among children of more educated mothers across countries and outcomes, then the retreat from traditional family structures could decrease children’s inequality.

Compensatory Hypothesis and Floor Effect Hypothesis

The sociological literature has developed two general perspectives about the heterogeneity of parental divorce and family structure effects by mother’s education: The “compensatory hypothesis” and the “floor effect hypothesis.” These perspectives are based on diverging interpretations of how various mediators of the effects of family structure on children’s well-being work according to different levels of mother’s education. These mediators are financial constraints, quality of parenting, mother’s psychological well-being, involvement of the noncustodial father, and social support and networks (Amato Reference Amato1993; Sigle-Rushton and McLanahan Reference Sigle-Rushton and McLanahan2004).

The compensatory hypothesis posits that mothers with a higher educational background are better equipped to buffer their children from the negative consequences of growing up in a single-parent family and, consequently, there are no – or few – differences in children’s outcomes by family types among those that have a mother with a higher educational level. On the other hand, this hypothesis states that lower educated mothers are more vulnerable to factors that intensify the negative consequences of growing up in a single-mother family. Single mothers with a low educational level are in a worse position than single mothers with high educational level, and are less likely to mobilize resources to compensate for their children’s disadvantages (Augustine Reference Augustine2014; Leopold and Leopold Reference Leopold and Leopold2016).

With respect to financial constraints, it is well-known that women with more education are more likely to be in the labor market and to be better paid (Pettit and Hook Reference Pettit and Hook2005). Highly educated women may, therefore, already have jobs before becoming single mothers. They also have better opportunities to re-enter the labor market after a period of nonemployment than women with a lower educational level (Drobnič, Blossfeld, and Rohwer, Reference Drobnič, Blossfeld and Rohwer1999). Further, research has found high levels of educational homogeneity within couples in Western countries (Blossfeld and Timm Reference Blossfeld and Timm2003). Consequently, children with a mother with a high educational level have a higher probability of having a father with a high educational level. Couples with high educational levels tend to be wealthier and, even when family income and wealth have to be divided after parental separation, mothers may retain more financial resources than their less-educated counterparts. Finally, resources may also increase mothers’ ability to navigate the legal system on behalf of herself and her child to obtain child support payments. Case, Lin, and McLanahan (Reference Case, Lin and McLanahan2003) show that mothers with a higher level of education have a greater chance of receiving child support payments in high amounts than mothers with a lower education, who often do not receive any child support.

With respect to quality of parenting, Augustine (Reference Augustine2014) argued, that better-educated single mothers are better placed to overcome many family-structure-related barriers to maintaining higher levels of parenting quality. The first barrier is financial resources. Single mothers have less time and energy than mothers in two-parent families, and this is mainly due to task overload since they have to obtain financial resources and take care of their children alone (Astone and McLanahan Reference Astone and McLanahan1991). Mothers with a high educational level have greater financial resources to pay for hiring domestic workers or good quality child care. These mothers also have larger and wealthier social networks that may help by taking care of the children directly or providing them with financial support. Economic resources and the related social networks of highly educated single mothers can help them to minimize their stress and task overload and hence, they may have more time and energy to provide better quality parenting to their children.

A second barrier that affects quality of parenting of single mothers is psychological well-being, and this barrier may be more consequential for less-educated mothers. In fact, research shows that mothers with lower socio-economic resources experience more psychological problems after dissolution of their unions than those with greater resources (Liu and Chen Reference Liu and Chen2006; Mandemakers and Monden Reference Mandemakers and Christiaan2010). Better-educated mothers may also be more conscious of the negative effects of divorce and single motherhood since they may be more familiar with psychological and sociological research that has been popularized on this topic (Mandemakers and Kalmijn Reference Mandemakers and Kalmijn2014). Therefore, despite the psychological problems that these mothers may experience, they may be more aware of the importance of providing high quality of parenting to counterbalance these effects.

In addition, several studies show that mothers with higher education are more likely to enroll their children in academically stimulating preschool programs and are more likely to sign their children up for extracurricular activities or summer programs (see Augustine and Crosnoe Reference Augustine and Crosnoe2010 for a review). These pro-academic experiences provide children with learning opportunities that may not be available at home due to the task overload or psychological problems that single mothers often face. For these reasons, highly educated mothers who cannot give these learning opportunities to their children directly may plan so that children receive them indirectly. They may also monitor the results in ways that enhance outcomes.

A third impact on children’s well-being is the quality of the father-child relationship (Amato and Gilbreth Reference Amato and Gilbreth1999). As mentioned, mother’s education is correlated with father’s education; this, in turn, is associated positively with fathers’ involvement (King, Harris, and Heard Reference King, Harris and Heard2004). Cheadle, Amato, and King (Reference Cheadle, Amato and King2010) show that children whose mothers have a high educational level also have a greater probability of maintaining a consistently high level of contact with their fathers over time – a precondition of having a good relationship with them. In addition, mothers who share joint physical custody have a higher educational level than those who are awarded sole custody (Juby, Le Bourdais, and Marcil-Gratton Reference Juby, Le Bourdais and Marcil-Gratton2005), and joint physical custody may be somewhat beneficial for children when compared to sole custody (see Baude, Pearson, and Drapeau Reference Baude, Pearson and Drapeau2016 for a review). In addition to parental relationships, several studies show that the amount of social support children receive outside the home is positively related to their adjustment after divorce (Zartler and Grillenberger Reference Zartler and Grillenberger2017 for review), and affluent children receive more social support (Putnam Reference Putnam2015).

On the other hand, in direct opposition to the compensation hypothesis, the “floor effect hypothesis” posits that the family structure penalty is smaller for children with less-educated mothers (Bernardi and Radl, Reference Bernardi and Radl2014; Leopold and Leopold, Reference Leopold and Leopold2016). This perspective maintains that children with less-educated mothers are less vulnerable to the negative effects of family structure given that their mothers are poor, have low psychological well-being, provide poor quality parenting, and their fathers have little involvement – regardless of whether their parents are together (Bernardi and Boertien Reference Bernardi and Boertien2017b). In other words, women with a low educational level are (Bernardi and Boertien Reference Bernardi and Boertien2017b) already in a bad situation that cannot become much worse. In contrast, children of highly educated mothers are better situated – have a higher level of family income, better maternal psychological well-being, a greater likelihood of good quality parenting, and are more likely to have an involved father – so those who become single mothers have more to lose; family structure matters more for their children.

Previous Research and Limitations

As mentioned, the few studies that have focused on how parental divorce and family structure effects differ by mother’s education have obtained mixed findings (Bernardi and Boertien Reference Bernardi and Boertien2017b). First, substantial research shows that children’s educational attainment suffers less from parental divorce if they have more highly educated mothers (Albertini and Dronkers Reference Albertini and Dronkers2009; Fischer Reference Fischer2007; Grätz Reference Grätz2015). However, the two studies that used test scores rather than educational attainment obtained contradictory results. Augustine (Reference Augustine2014), comparing children in the United States who live in intact married families to those who live in other family forms, found that the effect of family structure on math and reading achievement is greater for those whose mothers have a lower educational level. On the other hand, Mandemakers and Kalmijn (Reference Mandemakers and Kalmijn2014), using the British Cohort Study (1970), found that the effect of parental divorce on reading and math test scores did not vary by maternal education. These findings suggest that the choice of educational outcome affects the conclusions drawn from the research. In addition, outcomes such as mental health and behavior problems that have been extensively analyzed in the literature on family structure effects have not been tested for heterogeneous effects across maternal education levels (with the exception of Mandemakers and Kalmijn Reference Mandemakers and Kalmijn2014). For these reasons, more research on other outcomes alongside educational attainment is needed in order to have a more complete picture of how mother’s education conditions the effect of family structure.

An alternative explanation for the conflicting findings on test scores is that the two studies on this outcome are based on data from different countries. There are several reasons to argue that the role of mothers’ education may vary by country (Bernardi and Boertien Reference Bernardi and Boertien2017b; Mandemakers and Kalmijn Reference Mandemakers and Kalmijn2014). For example, less-educated mothers may be less vulnerable to separation-related declines in income when they live in generous welfare states where various social policies protect citizens against financial hardship (Leopold & Leopold, Reference Leopold and Leopold2016). Mothers with a low educational level would have similar income levels regardless of whether they live in two-parent or single-mother families, since family income cannot be lower than a state-guaranteed minimum. By contrast, children with highly educated mothers should suffer more from single motherhood because they are likely to experience lower income than in two-parent families.

A third reason for these conflicting findings is that there is substantial cross-national variation in the percentage of children living in joint physical custody (Bjarnason and Arnarsson Reference Bjarnason and Arnarsson2011), and children with highly educated mothers have a higher likelihood of living in joint physical custody (Juby, Le Bourdais, and Marcil-Gratton Reference Juby, Le Bourdais and Marcil-Gratton2005). The number of children with highly educated mothers in this living arrangement should, therefore, be greater in countries with a high proportion of children in joint physical custody. Taking into account that joint physical custody is beneficial for children (Baude, Pearson, and Drapeau Reference Baude, Pearson and Drapeau2016), mother’s education may be associated with smaller negative effects of parental divorce and single motherhood in countries with a high percentage of joint physical custody. Additionally, societal characteristics related to the outcome studied may affect the interplay between family structure, mother’s education, and children’s well-being (Bernardi and Boertien, Reference Bernardi and Boertien2017b). Mare (1993) argues that in a society with a high level of inequality in educational opportunity, only very talented children from poorly educated families may obtain higher education. Following this argument, Bernardi and Radl (Reference Bernardi and Radl2014) argue that floor effects are exaggerated in the most unequal societies – children from disadvantaged socioeconomic background are so unlikely to succeed in the educational system that parental divorce or the experience of single motherhood does not reduce their odds substantially. In spite of these theoretical reasons for cross-national variation, previous research was based on single-country studies (with the exception of Bernardi and Radl Reference Bernardi and Radl2014), and a cross-national approach is required to determine whether mothers’ education conditions the effects of family structure differently across countries or whether there is a similar pattern in Western countries.

