Selection of Unit Root Test on the Basis of Time Series Length and Value of AR(1) Parameter

Abstract

An important task in econometric modelling is to determinate the integration order of analysed time series through unit root tests. Statistical theory offers a wide range of tests where the most common are Dickey-Fuller tests, Phillips-Perron test, KPSS test, and their modifications ADF-GLS test and Ng-Perron test. The choice of an appropriate one depends primarily on a subjective judgement of the analyst. If we wish to avoid the subjective choice, we need to find an objective criterion that clearly defines which test is the most suitable for specific types of time series. The goal of the article is to answer this question by a simulation study and to provide the recommendations which test is possible to use. The conclusions will be applicable for time series of lengths T = 25, ..., 500 and positive values of the autoregressive parameter AR(1).

Keywords

• Unit root test
• augmented Dickey-Fuller test
• Phillips-Perron test
• KPSS test
• ADF-GLS test
• NGP test
• simulation
• power of test
• time series analysis