I know the state space framework is flexible enough to accept different representantions of the same question.
The following pieces of code get close variances parameters, but not the same. Is this a state space representation question or there's an addtonal parameter I should set in Eviews?
'Code 1:
depvar1 = SV1 + SV2 + [ENAME=E1, VAR=192*EXP(C(1))]
depvar2 = C(3)*depvar2(-1) + E1
@STATE SV1= SV1(-1) + [VAR=EXP(C(1))]
@STATE SV2 = SV2(-1) + SV3(-1) + SV4(-1) + [VAR=EXP(C(6))]
@STATE SV3= SV2(-1)
@STATE SV4= SV3(-1)
@PARAM C(1) -14.3 C(3) 1 C(6) -6.87
'Code 2:
depvar1 = SV1 + SV2 + [ENAME=E1, VAR=192*EXP(C(1))]
depvar2 = C(3)*depvar2(-1) + SV5
@STATE SV1= SV1(-1) + [VAR=EXP(C(1))]
@STATE SV2 = SV2(-1) + SV3(-1) + SV4(-1) + [VAR=EXP(C(6))]
@STATE SV3= SV2(-1)
@STATE SV4= SV3(-1)
@state SV5 = E1
@PARAM C(1) -14.3 C(3) 1 C(6) -6.87
The difference between them, is that, at "code 2", I have replaced "E1" for SV5 in the second equation. I also created a state variable "SV5" that is equal to "E1".
State Space equivalent representations?
Moderators: EViews Gareth, EViews Moderator
-
- EViews Developer
- Posts: 2676
- Joined: Wed Oct 15, 2008 9:17 am
Re: State Space equivalent representations?
Could you please post your workfile so we are looking at the same thing (and to make discussion easier). Thanks.
My first instinct is that there is a difference in the timing between the covariance between the states and the signals in the two cases. The second case has a correlation between the DEPVAR1 and SV5 covariances at (t, t+1) so the DEPVAR1 and DEPVAR2 covariances are at 1 period differences, where the former has the correlation between the two signal covariances is defined for (t, t).
My first instinct is that there is a difference in the timing between the covariance between the states and the signals in the two cases. The second case has a correlation between the DEPVAR1 and SV5 covariances at (t, t+1) so the DEPVAR1 and DEPVAR2 covariances are at 1 period differences, where the former has the correlation between the two signal covariances is defined for (t, t).
Return to “General Information and Tips and Tricks”
Who is online
Users browsing this forum: Bing [Bot] and 0 guests