Non-Technical Summary:
In our paper, we are analysing the link between gender of children and the divorce risks faced by the parents. In other words, we ask whether parents of girls divorce more often than parents of boys.
This question has received considerable attention in the academic literature. Several American studies have found that families with first-born girls are slightly more likely to divorce than families with first-born boys. Other American studies have however challenged this finding, and there is no evidence from other developed countries showing that such a relationship exists. Thus, the question remains largely unanswered.
In our paper, we leverage the Dutch administrative data to overcome major limitations of the previous studies, and we show that there is indeed a link between the child’s gender and the divorce risks faced by the parents: The Dutch families with daughters are more likely to divorce than families with sons. This relationship however does not hold universally. Up to the age of 12, the gender of a child has no influence over the parents’ divorce risks. It is only when the children become teenagers that the parents of girls start to face higher risks of divorce. We find that throughout the teenage years, parents of first-born girls face 5% higher risks of divorce than parents of first-born boys. This effect peaks at the age of 15, when the risks faced by parents of girls become almost 10% higher than the risks faced by parents of similarly-aged boys. In the following years, the excess risks decline, and eventually disappear when the children turn 19. The same age-specific associations are found among second- and third-born daughters, although the magnitudes of the effect are lower.
We show that our results cannot be explained by son preference of Dutch parents, or by a host of other previously-postulated mechanisms such as the non-random assignment of gender to children, or higher financial costs of daughters. Instead, our study suggests that the increased odds of divorce should be attributed to the relationship strains between parents and their teenage daughters. This claim is supported by an analysis of a large survey of Dutch households, which shows that parents with teenage daughters are more likely to disagree about the way their child should be raised, and they report lower satisfaction with their life and the quality of family relationships compared to parents with teenage sons. Teenage daughters themselves tend to report worse relationships with their fathers.
Additional analyses of the administrative data emphasize the role of the relationship between the father and his daughter. A particularly interesting result comes from an exercise in which we split the sample into two groups, distinguishing between fathers who grew up with sisters, and fathers who did not. This split yields a contrasting set of results: Fathers who grew up without sisters are subject to a pronounced increase of divorce odds when their daughters become teenagers, whereas the fathers with sisters are not. Remarkably, the fathers who grew up among sisters face no increase of divorce odds for having a daughter. This suggests that men who have been exposed to more daughter-specific parenting styles and mixed-gender relationships when growing up may be better prepared for similar relationships in their own households. We also find that the increased odds of divorce are more pronounced among parents who are likely to hold traditional gender norms, such as the first-generation immigrants, or parents born in the first half of the 20th century.
We also confirm that the increased odds of divorce associated with teenage daughters are not unique to the married couples in the Netherlands. We show that the same effect manifests among the Dutch couples who are cohabiting, and among the married couples in the U.S. In both cases, the increased odds of separation are considerably larger than those faced by the Dutch married couples. Similarly to our principal analysis, we do not find evidence of son preference among American couples. This suggests that the divorce effects of son preference – if existent – are marginal compared to the effects of having a teenage daughter.