Forecasting a series without regressing

For questions of any nature based on EViews 5 or earlier versions of EViews.

Moderators: EViews Gareth, EViews Moderator

thomaskm
Posts: 14
Joined: Thu May 24, 2012 4:46 am

Forecasting a series without regressing

Postby thomaskm » Thu May 24, 2012 4:53 am

Dear all,

I'm on Eviews 5 and i have a problem with forecasting. I want to see the Mean Root Squared Error for two series, namely
a) Overnight Interest rates
b) Taylor-rule with prespecified coëfficients.
I did some googling and it seems as if everytime you want to do a forecast, Eviews wants to estimate the coefficients first. I already know the coefficients i want to use, and hence, want to skip this feature. So basically: i want to do a forecast without regressing first. Can anyone help me with this?

Thanks in advance,

Kind regards,

Thomas Kooiman.

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 11372
Joined: Tue Sep 16, 2008 5:38 pm

Re: Forecasting a series without regressing

Postby EViews Gareth » Thu May 24, 2012 7:52 am

You'll have to do it manually.

Code: Select all

series y_f = coef(1)+coef(2)*x+coef(3)*z

etc..
Follow us on Twitter @IHSEViews

thomaskm
Posts: 14
Joined: Thu May 24, 2012 4:46 am

Re: Forecasting a series without regressing

Postby thomaskm » Fri May 25, 2012 2:15 am

Thank you kindly for your response. If i understand what you're saying i need to construct a serperate series that contains all the coefficients and variables that are in the Taylor rule. After that I do a regression with command

ls overnight_int overnight_int_taylor

and proceed with a forecast..

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 11372
Joined: Tue Sep 16, 2008 5:38 pm

Forecasting a series without regressing

Postby EViews Gareth » Fri May 25, 2012 6:33 am

No, I'm just saying you construct the forecast directly by hand.
Follow us on Twitter @IHSEViews

thomaskm
Posts: 14
Joined: Thu May 24, 2012 4:46 am

Re: Forecasting a series without regressing

Postby thomaskm » Wed May 30, 2012 2:25 am

Could you please clarify how exactly this is done? I appreciate your answers thusfar, but they haven't enabled me to figure out how to do this.
The following screenshot depicts the two series for which i want to know the root mean squared error.

Could you tell me the specific code on how to do this in this instance? If I know that i can continue to do it for the rest of my dataset.
Thanks a bunch in advance!

Image

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 11372
Joined: Tue Sep 16, 2008 5:38 pm

Forecasting a series without regressing

Postby EViews Gareth » Wed May 30, 2012 7:04 am

I'm confused. What are you trying to do? Where do those two series fit into the forecast you're trying to do?
Follow us on Twitter @IHSEViews

thomaskm
Posts: 14
Joined: Thu May 24, 2012 4:46 am

Re: Forecasting a series without regressing

Postby thomaskm » Wed May 30, 2012 7:13 am

Yes, I think i've been unclear. My apologies.
I think my use of 'forecasting' wasn't right, i don't think i want to forecast anything.
All i want to do is the following: calculate the RMSE for the two series depicted in the graph. So: for overnight_int and overnight_int_taylor, for the entire period (1991Q1-1998Q4). That's it. Hope this clears up what i'm asking.

startz
Non-normality and collinearity are NOT problems!
Posts: 3173
Joined: Wed Sep 17, 2008 2:25 pm

Re: Forecasting a series without regressing

Postby startz » Wed May 30, 2012 7:18 am

Subtract one from the other and look at the standard deviation?


Return to “EViews 5 and Earlier”

Who is online

Users browsing this forum: No registered users and 1 guest