Qualitative variables and their reduction possibility. Application to time series models

In this paper we will study the influence of qualitative variables on the unit root tests for stationarity. For the linear regressions involved the implied assumption is that they are not influenced by such qualitative variables. For this reason, after we have introduced such variables, we check first if we can remove some of them from the model. The considered qualitative variables are according the corresponding coefficient (the intercept, the coefficient of Xt −1 and the coefficient of t ), and on the different groups built tacking into account the characteristics of the time moments.