addinit so no residual when evaluate at actuals

For technical support, tips and tricks, suggestions, or any other information regarding the EViews model object.

Moderators: EViews Gareth, EViews Moderator

Elderfield.A
Posts: 78
Joined: Thu Jul 11, 2019 10:47 pm

addinit so no residual when evaluate at actuals

Postby Elderfield.A » Wed Dec 09, 2020 6:42 pm

Hi,

I've come across a bit of a peculiar problem, which I'll do my best to describe. In coming up with a baseline for my model, a few variables are excluded and overridden, some add-factors are adjusted etc. I then want to run scenarios off this baseline, but in the scenarios I want the full functionality of my model (no equations excluded).

To do this, and please advise if there is a better way to do this, I go through a number of steps:

1) create profiles of my baseline forecast with no suffix over the whole workfile sample. So for example, x_1 becomes x in the workfile, and instead of having NA in forecast period, it has been filled with my baseline forecast :

Code: Select all

smpl @all
model.scenario "Baseline"
model.makegroup blvars @endog
for !i =1 to blvars.@count
   %1 = blvars.@seriesname(!i)
   !a = @length(%1)-2
   %2 = @left(%1,!a)
   series {%2} = {%1}   
next


2) I then create a new scenario and initialise so that the equation has no residual when evalulated at actives

Code: Select all

   smpl 2020q4 2031q2
   model.scenario(n,a = "_3") "scenario name"
   model.addinit(v = n) @all



3) Solve model with new add factors


Now - this has worked (last quarter, for example it worked fine). However, I have updated my baseline and now this procedure doesn't work. I don't get any errors, but the models solution is not equal to "actuals".

I appreciate it is hard for Eviews to give me an answer without seeing my workfile, unfortunately I can't share it. However, is there anything generally speaking that would cause this not to work? Like things I could check and troubleshoot? Are there some type of equations that could cause error? We've changed a couple of equations since the last baseline, but nothing that I would deem as unconventional.

Thanks

Adam

EViews Matt
EViews Developer
Posts: 560
Joined: Thu Apr 25, 2013 7:48 pm

Re: addinit so no residual when evaluate at actuals

Postby EViews Matt » Mon Dec 14, 2020 5:34 pm

Hello,

You mention that you've made changes to some of the model's equations. Depending on how you did that, it's possible that the any add factors for those questions were dropped (the series will still exist in the workfile, but won't be used). I'd started by double checking that any modified equation that should have an add factor are still flagged as such.

Elderfield.A
Posts: 78
Joined: Thu Jul 11, 2019 10:47 pm

Re: addinit so no residual when evaluate at actuals

Postby Elderfield.A » Mon Dec 14, 2020 6:23 pm

Thanks for the reply Matt.

I thought somethings similar, I used the command

Code: Select all

 model.addassign(n) @all
 model.addinit(v = n) @all


Which I was hoping would create zero filled add-factors for all equations in the model, then addinit setting so there are no residuals at actuals.

Just as an aside, does my approach of "creating actuals" from my forecast - seem sensible, or is there a better way to do this?

I am using Eviews 10+ Feb 5 2020 build, by the way.

EViews Matt
EViews Developer
Posts: 560
Joined: Thu Apr 25, 2013 7:48 pm

Re: addinit so no residual when evaluate at actuals

Postby EViews Matt » Tue Dec 15, 2020 12:28 pm

That's a reasonable way to create your actuals, since you're utilizing overrides that cannot be applied in the standard actuals or baseline scenarios.

Without access to your workfile, I'm at a loss to explain the discrepancy between your model results and your updated actuals. Are the model solutions for all the variables different from the actuals? If not, you should be able to identify the "first" variable/equation with a discrepancy by cross-referencing with the variable evaluation ordering shown in the model block structure view. That at least may help direct you towards equations that may have had their settings changed in unexpected ways, producing these unexpected results.

Elderfield.A
Posts: 78
Joined: Thu Jul 11, 2019 10:47 pm

Re: addinit so no residual when evaluate at actuals

Postby Elderfield.A » Tue Dec 15, 2020 2:48 pm

Thanks for the suggestion Matt - I haven't checked if all variables are different, just some key ones. I'll try and find the equation where things fall over.

And, yep totally understand it is hard to offer advice without seeing the workfile. But thanks for helping where you can :)

Elderfield.A
Posts: 78
Joined: Thu Jul 11, 2019 10:47 pm

Re: addinit so no residual when evaluate at actuals

Postby Elderfield.A » Wed Dec 16, 2020 4:06 pm

Hi Matt,

Just closing the loop on this. Turns out your initial hunch was correct, there were some add-factors that were dropped at some stage. Using the addassign before the addinit command does actually work - must of bungled it when I first was experimenting.

Cheers,

Adam

EViews Matt
EViews Developer
Posts: 560
Joined: Thu Apr 25, 2013 7:48 pm

Re: addinit so no residual when evaluate at actuals

Postby EViews Matt » Thu Dec 17, 2020 10:30 am

Good to know. I'm glad you were able to sort out the problem.


Return to “Models”

Who is online

Users browsing this forum: No registered users and 10 guests