Tests with Stationarity as the Null Hypothesis

Abstract

So far the null hypothesis has been that of nonstationarity due to a unit root, the (usual) alternative being stationarity about a constant or a linear trend. The asymmetry in classical hypothesis testing favours the null hypothesis, which is the unit root in the set-up so far. There are, however, a number of tests that reverse the roles of the hypotheses, so that the null hypothesis is of stationarity (about a constant or a linear trend) and the alternative hypothesis is of a unit root, which is one example of nonstationarity; other forms of nonstationarity have also been considered (see for example, Lo, 1991; Lee and Schmidt, 1996; and see Wright, 1999, for the fractional integrated alternative).