Data and Variables

For the purposes of this study, we have used data from the 2012 Programme for International Student Assessment (PISA) organized by the Organisation for Economic Co-operation and Development (OECD). PISA data provide internationally comparable measurements on the socioeconomic background and cognitive and noncognitive educational performance of 15-year-old students from OECD countries. In this study, we focus on twenty-one countries that share similar Western cultural traditions and social institutions (Garib, Garcia, and Dronkers Reference Garib, Garcia and Dronkers2007). These countries follow the well-known welfare state regime categories (e.g., Armingeon, Reference Armingeon2001; Esping-Andersen, Reference Esping-Andersen1990; Ferrera, Reference Ferrera1996). The Liberal countries are Australia, Canada, United Kingdom, Republic of Ireland, New Zealand, and the United States. The Nordic countries are Denmark, Finland, Norway, and Sweden. The Continental countries are Belgium, France, Netherlands, Austria, Germany, Switzerland, and Luxembourg. The southern Europe countries are Spain, Greece, Italy, and Portugal.

PISA data have some strengths and some weaknesses. The main strength of PISA is its cross-national comparability. The most significant weakness is the limited nature of the data collected. It is a snapshot of 15-year-old students: No information about either the children’s further development or about their earlier experiences and outcomes is available (Garib, Garcia, and Dronkers Reference Garib, Garcia and Dronkers2007). For example, the causes of the current family structure are not known. Single-parent families may be due to divorce, cohabitants’ separation, parental death, or the parents never having lived together. Furthermore, the most recent PISA survey, the PISA 2015, contains no information about family structures. For this reason, PISA 2012 is used in this chapter.

There are several outcome variables that measure cognitive and noncognitive performance. Cognitive performance is measured using math tests developed by PISA since mathematical literacy was the focus of the PISA 2012 survey. The grade repetition variable takes into account students who repeated a grade in primary school or in secondary school (value “1”) and students who never repeated a grade (value “0”). Truancy is used as a measure of noncognitive performance. Students were asked if, in the last two weeks, they had played truant for a whole day or just from some classes. Students who reported that they had played truant from classes or for days of school at least once in the two weeks leading up to the PISA test have lower scores than students who did not (OECD 2013b). The truancy variable takes value “1” when they played truant all day or from some classes one or more times during the last two weeks and “0” when the student did not.

The family structure variable is based on the child’s response to the questionnaire item asking them with whom they live. This is made up of two categories: single-mother family referring to children who said that they live with only with their mother, and two-parent family referring to children who said that they live with their two biological parents or stepparents.Footnote ¹

Mother’s education is measured using the International Standard Classification of Education (ISCED) scale. Four categories are created: Lower secondary education or below (None education, ISCED levels 1 and 2), upper secondary education and non-tertiary postsecondary (ISCED levels 3 and 4), tertiary education (ISCED levels 5 and 6). The control variables are gender of the child (“1” female and “0” male), immigrant statuses of the mother and the child, and age of the child (measured continuously). The immigrant status of the child, used when predicting child outcomes, has three categories: (1) native student; (2) first-generation student; and (3) second-generation student. Mother’s immigrant status, used when estimating how much education affects the probability of single motherhood, takes value “1” if she is foreign-born and “0” if she is native-born.

Results

The Relationship between Single Motherhood and Education

Table 6.1 shows the percentages of different family types across the twenty-one countries in 2012. As previous studies have found, there is a substantial variation in the percentage of single-mother families. In 2012, the United States had the highest percentage and Greece the lowest.

We tested whether less-educated mothers were more likely to be single using logistic regression with family structure as the dependent variable. Separate models were done for each country, and the effect of mother education was estimated controlling for whether the mother was foreign-born. Table 6.2 only presents the coefficients for mother’s educational level.

In Anglo-Saxon and Nordic countries, single mothers are significantly more likely to have less education, while mothers in two-parent families are more likely to have higher educational levels. There are only two exceptions to this generalization. In the United Kingdom, the effect of tertiary education is not significant. In Denmark, lower education does predict single motherhood, but not significantly.

The Continental countries have two different patterns. In France, the Netherlands, and Belgium, less-educated mothers are more likely to be single, but the relationship is not statistically significant in Belgium. In contrast, in countries where German is spoken – Germany, Switzerland, and Austria – more educated mothers are more likely to be single, though the relationship is only significant in Switzerland. In Luxembourg, single motherhood is distributed almost evenly across the educational spectrum. There is no clear pattern in Mediterranean countries where more educated mothers are more likely to be single in Portugal, Italy, and Greece (not significantly in Greece), but in Spain, like Denmark, less-educated mothers are insignificantly more likely to be single mothers.

Overall, these findings indicate that in spite of the fact that there are still substantial cross-national differences in the relationship between mother’s education and single motherhood, there is a general pattern toward a negative relationship between mother’s education and single motherhood in most Western countries. Higher education significantly predicted greater odds of single motherhood only in Switzerland, Portugal, and Italy among the twenty-one countries analyzed, although in a few other countries, there was no significant effect. Because single motherhood is commonly concentrated at the bottom end of the educational spectrum, it makes sense to continue considering whether trends away from traditional family structure are contributing to an increase of inequality in children’s outcomes and life chances between those from different socioeconomic backgrounds. To answer this question, as mentioned, we need to know if single motherhood matters across several educational outcomes, and if its effects vary by maternal education.

The Relationship between Single Motherhood, Maternal Education, and Various Childhood Outcomes

Table 6.3 shows the main effects of growing up in a single-mother family and mother’s education on math test scores, grade repetition, and truancy controlling for sex, age, and immigration status of the child. We have performed three separate models for each of the twenty-one countries; OLS regressions for math test scores, and logistic regressions for the other two outcomes. The effect of being in a single-mother family is significant for math test scores in all countries except Germany, Spain, Greece, and Portugal. As previous research has shown, there is substantial variation in the magnitude of this effect across countries; the largest negative effects are observed in United Kingdom, the United States, Republic of Ireland, Belgium, and the Netherlands. In all countries, the effect of having a mother with tertiary education is significant and, with the exception of the Netherlands, tertiary education has a substantially greater positive effect on math ability than the negative effect of being in a single-mother family. That is, the magnitude of the effect of tertiary education is greater than the magnitude of the effect of single-mother family. For test scores, it seems that mother’s education is more important than family structure.

Turning to grade repetition, the effect of being in a single-mother family is significant across nineteen of the twenty of the nations studied (there is no grade repetition data for Norway). The estimated effect in Sweden is in line with the other countries, but not significant (b = 0.47, p = 0.125). Finland, Belgium, Austria, Greece, and Italy all show very large effects associated with single motherhood. Unlike math test scores for which the positive effect of mother’s tertiary education outweighed the negative effects of single motherhood in virtually all of the countries, this is only true in slightly more than half (12) of the countries when considering grade repetition. In fact, having a mother with tertiary education did not significantly affect grade repetition in Australia, New Zealand, United Kingdom, and Republic of Ireland, and the estimated family structure effect is larger than the estimated effect of tertiary education in Sweden, Austria, Switzerland, and Italy. In all eight of these countries, family structure seems more relevant in predicting grade repetition than mother’s tertiary education.

The effect of living in a single-mother family on truancy is significant in all countries with the exception of Greece (p = 0.140). The largest effects of family structure on truancy are found in Sweden, Belgium, Germany, and Norway. Unlike math test and grade repetition, the estimated effect of having a mother with tertiary education is not significant in eight of the twenty-one countries, and it is greater than the estimated effect of family structure in ten other countries. Only in the United Kingdom, New Zealand, and United States does the estimated positive effect of mother’s tertiary education exceed the estimated negative effect of being in a single-mother family.

Overall, our analysis reveals substantial differences in the importance of family structure depending on the outcome studied. Growing up in a single-mother family has negative effects on math test scores in only seventeen of the twenty-one countries analyzed, while children of single mothers are more likely to repeat a grade or play truant practically everywhere. In addition, the magnitude of the coefficient of having a mother with tertiary education is clearly more important than family structure on cognitive performance in all countries analyzed, while this is only true in about half for grade repetition and around a fifth for truancy.

To What Extent Does the Effect of Family Structure on Educational Outcomes Diverge by Mother’s Education?

We now turn to investigate whether the impact of being in a single-mother family depends on the mother’s educational level. We show the main and interaction effects between family structure and mothers’ education for each outcome (math test scores in Table 6.4; grade repetition in Table 6.5; truancy in Table 6.6) in every country.

Our results show important cross-country differences. The interaction between family structure and having a mother with a tertiary education is negative and significant in six countries, specifically in Republic of Ireland, United States, Belgium, France, the Netherlands, and Italy. This means that the negative effect of growing up in a single-mother family in these countries is larger when the mother is highly educated. In the United Kingdom, this interaction is also negative but nonsignificant (b = −17.55, p >0.10). We cannot rule out the possibility that the interaction would be significant if United Kingdom had a larger sample size. In contrast, having a mother with tertiary education positively interacts with family structure in Germany and New Zealand, showing that the negative effect of growing up in a single-mother family in these countries is smaller when the mother is highly educated. In the other twelve countries, the penalty associated with being in a single-mother family does not significantly vary by mother’s education. Unlike tertiary education, upper secondary education conditions the effect of family structure in only three countries. In the United States and the Netherlands, the negative effect of growing up in a single-mother family is greater for children who have a mother with upper secondary education than one with lower secondary education. The opposite is true in Australia.

