【Brownbag】Nonparametric Identification of Treatment Effect with Misclassified and Endogenous Treatment
2026/11/20
研討會日期 : 2026-11-20
時間 : 12:10
主講人 : Mr. Yan-Yu Chiou (邱晏瑜)
地點 : Conference Room B110
演講者簡介 : Mr. Yan-Yu Chiou is currently a Ph.D. student at Rice University. His research interests lie in Econometric Theory, Machine Learning, and Microeconomics.
演講摘要 : This paper provides new nonparametric identification results for misclassified and endogenous treatment effects. We propose a nonparametric K-means algorithm and establish the asymptotic properties of the algorithm. Instead of finding instrumental variables or repeated treatment measurements to deal with misclassification issues, we use a large number of repeated outcomes to recover the true treatment status of each individual by the nonparametric K-means algorithm. After we recover the treatment status of each individual, we can use a binary instrumental variable to identify the local average treatment effect (LATE) and local quantile treatment effects without the nondifferential error assumption. We show that the treatment estimator is asymptotically equivalent to the infeasible treatment estimator, so the statistical inference of the infeasible treatment estimators can be applied to our estimators. We also propose a test statistic for the nondifferential error assumption. The simulation results show that the nonparametric K-means algorithm performs well in finite samples.
