|Authors:||Pedro Carneiro , Costas Meghir and Matthias Parey|
|Date:||19 September 2007|
|Type:||cemmap Working Paper, W07/15|
There is a striking increase in inequality in children's home environments over the last 50 years (McLanahan, 2004). These are measured as differences in age of mothers of young children (below 5), maternal employment, single motherhood, divorce during the first 10 years of marriage, father's involvement, and family income, for mothers with different levels of education. This trend is cause for great concern because the home environment is probably the best candidate for explaining inequality in child development.
Proposals to address this problem often rely on changes to the welfare system. However, given that home environments are rooted in the experiences of each family, they are probably difficult to change if we rely only the welfare system, while more direct interventions require invading family autonomy and privacy and are notoriously difficult to enforce. Therefore, one possible alternative is to target future parents in their youth, by affecting their education, before they start forming a family. In this paper we assess the potential for such a policy, by estimating the impact of maternal education on home environments and on child outomes.
We provided a unified analysis of different aspects of child development, including cognitive, noncognitive, and health outcomes, across ages. We also estimate the impact of maternal education not only on parental characteristics like employment, income, marital status, spouse's education, age at first birth, but also on several aspects of parenting practices. Our paper provides a detailed analysis of the possible mechanisms mediating the relationship between parental education and child outcomes. Finally, we compare the relative roles of maternal education and ability, and we show how the role of maternal education varies with the gender and race of the child, and with the cognitive ability of the mother.
We show that maternal education has positive impacts both on cognitive skills and behavioral problems of children, but the latter are more sustained than the former. This is perhaps because behavior is more malleable than cognition. Especially among whites, there is considerable heterogeneity in these impacts, which are larger for girls, and for mothers with higher cognition.
More educated mothers are more likely to work and work for longer hours, especially among blacks. This is true independently of the child being in its infancy, childhood, and adolescence. Nevertheless, there is no evidence that more educated mothers do less breastfeeding, spend much less time reading to their children, or even taking them on outings. This is important because some studies suggest that maternal employment may be detrimental for child outcomes if it leads to reduced (quality) time with children.
Due to the nature of the data, this paper focuses on the effect of maternal, but not paternal, schooling. Due to assortative mating, part of the effects we find may be driven by the father's schooling through a mating effect. However, unless the effect of partner's schooling is incredibly large, assortative mating cannot fully explain our main results, as suggested in some of the literature.Download full version