Education, Gender, and Cohort Fertility in the Nordic Countries

Over the past 60 years, European countries have experienced a remarkable expansion of education at the upper-secondary and tertiary levels. The proportion of persons lacking degrees beyond the basic level has halved within one generation. In 2015, only 15% of 25–34-year-olds in Europe (EU 22) were without post-basic education compared to 30% of 55–64-year-olds (OECD 2016, Table A1.3). This change has been particularly pronounced among women. In almost all post-industrial societies, young women now outnumber men in attaining tertiary degrees (Vincent-Lancrin 2008; Charles 2011).

A vast array of literature has documented the fundamental role of educational attainment in shaping women’s childbearing patterns, especially their timing of births and their childlessness. The findings generally reveal that highly educated women tend to have their first child at an older age and that their levels of childlessness are higher than those of less educated women (for reviews, see Hoem et al. 2006; Kravdal and Rindfuss 2008; Tanturri et al. 2015; Nitsche et al. 2015).

Little is known about whether and how the associations between women’s increasing educational attainments and their ultimate fertility have changed over time (van Bavel 2012; Tanturri et al. 2015). However, recent studies suggest that educational expansion has had only limited effects on patterns and developments in ultimate fertility (Sobotka et al. 2017) and childlessness (Beaujouan et al. 2016) by educational level in continental European countries. In most European countries, highly educated women are still more likely to remain childless than low-educated women (Wood et al. 2014; Miettinen et al. 2015; Tanturri et al. 2015).

Compared to research on women’s fertility, research on the fertility of men is scarce and almost uniformly suggests a positive association between men’s educational attainment and their fertility outcomes (Kravdal and Rindfuss 2008; Trimarchi and van Bavel 2017; Tanturri et al. 2015; Miettinen et al. 2015; Nisén et al. 2014, 2017; Nisén 2016). However, almost no research has addressed the ways in which increases in educational attainment and the reversal of the gender gap in education to women’s advantage are related to changes in men’s fertility outcomes.

When women regularly attain educational levels that are similar to or higher than those of men and economic and social roles become more similar, gender differences associated with education and fertility would also be expected to diminish. Studies examining the timing of entry into motherhood and fatherhood in France and women’s and men’s fertility postponement and childlessness in Britain have resulted in some support for this ‘convergence hypothesis’ (Winkler-Dworak and Toulemon 2007; Kneale and Joshi 2008). However, whether the gendered patterns in fertility may indeed converge is an open question. It is furthermore unclear whether differences in women’s and men’s fertility would converge within educational groups only or whether educational differences in fertility outcomes would also diminish. In the first case, we would find ‘convergence across gender lines’, while educational differences persist; in the latter case, we would find ‘convergence across educational lines’ and consequently diminishing differences in fertility outcomes between educational segments of both women and men.

This study examines the ‘convergence hypothesis’ in greater detail by differentiating between ‘convergence across gender lines’ and ‘convergence across educational lines’ in cohorts born since the 1940s in the Nordic countries. The Nordic welfare states are often considered to be forerunners in developing both social and gender equality. Therefore, when researching the ways in which gendered associations between education and fertility have developed over time, the Nordic countries may establish a blueprint for similar changes in other post-industrial societies. In the Nordic countries, the aim of social equality is to reduce differences between social strata (here represented by the level of education). The aim of gender equality is to foster equal participation of women and men in paid work, family care work, and society at large. Considering both core aims, equality is generally understood as ‘equality of outcome’ (Phillips 1999). The aim is to create a society in which social and gender differences are minimized. Thus, the Nordic countries have been forerunners in Western Europe in promoting formal education and establishing a largely unified educational system. They have also been forerunners in implementing policies that reduce the opportunity costs of family formation and support both sexes’ participation in paid employment and childrearing. Given these aims and policies, we might expect relatively small differences in fertility patterns between educational groups, between men and women, and between the Nordic countries. Indeed, previous research suggests that unlike in many other societies, educational differences in women’s completed cohort fertility in the Nordic countries have diminished over time (Andersson et al. 2009; Kravdal and Rindfuss 2008; Miettinen et al. 2015).

