Testing for more than on unit root: the use of SAS® software for a routine to perform the Dickey-Pantula test

Abstract

The determination of a variable’s order of integration is a key issue for those dealing with time series econometrics. Any misidentification of the integration order (or of the number of unit roots) may lead to what has been called spurious regression. Although traditional statistical tests may seem to be significant, regression results have no economic meaning. To solve this problem, from the seventies onwards there were developed several unit root tests, among which the most widely used are the Dickey-Fuller (DF), Augmented Dickey Fuller (ADF) and the Phillips-Perron (PP) non-parametric test. Routines for these tests are available for SAS® users. However, DICKEY e PANTULA (1987) showedthat traditional unit root tests are inadequate when there is a possibility of more than one unit root. This paper presents the elaboration of a SAS routine designed to perform the Dickey-Pantula test in economic applications.