Forecasting from a Regression

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

gtoledo89
Posts: 4
Joined: Mon May 18, 2015 8:48 am

Forecasting from a Regression

Postby gtoledo89 » Mon May 18, 2015 9:02 am

Dear EViews Team,

I use EViews 8

I have been attempting to forecast commodity Prices based off of exchange rate0 movements using this formula:
∆cpt+1=β0+β1∆st+ε
where ∆cpt+1= change in commodity price at time t+1, the first beta is a constant and the second beta is the coefficient and is multiplied by ∆st = the change in currency at time t with ε being the error term.
I have been inputing into the estimate equation field:
log(cp) c log(curr(-1)) log(curr)
and have gotten strange results. I figure that I must be doing something wrong in terms of my inputs into the estimate equation field and was hoping that you may have a suggestion on what can fix the issue.

Thank you
Last edited by gtoledo89 on Mon May 18, 2015 9:14 am, edited 1 time in total.

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

Re: Forecasting from a Regression

Postby startz » Mon May 18, 2015 9:10 am

Dear EViews Team,

I use EViews 8

I have been attempting to forecast commodity Prices based off of exchange rate0 movements using this formula:
∆cpt+1=β0+β1∆st+ε
where ∆cpt+1= change in commodity price at time t+1, the two betas are constants, and ∆st = the change in currency at time t with ε being the error term.
I have been inputing into the estimate equation field:
log(cp) c log(curr(-1)) log(curr)
and have gotten strange results. I figure that I must be doing something wrong in terms of my inputs into the estimate equation field and was hoping that you may have a suggestion on what can fix the issue.

Thank you
Your equation specifies the change in cp, but your EViews command specifies the log. Perhaps you want

Code: Select all

dlog(cp) c log(curr(-1)) log(curr)

gtoledo89
Posts: 4
Joined: Mon May 18, 2015 8:48 am

Re: Forecasting from a Regression

Postby gtoledo89 » Mon May 18, 2015 9:16 am

Would you think then that I should also include the dlog for the currency as well as the equation also specifies the change in the currency?
dlog(cp) c log(curr(-1)) dlog(curr)

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

Re: Forecasting from a Regression

Postby startz » Mon May 18, 2015 9:44 am

You can, but it won't make any difference as you already have current and lagged log currency.

gtoledo89
Posts: 4
Joined: Mon May 18, 2015 8:48 am

Re: Forecasting from a Regression

Postby gtoledo89 » Mon May 18, 2015 9:51 am

I apologize for the incompetence, I have recently begun using the program. I do find a difference when I Input a dlog or just a log for the currency. Would the equation begin written as so make a difference?:
dlog(cp) c log(curr(-1))+dlog(curr)

Also, when I Input the equation, would the C coefficient be equivalent to my beta 0 and would the other coefficient be beta 1?

Thank you for your help, and once again I apologize for the simplicity of the questions.

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

Re: Forecasting from a Regression

Postby startz » Mon May 18, 2015 9:53 am

I apologize for the incompetence, I have recently begun using the program. I do find a difference when I Input a dlog or just a log for the currency. Would the equation begin written as so make a difference?:
dlog(cp) c log(curr(-1))+dlog(curr)
dlog(x) is defined to mean log(x) - log(x(-1))
Also, when I Input the equation, would the C coefficient be equivalent to my beta 0 and would the other coefficient be beta 1?
That's right.

gtoledo89
Posts: 4
Joined: Mon May 18, 2015 8:48 am

Re: Forecasting from a Regression

Postby gtoledo89 » Mon May 18, 2015 10:02 am

Thank you, that seems to have done the trick. The results are more representative of what I thought they should be.


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 2 guests