Estimating the Impact of Sterilization under a Government Campaign Using Contaminated Treatment Data

Abstract

We evaluate the effects of sterilizations resulting from a national campaign in Peru in the 1990s. We propose a modified inverse probability weighting estimator that accounts for data contamination that arises because while we observe sterilization status, we do not know which sterilizations occurred as a result of the campaign. We estimate that 70,000-90,000 women were sterilized because of the campaign and experienced an average decrease in completed fertility of approximately 0.8 children per woman. We find no effect of campaign-induced sterilizations on women's labor supply but find positive effects on children's health outcomes associated with better nutrition.