get ergodic probabilities from a MS estimation

For questions regarding the import, export and manipulation of data in EViews, including graphing and basic statistics.

Moderators: EViews Gareth, EViews Steve, EViews Moderator, EViews Jason

sanfrina
Posts: 3
Joined: Fri Oct 17, 2014 4:57 pm

get ergodic probabilities from a MS estimation

Postby sanfrina » Fri Oct 17, 2014 5:07 pm

Dear all,

I have a very urgent request of help!
I am running switching regressions (switchreg (type=markov, k=2)) on different series. I need to save the ergodic probabilities from each model so that, I can export those ergotic probabilities in excel.
How should I compute it in eviews? Is there a command in eviews?

thanks a lot in advance for your help

sanfrina
Posts: 3
Joined: Fri Oct 17, 2014 4:57 pm

Re: get ergodic probabilities from a MS estimation

Postby sanfrina » Fri Oct 17, 2014 5:45 pm

Just to give more details:
if there is no quick command to get the ergodic probability of such MS,2-state regression, how can I compute them?
The ergodic probabilitiy vector should be the eigenvector of the transition probability matrix of the markov chain associated with the unit eigenvalue.
Can I compute that in eviews with a matrix of transition probability P which is NOT symmetric?

Thanks a lot

EViews Glenn
EViews Developer
Posts: 2682
Joined: Wed Oct 15, 2008 9:17 am

Re: get ergodic probabilities from a MS estimation

Postby EViews Glenn » Sat Oct 18, 2014 7:53 am

There isn't anything built-in.

You'll have to grab the transition probability matrix and do the computation yourself. Unfortunately, EViews doesn't let you compute eigenvalue decompositions of non-symmetric matrices. Fortunately, for a two-state model, the computation looks to be pretty easy to do by hand.

sanfrina
Posts: 3
Joined: Fri Oct 17, 2014 4:57 pm

Re: get ergodic probabilities from a MS estimation

Postby sanfrina » Sat Oct 18, 2014 10:17 am

Hi Glenn,

thank you so much for your answer!
Yes, fortunately now I can compute the ergodic using the transition probabilities, but I wanted then to test the model with more states, i.e. k>3 (bear, bull and steady state) and I guess it will be hard to do in eviews!

Thanks a lot for your help!

EViews Glenn
EViews Developer
Posts: 2682
Joined: Wed Oct 15, 2008 9:17 am

Re: get ergodic probabilities from a MS estimation

Postby EViews Glenn » Sun Oct 19, 2014 4:45 pm

Even for 3-states it doesn't look particularly difficult.


Return to “Data Manipulation”

Who is online

Users browsing this forum: No registered users and 2 guests