SARIMA INTEGRATED FACTORS

[rafrods]

Hello. Its possible i have an SARIMA model like (1,1,1)(1,0,1)12, or (1,0,1)(1,1,1)12? In other words, have an d = 1, and D = 0, or d = 0, and D = 1?

How the integrated factor of seasonal part is represented in automatic arima in eviews, considering the appointed choosen models only apears like (p,q)(P,Q)?

[EViews Gareth]

https://eviews.com/help/helpintro.html# ... 23ww203177

[rafrods]

https://eviews.com/help/helpintro.html#page/content%2Ftimeser-Estimating_ARIMA_and_ARFIMA_Models_in_EViews.html%23ww203177

Ok. But how i can know if the automatic ARIMA is indicating to me an model with the D = 1 or D = 0? Because the "d" is easy to see considering the dependent variable, as appointed as selected dependent variable in example (https://eviews.com/help/helpintro.html# ... sting.html).

[EViews Gareth]

Automatic arima does not consider seasonal differencing.

[thorsten]

You write: "Automatic arima does not consider seasonal differencing."

Instead of using our variable y, would it be possible to just create a LHS variable that is d12y (that is e.g., series d12y=y-y(-12) ), and then use our seasonally differenced d12y in this automatic arima routine in EViews. Would this choose the best SARIMA (given that the true model says that we should have a seasonally differenced variable d12y on the LHS)?

[EViews Gareth]

It would choose the best ARIMA orders for the given seasonal difference of one.

[thorsten]

OK, so then this algorithm would not consider that a single ar(4) could be better than ar(1), or ar(1) ar(2) ar(3) ar(4)?