EcoSta 2019: Start Registration
View Submission - EcoSta2019
Title: Robust estimation of quantile treatment effects under unconfoundedness Authors:  Yu-Chang Chen - University of California, San Diego (United States) [presenting]
Abstract: Quantiles are important inputs to several inequality measures, and inspecting quantile treatment effects (QTE) is a useful way to characterize effect heterogeneity. It has been previously established the identification of QTE under unconfoundedness and provided an inverse probability weighting (IPW) estimator. Although the IPW estimator is semiparametric efficient, several simulation studies suggest that IPW estimators are sensitive to model misspecification and vulnerable to lack of overlap. Consequently, the procedure fails to balance the treatment and control group, leading to substantial bias. To address this issue, we exploit the recent advance in balancing techniques to achieve a robust estimation of QTE. Specifically, we discuss the potential application of the entropy balancing methodology to the QTE estimation problem. Simulations show that the proposed method has substantially less bias compared to the standard IPW estimators. Although the theoretical property of the proposed estimator is still not fully known, recent literature has shown that entropy balancing method has a connection to penalized regression. Therefore, we suspect that the proposed method can have a link to penalized regressions.