EFFECT OF PRELIMINARY UNIT ROOT TESTS
ON PREDICTOR OF MODEL

Monchaya Chiangpradit, Sa-Aat Niwitpong
Department of Mathematics
Faculty of Science
Mahasarakam University, 44150, THAILAND
Department of Applied Statistics
Faculty of Applied Science
King Mongkut's University of Technology North Bangkok
Bangkok, 10800, THAILAND

Abstract. This paper presents a method to estimate the predictor and the scaled prediction mean squares error of an AR(p) model after preliminary unit root tests by using Augmented Dickey-Fuller, Phillips Perron, KPSS and DF-GLS unit root tests. Monte Carlo simulation results are given to compare the relative efficiencies of one-step-ahead prediction using the scaled prediction mean squares error for an AR(2) model with a linear trend. All preliminary unit root tests considered here perform well to improve the predictors from trending AR(2) process when the root near unit root. Moreover, the preliminary unit root tests of KPSS and DF-GLS are slightly superior to other unit root tests.

Received: May 16, 2012

AMS Subject Classification: 62M10, 62M20

Key Words and Phrases: preliminary unit root tests, scaled prediction mean square error, AR(p) Model

Download paper from here.