EViews.com

Hello everybody,

I would like to estimate a VAR-Model comprising of two equations.
e.g.:
Var_1=C(1)+C(2)*Var_1(-1)+C(3)*Var_1(-5)+C(4)*Var_2(-1)
Var_2=C(5)+C(6)*Var_2(-1)+C(7)*Var_2(-3)+C(8)*Var_2(-4)+C(9)*Var_1(-1)

When I use the VAR-object I can only use the same amount of lags for each variable. When I use a system-object I can program the equations as described above. Thus, I would like to know, whether I can use a system (based on SUR or FIML estimation methods) to estimate a VAR-model or are there any limitations? After estimating the VAR model I will have to do some forecasting, which should be possible when I transform the system into a model.

Thank you very much and best regards.

Chris

You\'ll be able to estimate the VAR under the System object, and as you point out, you\'ll be able to forecast by using the Make Model proc. You will lose some of the VAR specific diagnostics though, notably Impulse Responses and things like Granger Causality.