Turning to grade repetition, being in a single-mother home increases the odds of grade repetition more for children of highly educated mothers in the Republic of Ireland, the United States, Belgium, Spain, and Greece (shown by the positive coefficient on the interaction term in Table 6.5). In other words, the cost associated with single motherhood is greater among children of the more highly educated. The opposite is true in Finland, Austria, and Luxembourg where children of less-educated mothers have a greater cost associated with being in a single-mother family. Germany shows the same pattern (b = −0.49, p = 0.16). In the remaining eleven countries, the effect of being in a single-mother family does not differ significantly between children with tertiary and lower educated mothers. In addition to that, in every country the effect of being in a single-mother family did not differ significantly between mothers with an upper secondary education and mothers with less education.

In most countries, the probability of truancy does not differ by mother’s education. There are only few exceptions. In Italy, the probability of truancy among children of single mothers is higher if the mother has upper secondary or tertiary education, and the same is true among those having a single mother with upper secondary education in Canada.

Discussion

The goal of this study was to determine to what extent the increase of single-mother families, especially among the less educated, is associated with an increase in children’s inequality in twenty-one Western countries. To do so, we first analyzed to what extent there is a negative relationship between single-mother families and mother’s education in these countries. This is important because most previous evidence on “diverging destinies” has come from the United States. We also investigated the effect of being in a single-mother family, and how this effect differs by mother’s education. To do so we tested the two main hypotheses developed by the literature: The “compensatory hypothesis,” which posits that mothers with a high educational level are better equipped to protect their children from the negative consequences of growing up in a single-parent family; and the “floor effect hypothesis,” which maintains that children with less-educated mothers are less vulnerable to single motherhood given that their mothers are already in a bad situation than cannot become much worse. We used multiple children’s outcomes and countries in order to overcome the limitations of previous research on how mother’s education conditions the effects of being in a single-mother family.

Our findings highlight substantial cross-national differences in the relationship between mother’s education and single motherhood. However, less-educated mothers are generally more likely to be single mothers in most Western countries. In eleven of the twenty-one countries, there was a significant negative relationship between mother’s education and the probability of being a single mother, and in four more this relationship is also negative but insignificant. More educated mothers are significantly more likely to be single in only three countries – Portugal, Switzerland, and Italy – and previous research has demonstrated that the positive gradient observed in Italy is decreasing (Garriga, Sarasa, and Berta Reference Garriga, Sarasa and Berta2015). Overall these findings indicate that the negative educational gradient toward single motherhood is not only an American phenomenon. However, to what extent does concentration of single motherhood among mothers with less education increase inequality in children’s outcomes between children from different socioeconomic backgrounds?

In all countries analyzed, living with a single mother has a negative effect on at least one of the three outcomes studied. However, we also found substantial differences in the importance of family structure depending on the outcome studied. Being in a single-mother family does not have negative effects on math performance in four of the twenty-one countries analyzed, while its effect on grade repetition and truancy is significant in practically all of them. In addition, mother’s tertiary education is clearly more important than family structure on cognitive performance in all countries analyzed, while this is only true in about half of them for grade repetition, and around a fifth for truancy. Overall, our results highlight that the effect of family structure is more important and consistent across countries for grade repetition and truancy than on cognitive performance. This finding accords with several literature reviews that have concluded there is less consistent evidence on the effects of family structure on test scores than on educational attainment and behavioral outcomes (Amato and Keith Reference Amato and Keith1991; McLanahan Reference McLanahan, Duncan and Gunn-Brooks1997; McLanahan, Tach, and Schneider Reference McLanahan, Tach and Schneider2013; Sigle-Rushton, Hobcraft, and Kiernan Reference Sigle-Rushton, Hobcraft and Kiernan2005). Most previous comparative work had used standardized test scores despite of the fact that grade repetition and truancy are both important outcomes since, as mentioned, they are strongly associated with labor market and socio-emotional outcomes and risk behaviors such as drug abuse or crime (Garry Reference Garry1996; Jones, Lovrich, and Lovrich Reference Jones, Lovrich and Lovrich2011; Range, Yonke, and Young Reference Range, Yonke and Young2011). In other words, these additional two outcomes tell us more about the likelihood that destinies will diverge than cognitive achievement alone does; they have strong behavioral components.

With respect to how the effect of family structure varies by mother’s education, our results show substantial variation across countries and outcomes. Consistent with the “floor effect hypothesis,” the negative impact of being in a single-mother family is greater among otherwise advantaged children on math performance in six countries, on grade repetition in four countries, and on truancy in one country. However, we also obtained a few results consistent with the “compensatory hypothesis”: The negative effect associated with being in a single-mother family is smaller among advantaged children on math performance and grade repetition in three countries.

When taking into account all three of the outcomes studied in each country, we can derive the extent to which the increase of single-mother families would increase inequality between those children from different socioeconomic backgrounds. According to Bernardi and his colleagues, the rise of nontraditional family forms will only increase inequality if single motherhood has a negative effect regardless of maternal education, or if these effects are greater among children with less-educated mothers (Bernardi and Boertien Reference Bernardi and Boertien2016; Bernardi, Boertien, and Popova Reference Bernardi, Boertien and Popova2014). One of these two possibilities is the case for all three outcomes in eleven of the twenty-one countries analyzed. For this reason, it is possible to argue that in countries such as Nordic countries, Australia, and New Zealand, there is evidence that an increase in single-mother families, especially among the less educated, implies an increase in inequality on children’s outcomes. In addition, it is important also to remark that only in Germany, having a mother with tertiary education compensates for the harmful effects of being in a single-mother family on math performance; the same is true for the other two outcomes, though the interaction between family structure and education does not reach statistical significance. These findings accord with those obtained by Grätz (Reference Grätz2015) for the probability of attending the upper track in secondary school (Gymnasium) and on school grades in German and Mathematics.

In contrast, it has been argued that if the costs associated with single motherhood are greater at higher maternal education levels, the growth of single-mother families may reduce inequality in children’s outcomes and life chances between children from different socioeconomic backgrounds (Leopold and Leopold Reference Leopold and Leopold2016). In no country are the negative effects of being in a single-mother family greater at higher maternal education levels across all of the outcomes studied. Therefore, we do not have any evidence that the growing number of single-mother family structures is consistently reducing inequality in societies. In fact, we obtained mixed findings in ten countries. For some outcomes, the negative effect of family structure is greater with higher maternal education (especially math performance) and for other outcomes, the conditioning effect of mother’s education is insignificant. For example, in contrast to Augustine (Reference Augustine2014) whose results supported the compensatory hypothesis, we found that in United States being in a single-mother family was associated with lower math test scores and more grade repetition only among children whose mothers had more education, and the odds of truancy among children in single-mother families did not depend at all on maternal education. Overall, our findings reveal that in around the half of the countries studied, the growth of single-mother families increases inequality in some outcomes and reduces inequality in others.

Alongside these contributions, the study has limitations. Foremost, due to the cross-sectional nature of the PISA data, we were not able to control for selection into single-mother families on unobserved variables and therefore, the interaction effects reported in this study may be spurious due to differences between social origin groups on the probability of being in a single-mother family (Grätz Reference Grätz2015). The data also did not allow testing how mother’s education mattered in different types of single-mother families – single mother at birth, single mother due to parental divorce, and single mother due to parental death – and between different types of two-parent families: Biological and stepfamilies.

The findings of this study demonstrate the importance of more cross-national research on how family structure effects differ by socioeconomic status. They also indicate the need for work across a broader range of outcomes than those analyzed here such as psychological well-being. Such research is essential in order to determine to what extent there is an increase of inequalities in children’s outcomes due to the growing number of single-mother families. Future research should also analyze contextual mechanisms that may explain why maternal education seems to condition family structure effects differently across outcomes and countries, such as the cross-national variations on the percentage of children in joint physical custody.

Family Structure and Child Outcomes

Many children growing up in households with nontraditional family structures, such as single-mother or stepfamilies, do at least as well as their peers (Amato Reference Amato2010). On average, however, they are disadvantaged on a wide range of outcomes compared to children growing up in traditional two-parent families. For instance, several studies have documented that they have lower levels of cognitive ability, noncognitive skills, educational attainment, income, and psychological well-being (Amato Reference Amato2000, Reference Amato2010; Härkönen, Boertien, and Bernardi Reference Härkönen, Bernardi and Boertien2017). These associations are in general relatively modest in size (Amato Reference Amato2000) in comparison to other socioeconomic background characteristics such as parental education (Bernardi and Boertien Reference Bernardi and Boertien2016), have been relatively stable across time (Gähler and Palmtag Reference Gähler and Palmtag2015; Li and Wu Reference Li and Wu2008; Sigle-Rushton, Hobcraft, and Kiernan Reference Sigle-Rushton, Hobcraft and Kiernan2005), but vary to some extent across countries (Hampden-Thompson Reference Hampden-Thompson2013; Pong, Dronkers, and Hampden-Thompson Reference Pong, Dronkers and Hampden-Thompson2003).

