American Review of Mathematics and Statistics

Asymptotic Properties of Local Composite Quantile Regression Estimation for Time-Varying Diffusion Model
Ji-xia Wang, Yuewen Xiao

Abstract
Based on discretely observed samples, this paper proposes local linear composite quantile regression estimation for time-dependent drift parameter of diffusion models. We verify the asymptotic bias, asymptotic variance and asymptotic normality of the local estimation proposed. The asymptotic relative efficiency of the local estimation with respect to local least squares estimation is discussed. The results show that the estimation proposed can be more efficient than the local least squares estimation for many commonly seen error distributions

Full Text: PDF     DOI: 10.15640/arms.v4n2a4