Skills, education and fertility and the confounding impact of family background

Skilled and educated women have on average fewer children and are more likely to remain childless than the less skilled and educated. Using rich Swedish register data, we show that these negative associations found in most previous studies largely disappear if we remove the impact of family background factors using twin (or sibling) fixed effects. For males, human capital measures are virtually unrelated to fertility, but this again masks the role of family background factors: more educated and skilled males tend to have more children than their less skilled peers once we use twin/sibling fixed effects to remove family background factors. Hence, for both men and women, human capital and fertility become more positively associated once the joint family components are removed, i.e. when studying the within-family associations. The one human capital measure which deviates from these patterns is non-cognitive ability, which has a very strong overall positive association with fertility, an association which instead is muted within families. We end by showing that these results can be reconciled in a stylized theoretical model where family-specific preferences for fertility shape the relative investments in different types of skills and traits when children are small as well as the choices, in terms of family formation and human capital investments, these children make when they enter into adulthood.