Maternal attributes such as education and age can impact early life health either directly or indirectly through the choice of familial inputs or endowments. For example, a mother’s education may affect her knowledge regarding the health impacts of maternal smoking. However, in the presence of assortative mating, her education may influence the education of the mate she chooses.
There is a recent growing interest in the impact of maternal education within economics. This is in part due to an expanding focus on the nonwage effects of human capital. Moreover, maternal education is one of the strongest predictors of infant health. Based on USA data, an extra year of schooling reduces the rate of low birth weight by 10%. These effects are surprisingly linear, implying that the effect of a year of high school education is roughly equal to the effect of a year of college education.
Of course, these correlations do not necessarily imply that there is a causal relationship between maternal education and infant health. Omitted variables bias is a concern, particularly because maternal education is positively related to other attributes such as family background that might improve infant health.
The recent economics literature has made great strides in identifying the causal effect of maternal education. Two of the more frequently exploited quasi-experiments are the construction of new schools and the expansion of compulsory schooling. In the USA, the expansion of higher education through the building of new universities and colleges between 1940 and 1990 led to reductions in the rates of prematurity and low birth weight. Outside of the USA, the construction of new schools in areas without schools has resulted in similar improvements in infant health. When interpreting these estimates, however, one should think about these two settings as possibly identifying different effects of education in the case that there are nonlinear effects of maternal education.
Compulsory schooling reforms in the twentieth century led cohorts born close to one another to have different educational requirements. These compulsory schooling laws dictate when individuals can legally drop out of school. In countries where many individuals drop out at the minimum schooling age and the compulsory schooling laws are enforced, increases in the compulsory schooling age are useful instruments for maternal education. In the USA, the size of the population affected by compulsory schooling reforms is rather small. In contrast, in Britain, at least historically, most individuals drop out of school at the minimum schooling age. Thus, one can use regression discontinuity techniques where contrasts are made between individuals proximate in date of birth who might be otherwise identical except for their level of schooling. The British compulsory reforms generally point to no effects of maternal education on infant health.
The discussed quasi-experiments increase education by extending the end of schooling. Alternatively, an increase in educational attainment could be achieved by reducing the age at school entry. Increases in schooling via augmenting either the beginning or end of schooling could potentially estimate different effects of education. As for the latter, there could be a mechanical effect of extra schooling. Being in school longer may act as an incarceration effect, reducing rates of sexual activity and thus, result in delayed fertility.
This conceptual difference may be an explanation for the difference between the conclusions reached from using school entry policies and other studies. School entry policies impact the start of schooling. Despite their differences in acquired schooling, comparisons of individuals born before and after school entry dates (i.e., the date by which a child must have reached age 5 to enter school) show no evidence of effects of maternal education on infant health.
One difficulty often neglected in this literature is that an instrument for maternal education may affect both fertility and infant health. In the case that there are fertility effects of education, the measured effect of maternal education on infant health suffers from a selection problem.
Similar to that of maternal education, the effects of maternal age could be direct or indirect. Women at either end of the childbearing age spectrum experience worse infant health outcomes. Support of the biological effects of maternal age has been confirmed with animal studies, but maternal age may also influence the choice of prenatal and postnatal inputs. Women who give birth at earlier ages may not have the income or access to adequate medical care that older mothers do. Thus, the fact that maternal inputs vary with maternal age obfuscates the causal effect of maternal age. Specifically, women who give birth at younger ages are of lower SES than women who give birth at older ages. Thus, the adverse impacts of giving birth at a younger age may be overstated in the cross-section although the opposite is true for older ages.
The main empirical evidence of the effects of maternal age comes from sibling-based comparisons. That is, one can compare the outcomes of children born to the same mother. Such an approach effectively controls for fixed differences (e.g., SES which may be fixed) across mothers. However, to the extent that maternal age is correlated with other attributes that vary across a woman’s lifecycle, these sibling contrasts will not capture solely the effect of maternal age. The sibling estimates do confirm the expected direction of biases – the effects of young maternal age are not as adverse as one would expect from correlations and the effects of advanced maternal age are worse than what the cross-sectional correlations imply.