Bootstrap In Unit Root Quantile Regression

The main content of this paper is the unit root test based on quantile regression.In the normal quantile regression unit root test,considering the model:y,=?yt-1+ut,t=1,…,n,where u,is an independent random variable from some distribution.Then,under the null hypothesis,?= 1,the asymptotic distribution of statistics is related to the distribution of the redundant parameter ut,so the distribution of ut should be estimated.In this paper,we use a bootstrap resampling process to avoid the estimation of ut and prove that the asymptotic distribution of statistics converges to the same distribution.We extend the result of Moreno and Romo(2000)[34]under least absolute estimation.