Page 1 of 1

How to make forecast after ARMA estimation?

Posted: Fri Apr 27, 2018 4:44 am
by ecofine
Dear Folks,

I am using eviews 10 version.

I have tried to estimate the forecast after ARMA model estimation.

My question is how to get the forecast estimation out of sample period, which will be a kind of real future forecast.

But eviews told that forecast is not possible out of the sample period.

For example, my sample period is between january 2010 and january 2017.

After ARMA estimation, I want to do a forecast between the period of Feb. 2017 and Feb. 2018.

Your kind suggestions or tips would be very much appreciated.

Thanks.

HIS

Re: How to make forecast after ARMA estimation?

Posted: Fri Apr 27, 2018 6:09 am
by EViews Gareth
What was the error message?

Re: How to make forecast after ARMA estimation?

Posted: Wed May 02, 2018 12:43 am
by ecofine
Dear Gareth,

Thank you very much for your kind reply.
I got his kind of error message:
Error in Sample: Attempt to set sample outside of workfile
range.

Structure/Resize (from \WORKFILE\PROC menu) can expand
the workfile range.

Actually, I tried to find the function of expand but nothing found.

Could you kindly introduce me how to expand the sample into out of sample period and get the forecast graph over the out of sample period?

Best,

Insang

Re: How to make forecast after ARMA estimation?

Posted: Wed May 02, 2018 7:12 am
by EViews Gareth
From your workfile click on the Proc button then select Structure/Resize.

Re: How to make forecast after ARMA estimation?

Posted: Fri May 04, 2018 3:21 am
by ecofine
Thank you for your kind reply.
I tired to revise the sample period covering out of the sample.
Then, I could get the graph for the extended period.

But my problem is that the prediction for the extended period was not changing at all. I simply extended the period without doing anything and any new data included.

Then, it simply comes that predicted Y variable comes as a fixed line after all without any changes; this means that there were not anything done in the forecast.

How can I get a proper forecast for the future period incorporating previously estimated coefficient results?

My questions is how to construct the predicted Y values in the extended period?

Usually, we could give a series of possible values of Y for the extended period. Is there any way automatically giving such a series of Y values for the extended periods?

My above question may be wrong.
Your proper suggestions would be so great.

Thank you.

HIS

Re: How to make forecast after ARMA estimation?

Posted: Fri May 04, 2018 3:27 am
by ecofine
Dear Gareth,

Could you check out wf1 attached?

Thank you.

HIS

Re: How to make forecast after ARMA estimation?

Posted: Fri May 04, 2018 8:15 am
by EViews Gareth
If you have no exogenous regressors, an ARMA forecast will become straight.

Re: How to make forecast after ARMA estimation?

Posted: Fri May 04, 2018 5:38 pm
by ecofine
Thank you Gareth.
I thought that that was my problem.

How can I make the exogenous data for the extended period using the estimated results is the next thing to do.
Well, I have no good data to replace.

Could you have some alternative suggestion for this case? What do people usually follow in order to make exogenous data of Y hat for the extended period?

Thank you very much.

HIS