Find higher fertility among men with higher cognitive ability among Swedish men using taxation & conscription registers; low income & low cognitive ability independently predict low fertility & high childlessness

Do income and marriage mediate the relationship between cognitive ability and fertility? Data from Swedish taxation and conscriptions registers for men born 1951–1967. Martin Kolk, Kieron Barclay. Intelligence, Volume 84, January–February 2021, 101514. https://doi.org/10.1016/j.intell.2020.101514

Highlights

• We find higher fertility among men with higher cognitive ability among Swedish men using population-level taxation and conscription registers

• Low income and low cognitive ability independently predict low fertility and high childlessness

• Strong evidence of lower access to marriage and reproduction among low status men in contemporary Sweden

Abstract: Recent evidence suggests a positive association between fertility and cognitive ability among Swedish men. In this study we use data on 18 birth cohorts of Swedish men to examine whether and how the relationship between cognitive ability and patterns of childbearing are mediated by income, education and marriage histories. We examine whether the expected positive associations between cognitive ability and life course income can explain this positive association. We also explore the role of marriage for understanding the positive gradient between cognitive ability and fertility. To address these questions we use Swedish population administrative data that holds information on fertility histories, detailed taxation records, and data from conscription registers. We also identify siblings in order to adjust for confounding by shared family background factors. Our results show that while cognitive ability, education, income, marriage, and fertility, are all positively associated with each other, income only explains a part of the observed positive gradient between fertility and cognitive ability. We find that much of the association between cognitive ability and fertility can be explained by marriage, but that a positive association exists among both ever-married and never-married men. Both low income and low cognitive ability are strong predictors of childlessness and low fertility in our population. The results from the full population persist in the sub-sample of brothers.

Keywords: FertilityChildlessnessCognitive abilityIncomeSweden

6. Discussion

In our paper, we show that while income is strongly associated with cognitive ability, men with below average cognitive ability have fewer children, even after adjusting for income. We also find that these differences are magnified for childlessness, and are also very strong for entry into marriage. Consistent with previous research, we find that income and fertility are very strongly associated (Chudnovskaya, 2019; Kolk, 2019), but that the relationship between cognitive ability and fertility persists net of the mediation of income. This is particularly true at lower income levels. Men with low cognitive ability who are above the median in cumulative income between age 18 and 45 have approximately the same number of children as men who score highly on cognitive ability. However, men with low cognitive ability are much less likely to find themselves in the top half of the cumulative income distribution. Among ever-married individuals, the association between cognitive ability and fertility is strongly attenuated, and only really suggests lower fertility among men with the lowest scores on cognitive ability. When comparing full biological brothers with each other, we find a strong positive fertility and cognitive ability gradient even after adjusting for income. Overall, our results indicate that the primary reason that we observe low fertility among men with lower cognitive ability is because of their failure to attract a partner for stable unions for childbearing. In addition to confirming previous findings on cognitive ability and fertility in Sweden (Kolk & Barclay, 2019), the findings of this study provide evidence for the importance of partnership formation for fertility, as well as showing that the intelligence-fertility association persists even after taking cumulative income into account.

Another intriguing empirical pattern that we have observed is that, although men with high cognitive ability have more children overall, men of average cognitive ability had more children than men with high cognitive ability scores among those who never married. These never-married men with high cognitive ability are too few to affect the population-level intelligence-fertility gradient, but may indicate a sub-population that either voluntarily abstains from childbearing and marriage, or in other ways have life trajectories that are associated with high education and income but not traditional patterns of family formation. In our full population analyses, when adjusting for cumulative income, we find that the men with the highest cognitive ability scores have slightly lower fertility and higher childlessness than men with median cognitive ability scores. After adjusting for income we observe slightly lower fertility among high IQ never-married men (left-panels of Fig. 6, Fig. 7), as well as slightly lower fertility among men with high cognitive ability (see Fig. 1). However, our finding that higher cognitive ability men have higher childlessness and lower fertility than men with similar incomes but average cognitive ability is not replicated in sibling comparison models.

In our sibling comparison models we consistently observe lower fertility among men in the bottom half of the cognitive ability distribution. The difference between our population level models and the sibling models is intriguing. Although the results from our population level models are key to understanding how cognitive ability may be distributed in the following generation (though without data on women we cannot speculate about this), the sibling comparison models effectively adjust for all factors shared in the family of origin. It is certainly possible that the results in the full population are confounded by factors that are jointly associated with both cognitive ability as well as fertility outcomes, for example negative experiences in early adulthood such as health shocks or substance abuse.

We believe that our study highlights the importance of examining and interpreting gross associations between cognitive ability and fertility by taking account of the associations between cognitive ability and mediating dimensions of social status and partnership formation. The sociological and demographic literature suggests great variation across the West in the associations between income and fertility, and education and fertility. The findings from our Swedish data may not generalize to other countries. In other high-income countries, the interrelationships between education, income, marriage, and fertility, differ in important ways from Sweden, and our results may to some extent be contingent on the aggregate positive relationships between status and marriage and family formation in Sweden. Nevertheless, we think that the fertility disadvantage of very low cognitive ability men is likely widespread across OECD contexts and that using datasets where such individuals are fully included is important if researchers are to be able to make population-level inferences. Future research on cognitive ability and fertility is therefore advised to pay careful attention to contemporary research in family sociology, demography, and economics on the overall relationship between status and fertility in the focal society. Importantly, the associations between income and fertility and education and fertility typically differ by gender. Unfortunately, we cannot examine any gender differences in the intelligence-fertility gradient in Sweden as we have male-only conscription data.

Our findings also contribute towards the increasing evidence for social polarization of childbearing in many Western countries. We find that the proportion childless and the proportion that never-marry is very substantial among men with lower cognitive ability. We find large separate effects where both low income and low cognitive ability are each strongly associated with a higher probability of childlessness and low completed fertility. When a man has both low income and low cognitive ability, fertility is even lower. This corresponds to the findings from a growing literature that shows that men with low income, low levels of education, worse health, and low cognitive ability, are much less likely to find a childbearing partner in Scandinavia (Barclay & Kolk, 2020; Jalovaara et al., 2019; Jalovaara & Fasang, 2020; Kolk, 2019). Fertility in Scandinavia has traditionally been characterized by relatively small social differences between groups. Our findings of differences by cognitive ability in probabilities of childlessness and ever-marriage of 20 to 30 percentage points clearly show that partnership and childbearing are increasingly less likely for many men with low cognitive ability in contemporary Sweden.