Estimating in VAR and single equation using OLS

[tinfah]

Dear Gareth,

Are there any difference in Eviews when we use a estimate var and estimate equaton for 2 variables. For the estimate equation we can use OLS and specify it like a VAR specifcation with the same lags. If there are, how can we use the estimate equation like a estimate var. I am keen to explore other estimators in estimate equation that is not available in estimate var like using a gmm.
Thanks

[EViews Gareth]

Why not try it and see? :D

[tinfah]

Thanks. Not exactly the same but close. I guess the differece is because the predetermined variables in a VAR formulation are solved iteratively (ie taken from the previous period) and fed into the next period?

[EViews Gareth]

They should be identical.

[tinfah]

Many thanks for your help. Yes, for the second equation, they are identical. For the first equation, very marginal difference.

[startz]

Unless by "marginal difference" you mean something like the fourth decimal place, something else has changed. I'd check that the VAR and OLS equations are all using the same sample.

[tinfah]

Dear Startz

If I run a 2 variable VAR, with 2 lags, one of the 2 equations is the same as if I run it in OLS in reduced form. But, the other equation has difference more than marginal, by your definition of 4 decimal place. If I run a 3 variable VAR, one of the 3 equations is identical, but the results for the other 2 has more than marginal difference. Perhaps, something has changed in the data set. Thanks.

[startz]

You might want to post your estimated results, or perhaps your workfile including the estimated equations.