VAR forecasting evaluation

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

Mihai
Posts: 6
Joined: Sat Apr 18, 2015 9:13 am

VAR forecasting evaluation

Postby Mihai » Sat Apr 18, 2015 9:36 am

Hello there and sorry if this question has been asked before. I am using EViews 7 and trying to make some forecasts with VAR models.
My question would be if it's possible to view the performance indicators for the forecasts such as RMSE, MAE, MAPE, THEIL IC in order to help me choose the most efficient model ? I know that this is available for the Forecasts from Equations but I can't find them anywhere for the VARs.

Thanks a lot,
Mihai

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 13604
Joined: Tue Sep 16, 2008 5:38 pm

Re: VAR forecasting evaluation

Postby EViews Gareth » Sat Apr 18, 2015 10:20 am

Not in EViews 7. You'd have to upgrade to EViews 9. Or calculate them manually.

Mihai
Posts: 6
Joined: Sat Apr 18, 2015 9:13 am

Re: VAR forecasting evaluation

Postby Mihai » Sat Apr 18, 2015 10:27 am

Thanks for the answer, so I guess it's the same for all of the previous EViews versions, earlier than EViews 9 correct ? :(
Or is there any other alternative way of evaluating the VAR forecasts with EViews 7 ?

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 13604
Joined: Tue Sep 16, 2008 5:38 pm

Re: VAR forecasting evaluation

Postby EViews Gareth » Sat Apr 18, 2015 10:46 am

You can try the VAR Forecast Add-in. I'm not sure if that works with 7 or not.

Mihai
Posts: 6
Joined: Sat Apr 18, 2015 9:13 am

Re: VAR forecasting evaluation

Postby Mihai » Sun Apr 19, 2015 8:47 am

Hi again,

I was thinking of trying something else but I'm not sure it would really be correct. Do you guys think that it would be ok to copy the desired equation from the VAR model and to paste it into a new generated simple eq, and to try to make the forecast from there ? Like this I would be able to get the needed forecast evaluation although I'm not so sure that the procedure is ok.

Thanks !

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 13604
Joined: Tue Sep 16, 2008 5:38 pm

Re: VAR forecasting evaluation

Postby EViews Gareth » Sun Apr 19, 2015 10:27 am

Should be ok

Mihai
Posts: 6
Joined: Sat Apr 18, 2015 9:13 am

Re: VAR forecasting evaluation

Postby Mihai » Sun Apr 19, 2015 10:47 am

Tried it and compared the forecasts and somehow they're not the same. Weird... anyways I'll work on this, Thanks !

startz
Non-normality and collinearity are NOT problems!
Posts: 3798
Joined: Wed Sep 17, 2008 2:25 pm

Re: VAR forecasting evaluation

Postby startz » Sun Apr 19, 2015 11:19 am

When you copy the equation into a simple eq, how do you get the forecasts of the RHS variables? If you're going more than period out, that might make a difference to the forecasts.

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 13604
Joined: Tue Sep 16, 2008 5:38 pm

Re: VAR forecasting evaluation

Postby EViews Gareth » Sun Apr 19, 2015 11:25 am

True - it only works if you're doing a static forecast

Mihai
Posts: 6
Joined: Sat Apr 18, 2015 9:13 am

Re: VAR forecasting evaluation

Postby Mihai » Mon Apr 20, 2015 1:10 pm

So what I tried was creating a system whith the variables from my estimated VAR, going to Proc - Make system - Order by value, and I copied the equation for my desired variable which I want to forecast into a new simple equation. After that I performed a dynamic forecast inside the sample, and compared it to the same forecast for the same time horizont performed with the VAR, also inside the sample. Probably the procedure is not correct but it was the only thing I could think of in order to be able to see the evaluation indicators.

startz
Non-normality and collinearity are NOT problems!
Posts: 3798
Joined: Wed Sep 17, 2008 2:25 pm

Re: VAR forecasting evaluation

Postby startz » Mon Apr 20, 2015 1:12 pm

Could you forecast from the VAR and then use

Code: Select all

forceval
to get the desired statistics?

Mihai
Posts: 6
Joined: Sat Apr 18, 2015 9:13 am

Re: VAR forecasting evaluation

Postby Mihai » Tue Apr 21, 2015 10:52 am

Hi again and thank you so much for your answer ! Unfortunately I'm not really a pro in using EViews or other statistical software and I don't really know how to use the code you suggested me. Where should I enter it or how to get there. I would be really grateful if you could point me the steps I have to follow.

Thanks a lot and sorry for my clumsiness :oops:

startz
Non-normality and collinearity are NOT problems!
Posts: 3798
Joined: Wed Sep 17, 2008 2:25 pm

Re: VAR forecasting evaluation

Postby startz » Tue Apr 21, 2015 10:55 am

I've never used this command either. Look in the Help system for guidance.


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 2 guests