Our study compares ultimate fertility by gender and educational attainment for cohorts born in 1940–1967/69 for men and 1940–1972/74 for women in four Nordic countries: Denmark, Finland, Norway, and Sweden. In other words, we compare fertility differentials between (a) birth cohorts, (b) men and women, (c) educational segments, and (d) four Nordic countries. This allows us to follow the change in fertility patterns across educational and gender lines and to draw conclusions about the potential forces behind these developments.

This analysis focuses on two aspects of ultimate fertility: completed cohort fertility, or Cohort Total Fertility (CTF, mean ultimate number of children), and the levels of ultimate childlessness, measured at age 45 for men and age 40 for women. We harmonize longitudinal individual-level data drawn from the national population registers of the Nordic countries and use a straightforward cross-sectional approach of measuring education simultaneously with ultimate fertility (at age 45 or 40).

We adopt a cohort approach. Period measures such as the period Total Fertility Rate (TFR) are otherwise common in official statistics. However, they mainly provide information on short-term changes in fertility behaviour. As a background, Fig. 1 shows the development of TFRs in the four Nordic countries during 1975–2015. The TFRs of each country show similar developments, with strong fluctuations in all countries (strongest in Sweden; see also Andersson 2000, 2004). Such fluctuations in the TFR are commonly due to changes in the timing of births, which does not affect the CTF.

Open image in new window

Our analysis partly builds on a previous study by Andersson et al. (2009), which focused on the cohort fertility of women in the Nordic countries. A key contribution of the present study is that it also covers male cohorts, with the aim to produce a gender comparison of fertility patterns over time. The present study also covers more recent cohorts and developments. Finally, it uses a somewhat different approach than Andersson et al. (2009), which defined education as the attainment at age 30 and used a prospective cohort approach (see our methods discussion).

Our empirical analysis begins with an overview of the development of CTF and childlessness among men and women in each Nordic country. Then, we summarize the developments in educational attainment. Finally, the analysis focuses on the developments of educational differences in CTF and ultimate childlessness across cohorts, comparing women and men within and between the four countries. Before moving on to the data and empirical analysis, we give a condensed overview of prevalent theoretical perspectives on the relevant associations.

Our analyses are based on individual-level data drawn from population registers and from the registers of educational degrees of each Nordic country. The Nordic countries have a long history of full and reliable registration of their resident populations and their vital events. Residents of the Nordic countries have a unique personal identification number, which enables an accurate, computerized linking of information from different data sources. Our data cover the entire population of Denmark, Norway and Sweden as well as an 11% random sample of the population of Finland.

We follow the logic of a true birth-cohort design. We focus on women and men who were born in their respective Nordic country. Further, we only include data on persons who were alive and resided in their birth country at age 40 for women and age 45 for men. Age is measured at the end of the calendar year. Return migrants, i.e. persons who were born in the respective country, emigrated and had returned by age 40 (women) or 45 (men), are included in the analysis. Other migrants, that is, immigrants born abroad and emigrants who did not return to their birth country, are not included.

The analysis covers 5-year cohorts born from 1940 to 1974. We have complete cohort data for all individuals born in 1940–44 up to the 1960–64 5-year cohort. For the youngest cohorts (1965–74), the study populations vary by country and by sex due to the different age limits for women and men and the last year for which data are available. For women born in 1965–69, we have complete data for all four countries. For those born in 1970 onwards, we have a 5-year cohort coverage (1970–74) for Denmark and Norway and a 3-year cohort coverage (1970–72) for Finland and Sweden. Likewise, for men born in 1965–69, our study comprises complete 5-year cohorts for Denmark and Norway and 3-year cohorts for Finland and Sweden.²

The two outcomes of our study, completed cohort fertility and ultimate childlessness, are based on the dates of all recorded childbearing events (live births). Births to return migrants while they were abroad are also included, although some births may be missing (e.g. if the child remained or died abroad). Unlike in survey data on men in which births in previous unions and non-marital births in particular often go unreported (Rendall et al. 1999), the registers cover men’s childbearing histories nearly as completely as those of women. Nevertheless, some slight proportion of the lower cohort fertility and higher childlessness among men than among women likely reflects the fact that fatherhood is sometimes not registered.³

We measure cohort total fertility and childlessness at age 45 for men and 40 for women. A small portion of women and men have additional children beyond these ages,⁴ but choosing higher age limits would lead to the exclusion of the most recent cohorts and thus limit our opportunities to assess the latest trends in completed fertility and ultimate childlessness.

