Learning from Teen Childbearing Experiences of Close Friends: Evidence Using Miscarriages as a Natural Experiment

This study builds on the peer effects in adolescent fertility literature by examining close friends and using miscarriages as a natural experiment to reduce potential confounding due to peer selection, simultaneity of peer influence, and shared environment. Our identifying assumption is that, conditional on a friend’s pregnancy, the occurrence of a friend’s miscarriage is an exogenous negative fertility shock that is uncorrelated with the unobservable factors. The study utilizes fertility and friendship nominations data from the core sample of Add Health. A total of 838 women had a friend with at least one teen pregnancy; of whom 647 had a friend who had a teen birth and 172 had a friend whose pregnancy ended in a miscarriage. A series of balancing and falsification tests support our identifying assumption of no correlation between the woman’s characteristics and the occurrence of a friend’s miscarriage in the pregnant friend subsample. The results suggest the existence of a sizable negative local treatment effect on teen childbearing among best friends. Multinomial logistic regression and IV results with controls for own and friend’s fertility history, extensive set of socio-demographic controls, local peer network structure, and school-level fertility measures suggest that a friend’s teen birth is associated with a 7-11 percentage point reduction in the likelihood of own teen childbearing. There is also evidence that this effect operates through a learning mechanism by changing attitudes toward early childbearing and reducing unwanted pregnancies.