Hi everyone,
I’m encountering an unusual issue while running a stochastic model in EViews. Out of a system of 62 equations, nearly all variables produce bounds as expected, with higher and lower bounds differing from the mean. However, for one specific variable, "solve solution" returns identical values (or values that closely overlap) for higher, lower, and mean.
This behavior seems strange, especially given that for all other variables the code works well.
Additionally, I verified that standard deviations are very close to zero and this would imply that the variance is close to zero as well, which could explain why the bounds appear identical to the mean. But I do not understand the reason behind it also considering that settings for this variable are the same as those for other variables in the model.
Has anyone experienced a similar issue or have any suggestions ?
Thanks so much for any insights!
Identical Higher, Lower bounds and mean in Stochastic Model
Moderators: EViews Gareth, EViews Moderator

 Fe ddaethom, fe welon, fe amcangyfrifon
 Posts: 13404
 Joined: Tue Sep 16, 2008 5:38 pm
Re: Identical Higher, Lower bounds and mean in Stochastic Model
The standard deviation is computed from the underlying equation. Does that equation have a very small residual variance?
Follow us on Twitter @IHSEViews
Re: Identical Higher, Lower bounds and mean in Stochastic Model
Thank you so much for your answer!!
To obtain the residual variance I've calculated 0.017239^2≈0.000297
This value is relatively small.
I report here the output from my equation:
Rsquared 0.742984 Mean dependent var 0.001580
Adjusted Rsquared 0.738838 S.D. dependent var 0.033733
S.E. of regression 0.017239 Akaike info criterion 5.263701
Sum squared resid 0.073703 Schwarz criterion 5.193871
Log likelihood 670.8582 HannanQuinn criter. 5.235606
DurbinWatson stat 1.957704
However, the size of the confidence interval, given by the higher and lower bounds can be selected: the default size (the one I've choosen) of 0.95 provides a 95% confidence interval with a weight of 2.5% in each tail.
To obtain the residual variance I've calculated 0.017239^2≈0.000297
This value is relatively small.
I report here the output from my equation:
Rsquared 0.742984 Mean dependent var 0.001580
Adjusted Rsquared 0.738838 S.D. dependent var 0.033733
S.E. of regression 0.017239 Akaike info criterion 5.263701
Sum squared resid 0.073703 Schwarz criterion 5.193871
Log likelihood 670.8582 HannanQuinn criter. 5.235606
DurbinWatson stat 1.957704
However, the size of the confidence interval, given by the higher and lower bounds can be selected: the default size (the one I've choosen) of 0.95 provides a 95% confidence interval with a weight of 2.5% in each tail.

 Fe ddaethom, fe welon, fe amcangyfrifon
 Posts: 13404
 Joined: Tue Sep 16, 2008 5:38 pm
Re: Identical Higher, Lower bounds and mean in Stochastic Model
We'd probably need to have the workfile to give more advice.
Re: Identical Higher, Lower bounds and mean in Stochastic Model
The point I wanted to stress more before is the following:
If I've understood correctly the error bounds are measured by using the tails of the distribution. Specifically,
 To simulate the distribution, the model object uses a Monte Carlo approach.
 Then, the Interval size (2 sided) box lets us select the size of the confidence interval given by the upper and lower bounds.
Therefore, I do not understand why these bounds are identical to the mean.
If I've understood correctly the error bounds are measured by using the tails of the distribution. Specifically,
 To simulate the distribution, the model object uses a Monte Carlo approach.
 Then, the Interval size (2 sided) box lets us select the size of the confidence interval given by the upper and lower bounds.
Therefore, I do not understand why these bounds are identical to the mean.
Re: Identical Higher, Lower bounds and mean in Stochastic Model
This is my code:
SOLVE MODELS
if(!update_model_base==1) then
smpl model_smpl
{%model_name}.update
{%model_name}.msg
{%model_name}.scenario "Baseline"
rndseed 123
{%model_name}.stochastic(d=t)
{%model_name}.solve(s=a)
smpl @all
endif
SOLVE MODELS
if(!update_model_base==1) then
smpl model_smpl
{%model_name}.update
{%model_name}.msg
{%model_name}.scenario "Baseline"
rndseed 123
{%model_name}.stochastic(d=t)
{%model_name}.solve(s=a)
smpl @all
endif
Who is online
Users browsing this forum: No registered users and 1 guest