What is it about family structures and transitions between them that could have an influence on children’s outcomes? Some authors have argued that it is the stability of a family structure rather than the particular characteristics of a family structure that matters for children’s development. The transition from one family structure type to another (e.g., the exit or the entrance of a parent or stepparent) creates a new situation to which children have to adapt, this might interfere with the development of cognitive and noncognitive characteristics (Fomby and Cherlin Reference Fomby and Cherlin2007; Waldfogel, Craigie, and Brooks-Gunn Reference Waldfogel, Craigie and Brooks-Gunn2010). To test this hypothesis empirically, several studies compared children living in stable nontraditional families to stable two-parent families and other family forms. In general, little empirical support has accumulated for the “family stability” perspective. Single-parent families often do worse compared to two-parent families also if they are stable throughout childhood (Magnuson and Berger 2010; Mariani, Özcan, and Goisis Reference Mariani, Özcan and Goisis2017), and the separation of a two-parent family appears to be more impacting for children’s outcomes than other family transitions (Bzostek and Berger Reference Bzostek and Berger2017; Lee and McLanahan Reference Lee and McLanahan2015).

The characteristics particular to certain family structures and transitions therefore appear to be responsible for its associations with child outcomes. The specific family structures and transitions that have received most attention are single-parent families (McLanahan, Tach, and Schneider Reference McLanahan, Tach and Schneider2013), the separation of two-parent families (Härkönen, Bernardi, and Boertien Reference Härkönen, Bernardi and Boertien2017), and the formation of a family including a stepparent (Sweeney Reference Sweeney2010). Characteristics held responsible for the effects of living with a single parent include less authoritarian parenting styles, obstacles to employment for the co-resident single parent, and access to resources of the non-resident parent (Amato Reference Amato2010; McLanahan and Sandefur Reference McLanahan and Sandefur1994; Seltzer Reference Seltzer, ldham and Melli2000). Parental separation, besides implying a transition to a single-parent family, can also come with family conflict and financial costs (Cherlin Reference Cherlin1999; Kalmijn, Loeve, and Manting Reference Kalmijn, Loeve and Manting2007; Pryor and Rodgers Reference Pryor and Rodgers2001; Uunk Reference Uunk2004). Many studies find the income losses related to parental separation to be responsible for a large part of its effects on educational outcomes (Jonsson and Gähler Reference Jonsson and Gähler1997; McLanahan and Sandefur Reference McLanahan and Sandefur1994; Thomson, Hanson, and McLanahan Reference Thomson, Hanson and McLanahan1994). Parental separation can also have a negative impact on psychological well-being, both in the short and the long term (Amato Reference Amato2010; Härkönen, Bernardi, and Boertien Reference Härkönen, Bernardi and Boertien2017), which can translate into poorer educational performance. Stepparents can provide time and financial resources that can compensate for some of the disadvantages experienced by single parents. Children living with a stepparent, however, appear to be more similar in their outcomes to their peers living with a single parent compared to peers living with two biological parents (Gennetian Reference Gennetian2005; Jonsson and Gähler Reference Jonsson and Gähler1997; Thomson, Hanson, and McLanahan Reference Thomson, Hanson and McLanahan1994).

The documented association between family structure and child outcomes could also be due to endogeneity, and hence be spurious. Variation in child outcomes across groups might reflect other processes that are both related to family structures and transitions as well as child outcomes. A major suspect in this respect is socioeconomic disadvantage of parents that might influence both child outcomes and the likelihood to enter a given family structure. In many countries, socioeconomically disadvantaged mothers are more likely to have children outside a union (Perelli-Harris et al. Reference Perelli-Harris, Kreyenfeld and Kubisch2010) and to separate after forming a union (Härkönen and Dronkers Reference Härkönen and Dronkers2006; Matysiak, Styrc, and Vignoli Reference Matysiak, Styrc and Vignoli2014). Associations between family structures and child outcomes might therefore reflect socioeconomic disadvantages that were already present before family formation or before a family transition took place.

In the study of the effects of parental separation, family conflict has been marked as an additional possible source of endogeneity. Many families who break up are likely to experience high levels of conflict before separation. In that case, parental conflict might both lead to a separation and have consequences for children’s outcomes. The actual separation of the parents could in that case have little extra consequences for children’s outcomes (Demo and Fine Reference Demo and Fine2010; Dronkers Reference Dronkers1999; Härkönen, Bernardi, and Boertien Reference Härkönen, Bernardi and Boertien2017).Footnote ¹

Several methods have been employed to monitor or control away the possible influence of these sources of endogeneity (see McLanahan, Tach, and Schneider Reference McLanahan, Tach and Schneider2013, and Härkönen, Bernardi, and Boertien Reference Härkönen, Bernardi and Boertien2017 for overviews). Whereas in some studies associations of family structure with child outcomes disappear, they persist, at least to some extent, in most studies (McLanahan, Tach, and Schneider Reference McLanahan, Tach and Schneider2013). Associations were more often found to be spurious once looking at cognitive ability, whereas they often appeared of a more causal nature once studying educational attainment (Bernardi and Boertien Reference Bernardi and Boertien2016; McLanahan, Tach, and Schneider Reference McLanahan, Tach and Schneider2013). The actual role of family structure in affecting inequality of opportunity is therefore likely to depend on the outcome variable considered. Nonetheless, given that educational attainment is a key socioeconomic outcome, family structure appears to matter for children’s chances in life at least to some extent.

Prevalence of Family Structure Types and Its Social Stratification

Family structure does thus appear to matter at least to some extent for children’s outcomes. However, whether and to what extent variation in family structure also contributes substantially to the observed inequality of opportunity between socioeconomic groups depends crucially on whether nontraditional family forms are common, and whether variation in family structures is socioeconomically stratified.

Giving birth as a single mother has been traditionally more common among women with lower socioeconomic status, but it is still an uncommon course of events in most countries, with the Czech Republic, Russia, the United Kingdom and the United States as some exceptions (Andersson, Thomson, and Duntava 2016; Mariani, Özcan, and Goisis Reference Mariani, Özcan and Goisis2017). Most episodes of living in a nontraditional family therefore start after the break-up of a two-parent family. The extent to which parental separation is socially stratified (i.e., correlated with socioeconomic characteristics) differs across countries, especially once looking at the socioeconomic status of the mother (Härkönen and Dronkers Reference Härkönen and Dronkers2006; Matysiak, Styrc, and Vignoli Reference Matysiak, Styrc and Vignoli2014). Explanations for variation in the socioeconomic gradient of divorce often go back to Goode (Reference Goode, Bendix and Lipset1962), who argued that when divorce is relatively uncommon, individuals need resources to overcome social, economic, and legal barriers to divorce. In such situations, the socioeconomic gradient of divorce will be more positive, but this gradient is expected to reverse to negative once barriers to divorce fade out and also those with fewer resources can divorce. The supposed greater stress experienced by disadvantaged couples will eventually cause them to divorce more once barriers to divorce cease to play a key role (Boertien Reference Boertien2012; Conger, Conger, and Martin Reference Conger, Conger and Martin2010; Härkönen and Dronkers Reference Härkönen and Dronkers2006). In addition to Goode’s long-standing narrative about why the socioeconomic gradient of divorce would become negative, contemporary reasons have been proposed as to why the socioeconomically advantaged are less likely to divorce. These include a higher prevalence of egalitarianism among the educated, which could stabilize relationships (Esping-Andersen et al. Reference Esping-Andersen, Boertien, Bonke and Gracia2013; Goldscheider, Bernhardt, and Lappegård Reference Goldscheider, Bernhardt and Lappegård2015), and greater internal barriers to divorce caused by common investments and commitment to relationships (Boertien and Härkönen Reference Boertien and Härkönen2018).

Table 7.1 shows the prevalence of different family types and the extent to which they are more common among low-educated mothers across countries. If we take mother’s education as a proxy for a family’s socioeconomic position, these numbers provide an indication of the scope for family structure to affect socioeconomic inequality of opportunity across countries. As discussed above, disadvantageous family types should be fairly common and concentrated among socioeconomically disadvantaged families in order to make a substantial contribution to inequality of opportunity.

The table combines information on the percentage of mothers who are single and its stratification by mother’s education. This classification is based on Härkönen’s (Reference Härkönen2017) results using cross-sectional data from the Luxembourg Income Study (LIS) for the period 2011–2015 (or 2006–2010 if recent data are missing).

The more prevalent single motherhood, and the more negative the association between mother’s education and single motherhood, the more likely family structure is to contribute to socioeconomic inequality of opportunity. Hence, in countries toward the bottom right corner of the table, such as Australia, Denmark, East Germany, and the United Kingdom, family structure is more likely to play a role in amplifying socioeconomic inequalities, whereas this is less likely to be the case in countries toward the upper-left corner, such as Hungary, Italy, and Serbia. Estimates of the accumulated exposure toward single parenthood across childhood based on union histories indicate a similar ranking of countries (for the countries with data available), but with France and the Czech Republic being among the countries with the highest percentage of children ever exposed to single parenthood (Andersson, Thomson, and Duntava 2016).

Differences in the Effects of Family Structure across Socioeconomic Groups

A final key factor that determines to what extent family structure contributes to socioeconomic inequality of opportunity is the heterogeneity in its effects on child outcomes across groups. It could be that family structure is socially stratified and that it matters for child outcomes on average, but that its effects are restricted to socioeconomically advantaged children. In that case, its effects on socioeconomic inequality of opportunity will still be limited. In contrast, if family structure especially matters for the disadvantaged, the contribution of family structure to inequality of opportunity might be bigger than expected.

