On Viernes, 24 de Julio de 2009 05:10:50 Allin Cottrell escribió:
On Tue, 21 Jul 2009, Ravi Mulani wrote:
> I am using the forecasting feature for ARIMA (using the icon
> view, as opposed to programming). Is there any way for me to
> forecast farther into the future than the lags of independent
> variables? For example, can I forecast GDP 3 quarters into the
> future using a 1-quarter lagged imports variable?
Only if you are able to supply values (possibly hypothetical
values) for the imports variable. Gretl can't make up those
values for you.
You can specify an univariate ARIMA model for imports, then forecast with this
model the periods you need for that variable and put this values as inputs in
the ARIMAX model for GDP. In this case, the point forecasts that gretl
produces are ok, but the forecast intervals are not good. The correct
intervals should have into account the variability in the forecasting of
imports, so they have to be wider than the intervals that gretl produces.
DEPARTAMENTO DE ECONOMÍA APLICADA III (ECONOMETRÍA Y ESTADÍSTICA)
Avda. Lehendakari Aguirre, 83 | 48015 BILBAO
T.: +34 946013732 | F.: +34 946013754