EcoSta 2019: Start Registration
View Submission - EcoSta2019
Title: Unconditional quantile regression with endogenous regressors: A simulation study Authors:  Yuya Katafuchi - Research Institute for Humanity and Nature (Japan) [presenting]
Abstract: Quantile regressions enable researchers to investigate the distributional effect of a target variable conditional on control variables, which complement the average effect identified by the ordinary least squares. Among them, the unconditional quantile regression (UQR) is becoming popular for empirical researchers who would like to measure the effect of a target variable on the unconditional quantile of the outcome distribution. In particular, to consistently estimate the causal effect of a variable of interest, the existing literature renders two approaches to deal with the endogeneity within the context of the UQR: the instrumental variable approach and the control function approach. There is, however, little research conducting a comparison study between these two approaches through simulation experiments. This issue is examined through Monte Carlo simulations. The cause of weak instruments with the UQR estimation is also investigated.