Should we expect heterogeneity in the effects of family structure according to socioeconomic status of families? An in-depth of discussion of this issue can be found in Bernardi and Boertien (Reference Bernardi and Boertien2017b). Two competing expectations can be formed in that regard. On the one hand, children from socioeconomically advantaged backgrounds might have more resources to deal with the challenges posed by living in a nontraditional family form. On the other hand, children from socioeconomically advantaged backgrounds might have more to lose from an absent parent. It could be harder for nonresident parents to transmit their cultural, social, and economic capital to their children (Coleman, Reference Coleman1988). Following Bernardi and Radl (Reference Bernardi and Radl2014), these competing expectations can be labeled as the “compensatory” and “floor effect” hypotheses respectively (see also Chapter 6). Kearney and Levine (Reference Kearney and Levine2017) described it in more economic terms as variation in the “marriage premium for children” according to socioeconomic background.

Studies on differences in the effects of family structure according to socioeconomic background have accumulated rapidly over the last years (Bernardi and Boertien Reference Bernardi and Boertien2017b) and do not all come to the same conclusions. Studies looking at educational attainment mostly find that children from advantaged backgrounds are affected more by parental separation (Bernardi and Radl Reference Bernardi and Radl2014; Kearney and Levine Reference Kearney and Levine2017; Martin Reference Martin2012; McLanahan and Sandefur Reference McLanahan and Sandefur1994). A recent study on the United Kingdom (Bernardi and Boertien Reference Bernardi and Boertien2016) documented how this pattern can to an important extent be explained by changes in family income following separation. Not only do children with higher educated parents lose more family income following separation, these losses in income are also more consequential for their college attainment. Given that family income matters less for the educational attainment of socioeconomically disadvantaged children (as family income could be too low to invest in education to begin with), losses in family income due to separation are less consequential for them (Bernardi and Boertien Reference Bernardi and Boertien2016).

Results from studies on other outcomes such as cognitive ability and psychological well-being come to more mixed conclusions with both possibilities finding support across studies (Augustine Reference Augustine2014; Grätz Reference Grätz2015; Mandemakers and Kalmijn Reference Mandemakers and Kalmijn2014; Ryan, Claessens, and Markowitz Reference Ryan, Claessens and Markowitz2015). Results depend crucially on whether one looks at heterogeneity according to maternal or paternal resources, as maternal resources are often directly accessible to children living with a single parent, whereas access to the resources of the father could be more complicated. The results of most studies can indeed be aligned with a narrative where effects of family disruption are larger when maternal resources are low and paternal resources are high (Bernardi and Boertien Reference Bernardi and Boertien2017b).

The context studied also appears consequential for conclusions. For instance, Grätz (Reference Grätz2015) provided one of the few results on Germany, and found that only the school performance of socioeconomically disadvantaged children is affected by parental separation. Studies on Italy, the Netherlands, and Sweden find smaller effects for children with resourceful mothers (Albertini and Dronkers Reference Albertini and Dronkers2009; Fischer Reference Fischer2007; Jonsson and Gähler Reference Jonsson and Gähler1997), but larger effects for children with resourceful fathers (Fischer Reference Fischer2007; Jonsson and Gähler Reference Jonsson and Gähler1997). Single-country studies on the United Kingdom and the United States in general support the conclusion that socioeconomically advantaged children are affected more by separation (Biblarz and Raftery Reference Biblarz and Raftery1993; Mandemakers and Kalmijn Reference Mandemakers and Kalmijn2014; Martin Reference Martin2012; McLanahan and Sandefur Reference McLanahan and Sandefur1994).

Bernardi and Radl (Reference Bernardi and Radl2014) documented the extent to which effects of parental separation on educational attainment differ by parental education across countries. They found, overall, that socioeconomically advantaged children were affected more by parental separation than socioeconomically disadvantaged children. Importantly, however, these differences in effects were smaller or even the opposite in countries where ability tracking in schools occurs at early ages. If crucial transitions in children’s school careers take place at an early age, separations taking place after that age will have small effects on children’s educational attainment (Bernardi and Radl Reference Bernardi and Radl2014), reducing the estimated influence of parental separation experienced during childhood when averaged across ages.

In general, our reading of the empirical evidence is that socioeconomic heterogeneity in the effects of family structure tends to limit the influence family structure has on inequality of opportunity. In any case, there is no strong evidence that the consequences of growing up in a nontraditional family are greater for children from socioeconomically disadvantaged backgrounds, and hence, that heterogeneity in the effects of family structure would be another factor contributing to the accumulation of disadvantages. Whether this is indeed the case is an empirical question that has so far been addressed only on few occasions.

Quantifying the Contribution of Family Structure to Inequality of Opportunity

We are now in a position to go back to the key question of this chapter: How large is the contribution of variation in family structures to inequality? This question has been central to many studies on income inequality and poverty. Various decomposition and simulation techniques have been used to estimate how much changes in family structure have contributed to changes in income inequality and poverty over time (M. A. Martin Reference Martin2006; McLanahan and Percheski Reference McLanahan and Percheski2008; Western, Bloome, and Percheski Reference Western, Bloome and Percheski2008). Studies on the United States in general come to the conclusion that family structure has been consequential for inequality. A review of the literature stated that between 11% and 41% of the increase in income inequality over the last decades in the United States can be attributed to increases in female-headed households (McLanahan and Percheski Reference McLanahan and Percheski2008). Evidence for other countries is more mixed with one study arguing that family structure only matters for income inequality in the United States (Esping-Andersen Reference Esping-Andersen2007) and other studies finding an income inequality amplifying effect for family structure across sixteen countries (Kollmeyer Reference Kollmeyer2013). A comparative study on poverty among single mothers comes to a similar conclusion (Härkönen Reference Härkönen2017).

That variation in family structures matters for income inequality and poverty, however, does not automatically imply that it also matters for inequality of opportunity between children coming from different socioeconomic groups. Few studies have, until now, aimed to quantify the extent to which family structure could explain differences in child and adult outcomes between individuals coming from socioeconomically advantaged and disadvantaged backgrounds.

Bernardi and Boertien (Reference Bernardi and Boertien2017a) presented such estimates of the contribution of family structure to socioeconomic background differences in educational attainment for four countries: Germany, Italy, the United Kingdom, and the United States. Their main question was to what extent differences in the likelihood of attaining tertiary education between children with higher and lower educated parents could be explained by family structure. Their results are summarized in Figure 7.1, which displays observed differences in college attainment between individuals with a lower (ISCED 1–2) and higher (ISCED 5–6) educated mother. These observed differences are compared to predicted differences between both groups of individuals in the hypothetical situation that all children would have grown up in a two-parent family.

Figure 7.1 Actual and predicted university attainment in hypothetical situation “where all children grow up with both parents in the household”

Figure 7.1 reveals that, in all four countries, differences in college attainment depending on maternal education are predicted to be very similar to observed differences in the hypothetical situation that all individuals would have grown up in a two-parent household. This suggests that the explanatory power of family structure is limited.

The reasons for this result differed according to country. In Italy, the number of children living in a nontraditional family was too small (see Table 7.1) to have a major impact on inequality of opportunity. In Germany, family structure was not (yet) clearly stratified according to parental education, preventing its influence on inequality of opportunity. In the United Kingdom and the United States children of lower educated parents were more likely to grow up in a nontraditional family structure. This was most clearly so in the United States where differences in college attainment between individuals with lower and higher educated parents were estimated to be 10% lower if family structure would not be stratified by parental education. However, in both countries children of higher educated parents were more negatively affected by growing up in a nontraditional family. This heterogeneity in effects almost entirely canceled out the effects of the stratification of family structure by parental education.

A lack of such “diverging destinies” due to variation in family structures in the United States has also been documented in another study using a similar approach (Alamillo Reference Alamillo2016). However, evidence quantifying the possible role of variation in family structures is limited to studies on educational attainment. It could well be that socioeconomic background differences in other outcomes are amplified by variation in family structures.

Discussion and Conclusion

Does the result that family structure can explain little of socioeconomic background differences in educational attainment imply that family structure does not matter for socioeconomic inequality of opportunity in general? More evidence is needed before such a conclusion can be reached. The existing evidence quantifying the contribution of family structure is limited to studies on educational attainment and current research is limited to a small set of countries and time periods.

Even though tertiary education is an important socioeconomic marker, it could be that family structures and transitions between them are important for socioeconomic background inequalities in other outcomes such as income, status, health, or even secondary education. Whether this is the case depends on how strongly family structure is related to these outcomes, and how this relationship varies between socioeconomic groups. Kearney and Levine (Reference Kearney and Levine2017) argued that the additional resources that a second parent (in their framework the father) brings to the household matter less for socioeconomically disadvantaged children when the outcome is only attained by relatively few people, as is the case with tertiary education. This may explain why higher educational attainment is the outcome for which the clearest evidence exists that socioeconomically advantaged individuals are more negatively affected by growing up in a nontraditional family (Bernardi and Radl Reference Bernardi and Radl2014; Martin Reference Martin2012; McLanahan and Sandefur Reference McLanahan and Sandefur1994). For other outcomes, such as psychological well-being and cognitive ability, evidence is less uniform (Grätz Reference Grätz2015; Mandemakers and Kalmijn Reference Mandemakers and Kalmijn2014).

Kearney and Levine made a contrasting argument for outcomes attained by the majority of the population, such as living out of poverty. In such cases, an inverted U-shaped pattern is predicted to be observed with both the most-disadvantaged and advantaged individuals benefiting the least from an additional parent’s resources. This is because socioeconomically advantaged single parents have sufficient resources to enable their children to attain such outcomes, while for many of the most-disadvantaged children, the additional resources of a second parent would still not bring them to a level that enables them to attain “basic outcomes.” If family structure indeed matters little for the attainment of “basic outcomes” by socioeconomically advantaged individuals, its contribution to inequality of opportunity might be greater once considering adult outcomes such as secondary education, employment, and living without debt. Conversely, its role in creating unequal access to outcomes attained by a smaller proportion of the population such as home ownership and other assets might be more limited because, like tertiary education, these are outcomes that advantaged individuals may be less likely to attain if they lose immediate access to the resources of a second parent in the household.

