Re: [Gretl-users] Standard error for confidence intervals
On Viernes, 16 de Octubre de 2009 12:35:09 Francisco Sosa escribió:
>
> I understand the concept, but:
>
> I have and AR(1), Prais-Winsten transformation, and I have seen that the
> error should be E(t)=rho*E(t-1)+U(t) (Please let me know if I am wrong).
I
> assume that U(t) is the standard error of the regression, and
E(t) is the
> error in T, for example:
But, why are you using Prais-Winsten to estimate an AR(1)? or am I
loosing
something?
You can estimate an AR(1) directly by OLS or using the arima command
of
gretl.
Is your model Y_t=const+\phi Y_{t-1}+\eps_t or is it
Y_t=alpha+beta X_t + E_t
with E_t=rho*E_(t-1)+U_t?
This last is the model that the Prais-Winsten estimator is for.
Actually my model is:
Y'_t=alpha+beta X'_t + E_t
When Y'_t= Y_t - rho*Y_t-1 and X'_t= X_t - rho*X_t-1.
And E_t=rho*E_(t-1)+U_t is the standard formula for one AR(1).
Prais-Winsten is an improvement of the algorithm for estimating regressions
(AR(1)) in the present of autocorrelated errors, I guess I have to do
something else to calculate the E_t+h, I need the variance to calculate the
confidence interval.
Many thanks