Normal Tests for a Unit Root in the Autoregressive Time Series Model

Abstract

Test for a unit root has been widely used to investigate the dynamic nature of economic time series. However, it is known to be rather cumbersome, since the asymptotic distributions of the conventional test statistics are non-normal and we have to resort to the special tables for critical values. Recently. Choi derived the test statistic whose asymptotic distribution is normal. This paper proposes similar test statistics and examine their empirical size and power through the Monte Carlo experiment, and compare them with those of the conventional Dickey-Fuller test statistics.