Can we expect family structure to matter more for socioeconomic background inequality of opportunity in contexts that have not been studied so far? Returning to the above discussion on the conditions under which family structure can matter provides clues to answer this question. First, nontraditional family structures have to be common and be socially stratified in order to impact on inequality of opportunity. Both the prevalence and social stratification in single parenthood have continued to increase in many countries during the latter decades (Härkönen Reference Härkönen2017), whereas most of the above results pertain to individuals born in the 1970s and 1980s (as educational attainment was measured around age 30). Stratification in family structures can therefore have become more important for inequality of opportunity over time. The role of family structure also remains unclear in countries where educational differences in family structures are relatively large but which have not featured in previous studies (such as Australia, Denmark, and Republic of Ireland [see Table 7.1]).

Second, the (negative) effects of family structures on the outcomes studies have to be relatively strong. Family structure effects are found in each country and they have remained relatively stable over time, despite the increase in nontraditional families (Härkönen, Bernardi, and Boertien Reference Härkönen, Bernardi and Boertien2017). An important implication of the insight that the role of family structure in the intergenerational reproduction of inequality is contingent on effect size is that intergenerational inequality can be potentially addressed by targeting the effects of family structure on child outcomes (cf. Cohen Reference Cohen, Amato, Booth, McHale and Van Hook2015). Family structure effects on economic outcomes – such as child poverty (Härkönen Reference Härkönen2017) – are readily modified by public policies, but findings suggesting that the effects on school performance can depend on social policies (Pong, Dronkers, and Hampden-Thompson Reference Pong, Dronkers and Hampden-Thompson2003) or the features of the educational system (Bernardi and Radl Reference Bernardi and Radl2014) support that public policies can address the consequences of family change more broadly. Family change need not inevitably lead to increasing inequality, and whether it does can depend on appropriate policy measures.

Third, the impact of variation in family structures on inequality of opportunity will be particularly large if family structure matters most for socioeconomically disadvantaged families. Many studies have shown, instead, that family structure effects are stronger for children from socioeconomically advantaged backgrounds. This particular heterogeneity in the consequences of growing up in a nonintact family reduces the contribution of family structures to the overall inequality of opportunity. It still remains unclear for many countries and outcomes whether heterogeneity in the effects of family structure exists and, if so, whether socioeconomically disadvantaged children are affected more.

All in all, however, the example from the United States is instructive. This is a context where effects of growing up in a nonintact family are large (Hampden-Thomson Reference Hampden-Thompson2013) and strongly socially stratified (Härkönen Reference Härkönen2017), but nonetheless the consequences for equality of opportunity are small because of larger consequences associated with nonintact families experienced by advantaged children. Moreover, inequality of opportunity would be only 10% lower if family structure effects were homogenous across socioeconomic groups (Bernardi and Boertien Reference Bernardi and Boertien2017a). Therefore, it is unlikely that the contribution of family structure to inequality of opportunity in education will be very large in other contexts. The overall conclusion of this chapter thus remains: Currently it does not appear to be the case that family structure contributes to inequality of opportunity between children of different socioeconomic groups in a major way. This conclusion does not mean that family structure does not matter per se for children’s outcomes. Children growing up in nontraditional families do, on average, differ in their outcomes from their peers growing up in stable two-parent families and are overrepresented among children living in poverty (Härkönen Reference Härkönen2017). Family structure is therefore a factor to take into account once studying income inequality, poverty or the characteristics of the most-disadvantaged children. Overall, however, the argument that variation in family structure is a major engine behind socioeconomic inequality of opportunity is not yet empirically supported.

Marriage, Family, and Economic Growth

The Noble Laureate, Robert Lucas Jr., has noted that once one starts to think about economic growth, it becomes hard to think about anything else (Lucas Reference Lucas1988). His comment reflects the fact that even small differences in economic growth rates can accumulate into very large differences in standards of living over time. Nations that create even small improvements in economic growth rates will see dramatic improvements in economic prosperity over time.

Previous research suggests that higher levels of household income and savings, male labor force participation, low levels of violent crime, and educational attainment all potentially play a role in fostering the conditions for economic growth. Here, we consider the ways in which marriage and the proportion of two-parent families in a nation may influence these factors. In this analysis, two-parent families include both married and cohabiting couples with children. Our assumption is that any effects of family structure on cross-national economic growth may be mediated in part by these mechanisms. That is, higher rates of marriage and two-parent families may foster more household income and savings, male labor force participation, public safety, and educational attainment among children, and in ways that – in turn – promote higher rates of economic growth. At the same time, more marriage and a higher proportion of two-parent families might also inhibit growth, insofar as they reduce female labor force participation. In this section, we explore associations between the proportion of children living in two-parent families and these potential mechanisms through which we expect stable families to encourage growth. In the section that follows, we test for net effects of family structure on economic growth, controlling for country-specific fixed effects and time-varying country-level characteristics that are also likely to contribute to economic growth.

Household Income and Savings

Compared to single individuals, married households enjoy greater economies of scale, often access to more income, and higher savings rates. One might note that marriage creates a mechanical bias in household income since there are two potential earners in the family compared with only one potential earnings’ stream in households with a single adult. That is exactly the point. Marriage can create two sources of income and allows the household to take advantage of economies of scale (Lerman and Wilcox Reference Lerman and Wilcox2014). Even in married households where only one spouse is in the labor force, the presence of both individuals allow them to specialize to maximize the welfare of the household and provide a natural source of insurance if the primary earner loses his or her job (since the spouse that was previously not working can enter the labor force to provide additional income) (Becker Reference Becker1993). All these things are conducive to growth because they foster positive economic outcomes at the household level (which can aggregate up to the national level) (Samuelson and Modigliani Reference Samuelson and Modigliani1966). They may also promote economic growth by encouraging a stronger work ethic and by reducing the need for government-funded social welfare programs (Plümper and Martin Reference Plümper and Martin2003).

Even compared to comparable cohabiting couples, married couples have an economic advantage. Because they enjoy more commitment and, in many cases, more legal recognition and benefits, married couples enjoy more stability. New research from the Social Trends Institute, for instance, indicates both that cohabiting families are less stable than married families in much of Europe and North America, even when they have children together, and that the growth of cohabitation across much of the globe is linked to increased family instability in countries around the world (DeRose et al. Reference DeRose, Lyons-Amos, Wilcox and Huarcaya2017). If research in the United States is representative of trends across the globe, marriage’s comparative advantage in commitment and stability should translate into higher levels of savings; married couples are also less likely to incur expenses from a union dissolution, insofar as union instability is markedly higher in cohabiting families than in married families (DeRose et al. Reference DeRose, Lyons-Amos, Wilcox and Huarcaya2017). Indeed, stably married Americans typically accumulate more assets than men and women who are single, divorced, remarried, or cohabiting, even controlling for a range of background factors (Lupton and Smith Reference Lupton, Smith and Grossbad2003; Wilmoth and Koso Reference Wilmoth and Koso2002). The income, savings, and stability advantages associated with marriage should help to explain any advantages that countries with more married adults and families enjoy, compared to countries with more single or cohabiting adults.

Finally, the links between marriage and economic growth should also be paralleled when we turn our attention to the proportion of two-parent families in a society. Specifically, a greater proportion of two-parent families – especially when such families are stable – should promote economies of scale, income pooling, and higher levels of savings in societies across the globe. That is, we expect that adults in two-parent families are more likely to pool income and devote more money to savings than adults living by themselves or heading up single-parent families.

So, what does the cross-national evidence suggest about the links between marriage, two-parent families, and household economic patterns? We do not have access to data about household income in countries across the globe, but data taken from the World Bank, the Organisation for Economic Co-operation and Development (OECD), and the Demographic and Health Surveys (DHS) indicate an association between family structure and one important economic outcome, the savings rate, conducive to growth. (For a list of the countries and years included in our data analyses for Figures 8.1, 8.3, 8.4, 8.5, as well as for Tables 8.1 and Table 8.2, please see our online Appendix, Table 8A.1: http://sociology.virginia.edu/media/2696.) Specifically, Figure 8.1 indicates that the proportion of children living in two-parent families is associated with higher savings rates in nations across the globe. That is, in more than ninety countries, countries that have more children living in two-parent families also have higher rates of savings. This pattern is consistent with our hypothesis that two-parent families foster economic behaviors conducive to economic growth. Marriage is associated with two-parent families, but future research will have to determine if marriage itself has a similar relationship with household saving. However, in general, we suspect that strong and stable families foster patterns of household income and savings, both of which promote economic growth at the national level.

Figure 8.1 Gross savings as percentage of GDP, by proportion of children being raised by two parents: 2001–2015

Note: The y-axis in this figure provides the gross saving rates as a percent of GDP(World Bank national accounts data for 2001–2015). The x-axis splits the sample of countries into terciles based on the fraction of children being raised by two parentsbased on data from the World Bank, OECD, and the DHS. The sample includes 429 country-year observations from 90 countries representing the following regions: Africa (30), the Americas (11), Asia (16), Europe (30), and Oceania (3).

