Based on a new, extensive database on male fertility in 17 high-income countries, we explore trends in the differences in the quantum and timing of male and female fertility. We find that, first, the level of male fertility relative to the level of female fertility can vary considerably across countries and over time. Male fertility, as measured by the total fertility rate (TFR), can be both higher and lower than female fertility; although in recent years it has generally been lower rather than higher. This observation is consistent with the results of Schoumaker (2019) based on survey and census data, and our further analysis for a subset of six countries shows that it also holds for cohort fertility rates (CFRs).
Our counterfactual calculations provide evidence for the claim that temporal fluctuations in the ratio between male and female TFR are often driven by age differences between partners and differences in postponement behavior, and our analysis of CFRs also supports this conclusion. Our second main finding is that age differences between fathers and mothers stayed constant or decreased in most of the countries in our study, except in the Eastern European countries and eastern Germany. While this general tendency is in line with expectations based on gender theories, the differences that currently exist across countries seem to be affected by more factors than just differences in gender equality levels.
While our findings indicate that male fertility has generally been lower than female fertility in recent years, this trend does not seem to hold globally. Schoumaker (2017, 2019) showed that in contexts in which polygyny is practiced and populations are rapidly growing, male fertility levels can sometimes be twice as high as female fertility levels. His results suggest that this pattern is driven to a large extent by age differences between couples. This finding is in line with the results of our counterfactual analysis, which indicate that extreme cases tend to disappear when the fathers are assigned to the same cohort as the mothers, and that large gender differentials in fertility levels are often driven by differences in fertility timing. At the other extreme, the lowest TFR ratio reported in the literature is for England and Wales in 1973, at around 0.89 (Schoen 1985). The value we have found for eastern Germany, 0.84, is below this level, and might indicate that eastern German males have been experiencing what Schoen (1985) called a “birth squeeze”: i.e., in eastern Germany, the unequal number of men and women in the reproductive age range is having an impact on the fertility of men. Generally, the variation in TFR ratios we observe across countries and over time is not negligible. Country groups with cultural and/or political similarities seem to have more similar trend patterns. This finding requires further investigation.
The mean age differences between men and women we have detected in our data are close to the orders of magnitude found in other studies (e.g., Kolk 2015). They are, however, lower than the high values observed in some African countries, where mean differences of more than 10 years have been estimated (Schoumaker 2017). Nevertheless, the variation across countries and over time in our 17 high-income countries is substantial. This heterogeneity of age differences is somewhat puzzling. In particular, our findings suggest that while European countries that score high on the gender equality index have smaller age differences, Japan does not seem to fit this picture. Even though Japan scores low on gender inequality indices (World Economic Forum 2017), the age differences among parents in this country are the lowest we have found in our study. This finding requires further investigation. Another factor that might be important is the cross-country variation in the share of migrants and the parental age differences that prevail in their countries of origin. We are, however, unable to study this potential factor, given that for most of the countries and years in our sample, we are unable to separate births by nationality or migration background status. We thus leave the exploration of this question to future research.
Looking at gender differences in fertility postponement, we find that gender differences tend to decrease over time, except in Eastern Europe, where age differences have been increasing. In Eastern Europe, this process has been accompanied by a process of societal restoration (Fodor and Balogh 2010). We also find that in some countries (e.g., Finland and France), the mean age difference has been stable over the period for which we have data. Our comparison of countries suggests that the Danish case, described by Nordfalk et al. (2015), is among those with rather large gender differences in fertility postponement. More research is needed to explain this variation across countries.
It is difficult to assess to what degree the observed patterns reflect gender differences in fertility preferences with respect to both the quantum and the timing of fertility. Surveys in low-income countries designed to assess the ideal number of children tend to report that the ideal number is usually much higher than the number of children actually born (Esteve et al. 2020). Longitudinal research has also shown that individuals tend to adapt their family size ideals to the number of children they were able to have (e.g., Kuhnt et al. 2017). This implies that the ideal number of children is not only a reflection of people’s preferences, but is also moderated by triggers and constraints that affect their fertility biographies. For Europe, data from the Eurobarometer survey 2011 indicate that in most countries, men tend to report lower ideal family sizes than women (Testa 2012). These results are generally consistent with our findings, which show that men have lower fertility than women. In addition, our TFR estimates for 2011 are highly correlated across countries with the Eurobarometer results on ideal family size for both men and women (> 0.9). However, the Eurobarometer results also show that in most countries, the gaps between the ideal and the actual number of children are larger for men than for women (Testa 2012). This observation could be interpreted as providing support for the view that men might be less able than women to realize their fertility desires, which could, in turn, be a result of the sex ratio imbalances.
Regarding preferences on the timing of births, the Eurobarometer survey data show that across member states of the European Union (EU25), both the ideal age to become parent and the latest age at which a person should have children are higher for men than for women (Testa 2006). The ideal age to become a parent is on average around 2 years higher for men than it is for women (Testa 2006), with women setting the ideal age for men higher (about 27.5 years) than men themselves (around 27 years). For the latest age at which a person should have children, the gaps between women and men are even larger. Both women and men set the age deadline for women on average at around 41 years, while both men and women set the age deadline for men at around 46 years (Testa 2006). Billari et al. (2011) used data from the European Social Survey and found slightly higher mean numbers both for women (41.7 years) and men (47.3 years), with some variance across countries. These survey results are generally in line with the gaps we identified. They are also in line with our finding that fertility decreases for men as well after age 45, even though many men are biologically able to have children at ages above 45.
The main outcomes we study in this paper—the TFR and the mean age at childbirth—are not the only measures of the quantum and the tempo of fertility. For instance, the proportion of childless individuals and the age at first birth have also received considerable interest in the literature. While this research has mostly focused on women, a small number of studies have also included men (Paavilainen et al. 2016). However, one of the limitations of the register data we employ is that we cannot use them to calculate the number of childless individuals. In addition, analyses by parity, such as by first birth, are only possible for women, if at all, as none of the birth registers records the parity of the father. Moreover, the mean parental age difference is not the only measure of the extent to which the paternal and the maternal age differ (e.g., Kolk 2015). However, using other measures, such as the variance or the standard deviation of the age difference, requires information on the joint age distribution of mothers and fathers, which cannot be derived from the ASFRs we provide as part of the HFC.
The new database we created to study male fertility is not limited to the questions we have investigated in this paper. It can, for example, be used in macro-level investigations of associations between male fertility levels and important economic and social indicators, including gender equality measures. There are many influential publications that have studied these relationships among women (e.g., Brewster and Rindfuss 2000; Myrskylä et al. 2009), and comparing the outcomes for female fertility with the results for male fertility might provide important insights. Such analyses will likely further improve our understanding of the role of gender in current fertility trends. The database can also be used in comparative analyses of trends in the paternal age at childbirth. Advanced paternal age is an important predictor of health outcomes of children, and has been attracting increasing attention in recent years (e.g., Khandwala et al. 2017). Thus, the database offers many promising avenues for future research, and we invite other researchers to make use of this new data resource available as part of the Human Fertility Collection (2019).
Posts
