Examining the power of stochastic unit root tests without assuming independence in the error processes of the underlying time series

Abstract

Many studies have examined the power of Stochastic Unit Root (STUR) tests. However, these studies assume that the two error processes of the underlying time series are independent. In this study, we undertake a Monte Carlo study on the power of STUR tests without the condition of independence among the error processes. The results show that the correlation between the two error processes may profoundly impact the power of STUR tests. Given the extensive use of STUR tests as both a diagnostic tool and a tool of analysis in economics and finance, this result therefore has very important implications for both theory and practice.