We excluded data on the very few cases with missing information on educational attainment. The exception is Finland, whose data do not allow us to distinguish persons with missing educational information from those with the lowest educational level. Although the data were maximally harmonized, the country comparisons may be somewhat hampered by differences and changes in educational systems, educational codings and classification procedures. However, the comparisons benefit from the fact that the Nordic educational systems are fairly similar and that most data differences do not affect the comparability of trends.

Compared to the study by Andersson et al. (2009), we have slightly different education data for Norway. For this country, we could apply a revised classification of education, which is better adjusted towards international standards (Statistics Norway 2006). Because of this revision, the proportion of low educated persons in Norway is lower and the proportion of medium educated is higher. The new classification is more in line with the standard of the other Nordic countries.

Methodologically, one central question is how and when to measure education and fertility. Education and childbearing are often interacting processes: education affects childbearing, but childbearing, especially at an early age, may also affect further education. Results on the association between education and fertility are therefore sensitive to when (at what age) and how (by which measure) education is measured (Hoem and Kreyenfeld 2006; Kravdal 2004), and the same applies to different methods of studying fertility outcomes. Our approach differs somewhat from that by Andersson et al. (2009), where education was measured at age 30 and a prospective study design was applied. In the Nordic countries, movements in and out of education occur frequently, and many people become parents prior to finishing their highest level of education (Hoem et al. 2006; Tesching 2012). Therefore, determining the ideal age to measure education is difficult. We explored alternative possibilities. Some cases resulted in minor differences compared to our study design and to the previous study by Andersson et al. (2009). Since the differences did not substantially affect the overall results and their interpretation, we settled for measuring educational attainment at the same age as when we measure fertility.

In what follows, we present our results as figure diagrams. The data are also made available as tabulations in the Electronic Supplementary Material.

Using harmonized longitudinal individual-level data drawn from national population registers, we explored the development of ultimate cohort fertility and ultimate childlessness for cohorts born since the 1940s in four Nordic countries and compared the potentially gendered developments of educational differentials in fertility outcomes. In all four countries, the segment of persons with only a basic education has shrunk among both sexes. Among men, upper-secondary education is now the most common level attained, whereas among women, tertiary education has expanded substantially. These gendered developments have led to a reversal of the educational gender gap to women’s advantage. Comparing the development of ultimate fertility and childlessness across cohorts in the Nordic countries, we find similar trends in all countries, with the partial exception of Finland. The levels of ultimate fertility have remained fairly stable, especially in female cohorts born in the 1950s onwards, and differences between countries have effectively disappeared, especially with respect to female CTF. Childlessness increased in the early cohorts in all countries and more so among men than among women. In Finland, it increased across all cohorts.

The associations between gender and education in fertility outcomes have undergone fundamental changes across the cohorts. The developments are quite similar in the four countries. For men, educational fertility differentials have not changed substantially. In line with previous research on men’s fertility (e.g. Kravdal and Rindfuss 2008; Trimarchi and van Bavel 2017; Miettinen et al. 2015), we find that low-educated men have lower completed fertility and higher childlessness than medium or highly educated men. These disparities show a remarkable persistence across cohorts.

In contrast to men, women’s educational fertility differentials have changed substantively. Contrary to findings from many other European countries (Sobotka et al. 2017; Beaujouan et al. 2016), the negative educational gradients in CTF observed in older female cohorts have disappeared. Among women in the younger cohorts, no differences are discernible in ultimate fertility between educational categories of women in Denmark, Norway and Sweden.

