How children affect women's labor market outcomes: estimates from using miscarriage as a natural experiment

In this study, I empirically estimate the impact of children on women's labor market outcomes (such as work status, work hours and earnings). The identification in a female labor supply model where the fertility is endogenous comes from the assumption that a miscarriage – spontaneous loss of the pregnancy – occurs mostly randomly. Medical research in general provides supportive evidence for this assumption. One advantage of using the occurrence of a miscarriage to estimate the impact of children is that it allows one to estimate the impact of a first child versus no children at all, which is not possible when using other natural experiments such as twin births or the gender combination of the first two children. An instrumental variable based on the outcome of the first pregnancy is constructed and is used to estimate the coefficient of the endogenous fertility variable. The result shows that in general children have a modest negative impact on women's labor supply. It also shows that the IV estimates tend to be smaller in scale than the OLS estimates using the same data, which suggests it is indeed important to address the problem of endogenous fertility when estimating a female labor supply model.