Coef uncertainty in SE calc for equations with ARMA

[irulishka]

Hello!
As I understand, there are standard formulas to calculate a forecast standard error and they are in EViews Users Guide. But as I can see from results, when I calculate SE for forecast by equations with ARMA, it doesn’t converge with given formulas. Could you explain, which formulas does EViews use to calculate it?

[EViews Glenn]

The simple formula we provide in the manual is for the non time-series case. The time-series form has to account for the uncertainty generated by the AR and MA terms. It's a complicated recursive formula which is why we don't write it out in the manual...

[shawnmu]

Glenn,

I have a simple linear trend with AR2 (or, precisely lagged dependent variable) model as follows:
yt=a0+a1*y(t-1)+a2*y(t-2)+bt+et. Can you please explain how the eviews come up with a three-step-ahead forecast s.e. as an illustation? I assume it is not based on the delta method. Many thanks.