A Monte Carlo comparison of parametric and nonparametric quantile regressions

Abstract

This study compares parametric and nonparametric quantile regression methods using Monte Carlo simulations. Simulation results indicate that the nonparametric quantile regression approach is more appropriate, particularly when the underlying model is nonlinear or the error term follows a non-normal distribution.

Acknowledgement

The authors wish to acknowledge the helpful comments of Qi Li.

Notes

¹ In this article, a local constant kernel estimator is employed while Yu and Jones (1998) used a local linear kernel estimator.

² The standard normal density function is used as a kernel function.