Identifying the Consequences of Teenage Childbearing

Using miscarriage as an instrument for birth suffers from a censoring problem created by the fact that women wanting the pregnancy to end in an abortion frequently pre-empt an "assigned" miscarriage. I describe evidence that this problem is severe, creates a significant bias in IV estimates, and develop non-parametric estimators to correct for censoring. While simple IV estimates previously suggested teenage childbearing had little effect on high school completion or labor market outcomes later in life, non-parametric IV estimates indicate significant costs of teenage childbearing, particularly for high school aged girls