Changes in women’s education-specific levels of childlessness are even more pronounced than the changes in the CTF levels. In all four countries, the childlessness of highly educated women remained fairly constant. Childlessness levels of medium-educated women converged towards those of the highly educated. In contrast, childlessness among low-educated women has been on an almost continual rise and has surpassed the levels of highly and medium-educated women. These differential trends have led to a reversal of the educational gap in childlessness among women in Denmark, Norway and Sweden. Highly educated women from the older cohorts had higher levels of childlessness than lower educated women, but in the youngest cohorts in Denmark, Norway and Sweden, the levels are highest among low educated women.

In Finland, we mainly observe a convergence in childlessness levels between different educational categories of women. In the case of CTF levels, slight differences in ultimate fertility by the educational categories remain. Evidently, the gendered patterns of fertility dynamics appear to evolve at a somewhat slower pace in Finland than in the other Nordic countries (for another similar example of differential change in gendered family dynamics, see Andersson et al. 2006).

The developments in women’s and men’s educational fertility differentials have brought about a new pattern of gender similarity in the Nordic region. The ‘convergence across gender lines’ in childlessness among the low educated is particularly noteworthy. In our oldest study cohorts, childlessness levels were highest for low-educated men and high-educated women, whereas in younger cohorts in all countries, childlessness is highest for the low educated regardless of gender. The new gender similarity accompanies a new ‘divergence across educational lines’ in fertility outcomes. The development of the youngest cohorts may well indicate that social inequality in childbearing is intensifying in that lower educated persons are increasingly left behind in family formation and that this applies to women and men equally. This, of course, goes hand in hand with the strongly diminished sizes of the low educated segments in each country.

Compared to the study by Andersson et al. (2009), including men and more recent cohorts provided novel insight into Nordic fertility dynamics and input to their potential explanation. Andersson et al. (2009) emphasized that welfare-state policies that focus on supporting gender equality may lead to a convergence in fertility across different educational groups. Extending their cohort series, we revealed that low educated women increasingly differ from the medium and higher educated. Adding analyses of male fertility allowed us to detect emerging gender similarities and persisting differences in completed fertility and childlessness.

How can we interpret these results in light of the outlined theoretical framework, namely, economic considerations, gender-equity assumptions, and institutional aspects, that links education and fertility? Our study can only provide hints about the numerous factors at play behind the developments we have observed. Our findings suggest that the fertility-stimulating effects of socioeconomic resources may have become more important for both women and men. If education is taken as a proxy for a person’s earnings potential, having a sufficiently high income or having the economic means to sustain a family may have become an increasingly important prerequisite for having (more) children for both women and men. The Nordic countries have numerous services and income transfers that should encourage childbearing among all social groups, but having children still appears to have become increasingly difficult for low-educated women and men. Additionally, the difficulties reconciling a career with family building that previously hampered the childbearing of highly educated women in particular seem to have been overcome.

We find some support for recent family-demographic gender theory and for the importance of institutional factors in shaping family-demographic outcomes. We assume that the gender-egalitarian ideology of the Nordic welfare regime, which originally targeted mainly women (Hernes 1987), contributed to halting increases in childlessness and declines in ultimate fertility. Institutional support for parents, particularly day care for children and support for mothers to remain in the labour force and for fathers to engage in childcare, eased the burden for mothers, countered the negative career consequences of motherhood, and reduced conflicts between women’s employment and family care. This orientation of the Nordic welfare regime seems to have mitigated the negative consequences of increased female education and labour-force participation on fertility.⁶

However, it is also noteworthy that even in the Nordic countries, some gender differences prevail. For instance, among men there is a clear positive education gradient in CTF, characterized by a large gap between the low- and middle-educated, while for women, the most prevailing pattern is a convergence in cohort fertility. Despite the fact that men and women in the Nordic countries are comparatively similar in their economic and domestic roles, differences prevail in gender norms, expectations and the potential to realize parenthood. In Nordic families, men are rarely the sole breadwinners, but their economic situation still seems to play a crucial role in family formation. Since there are more men than women in each cohort and women have higher education than men, it could well be that men who have less potential of being an economic provider (Chudnovskaya 2017b), who share (or can share) less in parenting (Duvander and Johansson 2014) or have other cognitive traits related to low education (Kolk and Barclay 2017) may find it increasingly difficult to form a family and have children.

