Page 1 of 1

Forecast values are very large

Posted: Wed May 16, 2012 8:34 am
by AnnH
I am running Eviews 7. I have built a model and the program runs fine with no errors and I get forecast values for all of the variable for which I have appended equations to the model.
ModelA.SOLVE(M=5000, C=.001, D=d, S=d, N=t, O=m)

The problem is that for about 1/3 to 1/2 of the variables that eviews finds forecast values for, the forecast values are HUGE. I have checked my exogenous variables and I do not see any issues with the values we have chosen for those factors, which makes me think as Eviews is solving something is going out of control.

I'm not sure where to start checking to see what I've done that may be causing this. Can anyone point me in the right direction?

Thanks!

Re: Forecast values are very large

Posted: Wed May 16, 2012 8:43 am
by EViews Gareth
The most likely cause is that your equations are simply explosive. Are the equations in question independent from the other equations? If so, try using the equation forecast rather than the model solution and check that the numbers are the same.

By the way, I'm not sure what the "o=m" option is doing.

Re: Forecast values are very large

Posted: Fri May 18, 2012 5:51 am
by jlbrillet
You can also have access to the sequence of values the variables take, using before solving the model :

model solution > diagnosis > track selected variables > names of the tracked elements.

The values for the tracked elements will be obtained later in

View > Trace output.

This is explained in paragraph 7.1.6.2.1 of my book (cf message of April 13).

Jean Louis Brillet.

Also, I do not know about o=m (EViews Gareth too, which says something). Use rather o=b or o=n.

Re: Forecast values are very large

Posted: Fri May 25, 2012 7:07 am
by AnnH
Thank you!
I think there is a problem with the equations so I am currently working on re-estimating them. I also am not sure what the option o=m means (I am trying to rework some old programs that were written by someone else - that can partly explain my cluelessness.....:-)

Re: Forecast values are very large

Posted: Tue May 29, 2012 2:24 pm
by AnnH
I believe I have my equations fixed so they are not explosive, however, now my model won't even solve, I get the error message "Solver has stalled: starting values may be too far from a solution".
Any ideas on where I can start to try and fix it?
Thanks so much for your help!

Re: Forecast values are very large

Posted: Tue May 29, 2012 2:30 pm
by EViews Gareth
You could try changing the solver method (Solver tab of the model - change the Solution Algorithm combo box to something else), or, as the message indicates, change the starting values.