Re: FW: [Gretl-users] Back forecasting
El Tuesday 24 June 2008 15:03:48 Nieves Sánchez Martínez escribió:
Hello again,
I have a problem with your help task. My model is something like that:
log[y(t)] = c + (alpha+ betaB)log[x(t)] + N(t)
(1 - phi(1)B - phi(2)B^2)(1-Phi(1)B^4)N(t)=a(t)
So, past values of x are needed to do back forecast. Do I need to create a
new model for the new time-serie? But it doesn't have sense... Do you know
how can I do that?
Thank you very much.
Nieves.
Yes, probably you will do the best identifying a new model with the data in
reverse order.
--
Ignacio Diaz-Emparanza
DEPARTAMENTO DE ECONOMÍA APLICADA III (ECONOMETRÍA Y ESTADÍSTICA)
UPV/EHU
Avda. Lehendakari Aguirre, 83 | 48015 BILBAO
T.: +34 946013732 | F.: +34 946013754
www.et.bs.ehu.es