Our findings have major implications for demographic research. They underline the need to carefully scrutinize the schemas of gender and social strata (here, education) that are applied in fertility research. As other researchers have also noted, we should be wary of the widely held notion that highly educated women necessarily have higher opportunity costs than low-educated women and that they therefore have fewer children than low-educated women (see also: Hoem et al. 2001; Andersson et al. 2009). This notion rests on the assumption that mothers are the sole caregivers of children and that motherhood and paid employment are strongly at odds, and it overlooks the value of available childcare, parental leave regulations, and partner’s engagement in childcare, all of which may modify the opportunity costs. In principle, the result may be that highly educated women have lower opportunity costs than the less educated (see Hoem et al. 2001; Kravdal and Rindfuss 2008). Overall, opportunity costs may not be as important as is usually assumed. Even if highly educated women have the largest earnings losses following from family leaves (Evertsson 2016; England et al. 2016), their remaining income may still be more than sufficient to maintain the family’s previous level of living, while the comparatively lower earnings loss of low-educated mothers may make it difficult for these women to make ends meet. Our research results clearly suggest the need to modify the existing economic concept of motherhood and to view women as breadwinners just as men are. The prevailing or even intensifying differences in CTF and childlessness between low-educated men and men with more education call for more thorough reflections and more stringent theoretical approaches to the links between changing gender relationships, manhood, fatherhood and men’s fertility (Hobson 2002).

The widening of the educational gap in childlessness for both women and men supports our presumption that fertility research should focus more attention on family-demographic developments among the low educated. With the increase in women’s and men’s educational attainment, the results indicate that the low educated have become an increasingly marginalized population segment not only with regard to their labour market position but also with regard to family formation. In the Nordic countries, affordable high-quality childcare has long been available to all children, and most fathers participate in childrearing. Therefore, the increase in childlessness among the lowest educated may not be readily explained by existing theories, such as gender theories on the division of work and care within the family. In particular, previous decline in marriage and increasing levels of union instability are other developments that may have hindered childbearing, particularly among the lower educated (Kravdal and Rindfuss 2008; Jalovaara and Fasang 2015, 2017). Research on the Nordic countries shows that the lowest educated men as well as women have the lowest likelihood of entering cohabitation and marriage, and the highest likelihood of union dissolution, and that stable employment promotes union formation and union stability regardless of gender (Bracher and Santow 1998; Jalovaara 2012, 2013; Jalovaara and Fasang 2015; Cooke et al. 2013). Indeed, one recent study suggests that the negative educational gradient in men’s childlessness operates chiefly via selection into union (Trimarchi and Van Bavel 2017). The importance of unions as mediating factors is largely missed by the theoretical discussion on the changing links between gender, education and fertility, as it focuses on couples. Overall, our research suggests that understanding fertility development in light of increased educational attainment and the reversal of the gender gap in education requires more thorough investigations of the demographic consequences of the increasing economic and social marginalization of low-educated women and men. Clearly, the currently prevailing demographic research perspective ‘from above’ needs to be complemented by a perspective ‘from below’.

Theories on men’s fertility highlight men’s position as family breadwinners and, more recently, as caregivers as well. Many researchers have noted the need to expand fertility research to include men (Bianchi 1998; Goldscheider and Kaufman 1996). Our research supports this view and further suggests that a gender comparison is essential. Our gender comparison of fertility developments allowed us to detect the increasing gender similarity, the increasing childlessness among the low educated that embraces both men and women, and the profound change in the link between education and fertility among Nordic women but not among men. It also demonstrated that despite these fundamental changes there are consistent patterns of gender differences which call for further investigation and explanation. Our research suggests that empirical and theoretical efforts to illuminate men’s fertility behaviour and the intersection of gender and social strata should not concentrate only on men but should use a gendered perspective.