Hi Matthieu (again),
Since this is a separate topic, I thought I should open a separate issue for it.
The documentation of tsDyn::rank.test() states that it does not allow for seasonal or exogenous regressors (while urca::ca-jo()does). Are you referring here to the dummy variables that the urca-package allows? When I tried out a model with exogenous regressors (and I guess these could technically also be seasonal dummies) using tsDyn::VECM() and then applied the tsDyn::rank.test() to the output, I did get some results. Should I assume that these results are wrong?
As a follow-up, is the reason for not allowing for exogenous variables (including seasonals) a conceptual one? If my understanding is correct, the presence of stochastic exogenous regressors in the VECM can alter the asymptotic distribution of the rank test statistic. Is there a similar issue with seasonal dummies and this is why your helpfile says your function cannot be used in this case?
Thanks again!
Hi Matthieu (again),
Since this is a separate topic, I thought I should open a separate issue for it.
The documentation of
tsDyn::rank.test()states that it does not allow for seasonal or exogenous regressors (whileurca::ca-jo()does). Are you referring here to the dummy variables that theurca-package allows? When I tried out a model with exogenous regressors (and I guess these could technically also be seasonal dummies) usingtsDyn::VECM()and then applied thetsDyn::rank.test()to the output, I did get some results. Should I assume that these results are wrong?As a follow-up, is the reason for not allowing for exogenous variables (including seasonals) a conceptual one? If my understanding is correct, the presence of stochastic exogenous regressors in the VECM can alter the asymptotic distribution of the rank test statistic. Is there a similar issue with seasonal dummies and this is why your helpfile says your function cannot be used in this case?
Thanks again!