I'm trying to estimate the standard error of the intercept in ARIMAX(1,0,2) model using R.
Since the reported 'intercept' in regression output is actually some kind of mean value I applied this formula to get the 'real' intercept:
Constant = Mean*(1-AR coefficient)
The problem is that I'm not sure how to calculate the standard error of this constant. I was thinking to use propagation of uncertainty. Does anyone know how the error propagation formula would look like in this case? Or if there is any other way to calculate the standard error of the constant in ARIMAX model?
Any help would be greatly appreciated.