DETERMINISTIC SEASONALITY AND VEQCM

[MMG1]

Hi friends!! I am thinking in something and i wanna discuss with you. The think is: Imagine that you have 3 montly series, one of them has
deterministic sesonality. You want to test the possible cointegration between them through Johansen approach, so in eviews in this analysis you have the chance to include exogenous variables. In this part we can include the seasonal dummies that we have created in order to capture the deterministic sesonality but my question is: Where are included this dummies? inside or outside the cointegration relation? i mean I dont really know, where the eviews locate the dummies in this analysis. someone could write the cointegration relation with 11 centered dummies and constant term?

[trubador]

We are interested in the cointegration equation, because it represents the long-term relationship among the variables. Since seasonality is a short-term phenomenon, it would be wise to keep it outside as an exogenous variable. In that case, however, including seasonal dummies as regressors for each equation will cost you too many degrees-of-freedom. If you are lucky enough to have a deterministic seasonality, you can remove it by a simple regression. Otherwise, I suggest you to seasonally adjust your series with a formal procedure (X12 or TRAMO) before using them in a model.