Table 8.1 GDP growth by proportion of adults who are married, country-level regression

	(1)	(2)	(3)	(4)
Proportion of Adults Married	0.0819***	0.103***	0.102***	0.106***
	(0.0297)	(0.0288)	(0.0287)	(0.0286)
Population (log)		−1.626***	−1.702***	−1.764***
		(0.377)	(0.379)	(0.377)
Percentage of Population in Cities		0.0830	0.106	0.0640
		(0.138)	(0.138)	(0.138)
Proportion of Population under 15		−0.507***	−0.526***	−0.528***
		(0.119)	(0.119)	(0.118)
Proportion of Population over 65		−0.306*	−0.304*	−0.295*
		(0.168)	(0.167)	(0.166)
Average Years of Education			−0.192	−0.206*
			(0.117)	(0.116)
Life Expectancy				0.200**
				(0.0897)
N	401	401	401	401

Note: The outcome variable is the log of per capita GDP. Each regression includes country fixed effects. All of the control variables are measured in standard deviation units. ***, **, and * indicate statistical significance at the 1, 5, and 10 levels, respectively. This includes 129 different countries. The regions represented by these countries are Africa (37), the Americas (24), Asia (32), Europe (32), and Oceania (4).

Table 8.2 GDP growth by proportion of children in two-parent homes, country-level regression

	(1)	(2)	(3)	(4)
Proportion of Children with Two Parents	0.126**	0.185***	0.192***	0.157***
	(0.0516)	(0.0505)	(0.0513)	(0.0517)
Population (log)		−0.189	−0.239	−0.242
		(0.470)	(0.475)	(0.468)
Percentage of Population in Cities		−0.0774	−0.0940	−0.134
		(0.186)	(0.187)	(0.185)
Proportion of Population under 15		−1.054***	−1.038***	−1.055***
		(0.166)	(0.168)	(0.165)
Proportion of Population over 65		−0.510***	−0.522***	−0.435***
		(0.112)	(0.113)	(0.115)
Average Years of Education			0.106	0.0232
			(0.143)	(0.143)
Life Expectancy				0.399***
				(0.121)
N	416	416	416	416

Note: The outcome variable is the log of per capita GDP. Each regression includes country fixed effects. All of the control variables are measured in standard deviation units. *** and ** indicate statistical significance at the 1 and 5 percent levels, respectively. This includes 87 different countries. The regions represented by these countries are Africa (30), the Americas (10), Asia (16), Europe (29), and Oceania (2).

Male and Female Labor Force Participation

The research on men and work in the United States indicates that marriage tends to have a transformative impact on men. After marrying, men tend to work harder, smarter, and more successfully; they also are more likely to steer clear of risky activities (Nock Reference Nock1998; Lerman and Wilcox Reference Lerman and Wilcox2014). In the words of Nobel laureate George Akerlof (Reference Akerlof1998, p. 290), “men settle down when they get married,” adding, “Married men are much more attached to the labor force; they have less substance abuse; they commit less crime, are less likely to become the victims of crime, have better health, and are less accident prone.”

The United States experience is instructive here. Over the last forty years, the drop in male labor force participation has been largest among men who are not married (see also Chapter 5). In Figure 8.2, we plot the fraction of men by family status from 1979 to 2013. This figure indicates that married fathers had the smallest declines in labor force participation, whereas unmarried men who do not live with children have seen the biggest declines in labor force participation. That is, the fraction of married men with children who are in the labor force has stayed relatively stable during this time period (at about 90 percent), whereas the fraction of unmarried men without children in the home in the labor force has fallen to about 75 percent. Lerman and Wilcox (Reference Lerman and Wilcox2014) estimate that about one third of the decline in men’s labor force participation since 1979 in the United States is associated with the retreat from marriage in the United States.

Figure 8.2 Percentage of 25–50-year-old men employed, by marital status and fatherhood: 1979–2013

Note: This figure is from Lerman and Wilcox (2014).

To be sure, there is a debate about whether marriage causes higher labor force participation or vice versa. Several recent studies have noted that a contributing factor to lower marriage rates are the lower employment prospects for less-educated men (Carbone and Cahn Reference Carbone and Cahn2014; Wilson Reference Wilson1987); however, research by economists Ahituv and Lerman (Reference Ahituv and Lerman2007) indicates that stable marriage increases men’s attachment to the labor force and their income in the United States, even after taking into account background factors like education and job experience. Chapter 5 also shows that less-educated men who are married are much more likely to be working than are less-educated men who are not working.

Men’s stronger connection to the labor force is also linked to a marriage premium in personal income. Because men work more hours and work more strategically when they are married, they tend to enjoy higher incomes than their equivalently credentialed single peers (Lerman and Wilcox Reference Lerman and Wilcox2014). Research in the United States suggests this marriage premium is greater than 10 percent (Lerman Reference Lerman, Cohen and Wright2011). The evidence, then, suggests that marriage is associated with both more work and more income for men in the United States. Cross-national research in fifteen countries in Europe and North America also indicates that married men enjoy an income premium in most of these countries (Geist Reference Geist2006). However, much of the premium can be attributed to underlying human capital differences between married and unmarried men or to increased engagement or better opportunities at work associated with the transition to adulthood for men – rather than marriage per se (Geist Reference Geist2006; Killewald and Lundberg Reference Killewald and Lundberg2017). In other words, from this research, it is not clear if marriage exercises a causal role on men’s wages or, if instead, men with more human capital or better work opportunities are more likely to get and stay married.

Finally, the link between family structure and men’s work seems to be particularly strong for men who are married fathers. Evidence from the United States indicates that men tend to work the most hours and garner the highest income premiums when they are married fathers than when they are in other family statuses (Nock Reference Nock1998; Killewald Reference Killewald2012). Men may feel particularly strong internal and external pressures to provide when they are married with children to support financially. In other words, married fatherhood is associated with the greatest premium in work hours and income for men, at least in the United States.

If our expectations about marriage and fatherhood are correct, we would expect to find higher levels of labor force participation and income for men in countries with more married men or two-parent families. Here, we look at measures of men’s labor force participation by the proportion of children who are in two-parent families. Specifically, we use data from the World Bank and the Demographic and Health Survey to examine the relationship between the proportion of children within two-parent families and male participation in the labor force, hypothesizing that men who are helping raise their children have additional motivation to support their families and seek employment. Contrary to our expectations, Figure 8.3 indicates that men’s labor force participation for ages 15 and older is negatively associated with children living in two-parent families. Thus, at least in this international sample of more than ninety countries, we do not see evidence that more two-parent families foster higher paternal labor force participation. This runs counter to our predictions.

Figure 8.3 Male labor force participation, by proportion of children in two-parent families: 2001–2015

Note: The y-axis in this figure provides the average male labor force participation is based on the International Labor Organization database for 2001-2015 available from the World Bank. The x-axis splits the sample of countries into terciles based on the fraction of children being raised by two parents based on data from the World Bank, OCED, and the DHS. The sample includes 443 country-year observations from 97 countries representing the following regions: Africa (35), the Americas (11), Asia (18), Europe (30), and Oceania (3).

There is another way in which strong families, however, might weaken the economy. Strong families may reduce female labor force participation and income, which could serve as a drag on the economy. The evidence in the United States and Europe is ambiguous about the effect that marriage has upon women’s labor force participation and income (Geist Reference Geist2006; Killewald Reference Killewald2012; Lerman and Wilcox Reference Lerman and Wilcox2014). Motherhood, however, is associated with fewer hours, less work, and lower income for women in many countries in the developed world (Budig and England Reference Budig and England2001; Budig, Misra, and Boeckmann Reference Budig, Misra and Boeckmann2012; Gash Reference Gash2009; Harkness and Waldfogel Reference Harkness, Waldfogel and Polacheck2003). The maternal penalty is reduced in countries where norms and public policies support mother’s work (Budig, Misra, and Boeckmann Reference Budig, Misra and Boeckmann2012; Gash Reference Gash2009), but the broader set of findings in this research lead us to expect that motherhood is generally associated with lower labor force participation, fewer hours, and less personal income for women.

Indeed, our analysis of data taken from the World Bank, the OECD, and DHS indicates that a motherhood penalty exists at the national level when it comes to women’s labor force participation. Figure 8.4 indicates that women are less likely to be working when more children live in two-parent families. Accordingly, if strong and stable families discourage women from working, this may offset some of the positive effects that strong and stable families have on the economy.

Figure 8.4 Female labor force participation, by proportion of children in two-parent families: 2001–2015

Note: The y-axis in this figure provides the average female labor force participation based on the International Labor Organization database for 2001–2015 available from the World Bank. The x-axis splits the sample of countries into terciles based on the fraction of children being raised by two parents based on data from the World Bank, OCED, and the DHS. The sample includes 443 country-year observations from 97 countries representing the following regions: Africa (35), the Americas (11), Asia (18), Europe (30), and Oceania (3).

Public Safety

The strength of the family may also influence patterns of crime at the national level. This, in turn, could have implications for the health of the economy, insofar as crime discourages economic growth (Detotto and Otranto Reference Detotto and Otranto2010). A number of scholars (Akerlof Reference Akerlof1998; Nock Reference Nock1998; Sampson, Laub, and Wimer Reference Sampson, Laub and Wimer2006) have argued that marriage fosters male responsibility and discourages crime. As Sampson, Laub, and Wimer (Reference Sampson, Laub and Wimer2006, p. 467) note, marriage discourages crime among men inclined to criminal activity because “it creates interdependent systems of obligation, mutual support, and restraint that impose significant costs for translating criminal propensities into action.” Marriage also encourages men to engage in ordinary work-related routines, rather than deviant activities, and to spend less time with friends and acquaintances who might encourage criminal activity (Nock Reference Nock1998; Sampson, Laub, and Wimer Reference Sampson, Laub and Wimer2006). Marriage, then, is an important social control mechanism reducing the likelihood that men engage in delinquent or criminal acts.

