Sspace - Kalman Gain

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

workhard
Posts: 8
Joined: Sun Jul 12, 2009 10:51 pm

Sspace - Kalman Gain

Postby workhard » Wed Sep 29, 2010 12:06 am

Dear all,

May I know how Eviews 7 calculate the Kalman Gain? What is the formula?
I use the command "@curr_gain" to get the Kalman gain. However, I am not sure how eviews get the values.
The Eviews User Guide does not mention about the this. Also, is it we only able to get the Kalman gain by using command?

Any prompt reply is greatly appreciated!!

EViews Glenn
EViews Developer
Posts: 2672
Joined: Wed Oct 15, 2008 9:17 am

Re: Sspace - Kalman Gain

Postby EViews Glenn » Wed Sep 29, 2010 9:39 am

As you have noticed, the EViews manual does not offer a comprehensive discussion of Kalman filtering and state space estimation. For that, we highly recommend that you consult one or more of the many references outlined in the manual. The Kalman gain should be described in detail in any one of those sources.

The @curr_gain, like many of the other state and signal results, are available only through the command interface. Only the one-step-ahead, filtered, and smoothed signal and state results are available through the dialog interface. The notion is that something like the Kalman gain is generally of interest for subsequent calculation, whereas the one-step, filtered, and smoothed results are most often the results of interest. A simple one line command should extract the results for you.

atbarbar
Posts: 6
Joined: Thu Aug 24, 2017 8:58 am

Re: Sspace - Kalman Gain

Postby atbarbar » Thu Mar 07, 2024 8:34 pm

I think that state space modeling with the Kalman filter is now well ingrained in applied econometrics that updates of Eviews should definitely provide easier access to the Kalman gain and other Kalman filter quantities. The comparative advantage of Eviews, relative to other packages, is that it is quick to do stuff and results are easily accesses and perused by clicking. By confining the gain to command window you kill that comparative advatange. Please reconsider this point and make the Kalman quantities accessible by clicking. Also, you should implement the "decomposition by observables" by Koopman, S.J., and A. Harvey, 2003, “Computing observation weights for signal extraction and filtering,” Journal of Economic Dynamics and Control, Vol. 27, pp. 1317–1333.as well. That is tremendously useful.


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 17 guests