Moreover, strong families reduce the odds that children engage in delinquent or criminal behavior as adolescent and young adults. Two-parent families tend to provide more attention and monitoring of children and adolescents (McLanahan and Sandefur Reference McLanahan and Sandefur1994), both of which discourage delinquent and criminal activity. Boys, for instance, who are raised in intact-married families are less delinquent, less criminally active, and less likely to be incarcerated, according to research in the United States (Antecol and Bedard Reference Antecol and Bedard2007; Harper and McLanahan Reference Harper and McLanahan2004). In Sweden, substance abuse and suicide are higher among children raised in single-parent homes (Weitoft et al. Reference Weitoft, Hjern, Hagland and Rosén2003). At the community level, communities with more two-parent families have less violent crime, at least based on research at the neighborhood and state levels in the United States and Canada (Lerman, Price, and Wilcox Reference Lerman, Price and Wilcox2017; Sampson Reference Sampson1987; Wong Reference Wong2011).

The connections between family structure and crime, both at the individual and the community levels, matter because crime tends to inhibit prosperity. Specifically, high levels of crime discourage male labor force participation (Eberstadt Reference Eberstadt2016), force businesses to spend more on security, and lead to significant public-sector costs. As the economists Detotto and Otranto (Reference Detotto and Otranto2010) note, “Criminal activity acts like a tax on the entire economy: it discourages … direct investments, it reduces firms’ competitiveness, and reallocates resources creating uncertainty and inefficiency.”

How, then, is family structure associated with violent crime? Using data from the United Nations Office on Drugs and Crime (UNODC), we find a strong negative association between the proportion of children raised in two-parent families and violent crime rates. Figure 8.5 indicates that the average homicide rate is more than four times larger in the countries in the bottom third of countries in terms of children living with two parents compared to countries in the top third. This association leads us to hypothesize that strong and stable families foster higher levels of public safety. In turn, we think that lower levels of crime are linked to stronger economic growth, ceteris paribus, thereby helping to explain any associations between family structure and economic growth in nations across the world.

Figure 8.5 Homicide rate, by proportion of children in two-parent families: 2001–2015

Note: The y-axis in this figure provides the homicide rate based on data from the United Nations Office on Drugs and Crime for the years 2001–2015. The x-axis splits the sample of countries into terciles based on the fraction of children being raised by two parents based on data from the World Bank, OCED, and the DHS. The sample includes 397 country-year observations from 83 countries representing the following regions: Africa (24), the Americas (11), Asia (16), Europe (30), and Oceania (3).

Marriage and Country-Level Economic Growth

To test our theories about the relationships between family structure and economic growth in countries across the globe, we compile data from a number of sources. We obtain historical data on marriage rates from the United Nations Statistics Division as well as the World Values Survey (WVS), which is a survey started in 1981 designed to gather national representative samples of individuals from almost 100 countries, which collectively constitute 90 percent of the world’s population, and the sample we use includes 340,000 total respondents providing on average about 4,000 respondents per country. Additionally, we merge in marriage statistics from census data from 81 countries obtained from the Integrated Public Use Microdata Series-International (IPUMS-I) database. Using the latest wave of data from all three data sets, we have marriage statistics for 129 countries scattered over a 46-year time period between 1968 and 2014, with a total of 401 unique country-year observations. Within these observations, the marriage rate is calculated as the fraction of adults who are legally married out of all adults.

We obtain historical data on the fraction of children who are living with two parents from the Organisation for Economic Co-operation and Development (OECD) and the Demographic and Health Surveys (DHS), which is a survey compilation beginning in 1984 including over 300 nationally representative surveys in over ninety countries. Combining the most recent OECD and DHS data sets, we have statistics on the fraction of children living in two-parent families for eighty-seven countries between 2001 and 2014, with a total of 416 unique country-year observations.

The other key measure of interest in our analyses is the gross domestic product (GDP) growth experienced by each country. We use data from the World Bank’s World Development Indicators to obtain data on per capita GDP, adjusted for inflation. As additional controls, we also merge in information about each country’s population, the fraction of the population that lives in cities, the proportion of the population under 15, the proportion of the population over 65, the education index (which taps the expected years of schooling as well as the historical average of years of schooling within each country), and the average life expectancy. Some of these measures (such as education) are likely to be influenced by family structure and so our estimates are likely to slightly the understate the actual correlation between family structure and economic growth.

We start by using a fixed effects regression to model the relationship between the economic growth and the fraction of adults who are married. The results are found in Table 8.1, indicating a positive relationship between marriage and economic growth. The coefficient indicates that a standard deviation increase (or 13 percentage point increase) in the fraction of adults who are married in a country is associated with a per capita GDP that is 8 percent higher.

The subsequent columns in Table 8.1 indicate how the size and statistical precision of this relationship between marriage and economic growth differs based on the controls that we include. Since each regression already includes country fixed effects, those fixed characteristics about each country (history, geography, natural resources) are controlled for in all of the analysis that we do. We find that the relationship is robust to the inclusion of additional time-varying controls such as population, urbanization, age distribution, education, and life expectancy. In each model, a standard deviation increase in the fraction of adults who are married is associated with about a 10 percent higher per capita GDP.

We then use a similar fixed effects model to explore the relationship between the fraction of children who are being raised by two parents in a country and its economic growth. Again, the results indicate a positive relationship (Table 8.2). The coefficient in this model is even higher and indicates that a standard deviation increase (also about 13 percentage points) in the fraction of children living with both parents is associated with a per capita GDP that is 13 percent higher. This relationship gets even larger when we include additional time-varying controls for each country. In the model with full controls, a standard deviation increase in the fraction of children who are living with two parents is associated with about a 16 percent higher per capita GDP. Overall, then, our results indicate that marriage and family structure are both strongly linked to patterns of growth using a sample of more than eighty countries across the globe.

Our analysis has some important limitations. One of the limitations of this analysis is that we have a limited number of observations per country and can only exploit a limited amount of the full variation of data over time. In addition, the results we present are descriptive and may not represent a causal relationship for a variety of reasons, including unobserved heterogeneity and reverse causality. What we show here is that changes in the proportion of adults who are married and changes in the proportion of children who are living in two-parent homes are both strongly associated with more economic growth. We assume that shifts in family structure predict shifts in growth, but changing economic conditions could also affect family structure. For instance, the work of economist Autor and his colleagues (2017) indicates that regional declines in employment in the United States fueled declines in marriage and increases in single parenthood in the regions most-affected by trade-related losses in employment (Autor, Dorn, and Hanson Reference Autor, Dorn and Hanson2018).

Efforts in the future may be able to identify factors unrelated to economic growth that had an independent effect on marriage and two-parent family rates that could be used as an instrumental variable. The downside of such an approach is that it is likely that identifying factors that influence the marriage rates of a sufficiently large sample of countries may be difficult to find. We present these results as a set of interesting descriptive findings of a largely overlooked factor, family structure, which may contribute to economic growth. Future research on economic growth should seek additional ways to explore the family structure–growth connection around the world.

Conclusion

Research on economic growth around the globe has tended to overlook the role that family structure may play in fostering growth. This is surprising, given a large body of evidence connecting marriage and family life to factors – from education to household savings to crime – that have potential implications for economic growth. In this chapter, we hypothesize that strong and stable families are associated with higher levels of economic growth in countries across the globe.

Indeed, we find that a significant association between family structure and economic growth. Every 13 percentage point increase in the proportion of adults who are married is associated with an 8 percent increase in per capita GDP, net of controls for a range of sociodemographic factors. Likewise, every 13 percentage point increase in the proportion of children living in two-parent families is associated with a 16 percent increase in per capita GDP, controlling for education, urbanization, age, population size, and other factors. There is clearly a link between family structure and economic growth.

However, this association does not prove causation. For instance, better economic growth may encourage increases in marriage and two-parent families even as poor economic performance may discourage marriage and family stability (e.g., Autor, Dorn, and Hanson Reference Autor, Dorn and Hanson2018). Our results cannot definitively prove that family structure has a causal impact on economic growth.

Nevertheless, we also note that the cross-national relationship between family structure, household savings, and crime are generally consistent with our expectations about how marriage and two-parent families foster a social environment more conducive to economic growth in countries around the world. It is striking that more two-parent families are linked to less crime and more savings. If nothing else, the patterns documented in this paper suggest that stronger families, higher household savings rates, less crime, and higher economic growth may cluster together in mutually reinforcing ways.

On the other hand, when it comes to labor force participation, we do not find consistent evidence that two-parent families are more or less conducive to work on the part of men or women. Not surprisingly, women in countries with a greater proportion of two-parent families are less likely to work, and surprisingly, men in countries with a greater proportion of two-parent families are less likely to be in the labor force, contrary to our expectations. On the other hand, after controlling for a number of demographic factors, our ancillary analyses indicate a positive association between male labor force participation and the proportion of children in two-parent families. Additional research ought to explore the links between family structure, men’s labor force participation, and economic growth, if any.

In conclusion, this chapter indicates that strong and stable families are linked to higher levels of economic growth in nations across the globe, despite the fact that marriage and two-parent families are in decline across much of the globe. Given the potential economic importance of marriage and family stability to a nation’s economic life, policymakers, business leaders, and civic leaders should pursue a range of public and private policies to encourage and strengthen marriage and stable families. That is because what happens in the family may not affect only the welfare of private families but also the wealth of nations.