Markov switching regime probabilities

For notifying us of what you believe are bugs or errors in EViews.
Please ensure your copy of EViews is up-to-date before posting.

Moderators: EViews Gareth, EViews Moderator

startz
Non-normality and collinearity are NOT problems!
Posts: 3507
Joined: Wed Sep 17, 2008 2:25 pm

Markov switching regime probabilities

Postby startz » Fri Jul 08, 2016 10:02 am

I have a Markov switching estimate where if I ask for a graph of the smoothed probabilities the graph is labelled filtered. What's more, if I make series of filtered and smoothed probabilities the reported numbers are identical.

But this only happens for one case, not for others. In the attached file China works fine and Costa Rica exhibits the problem.

...I wish you luck with this one.
Attachments
pwt81.wf1
(3.64 MiB) Downloaded 74 times

startz
Non-normality and collinearity are NOT problems!
Posts: 3507
Joined: Wed Sep 17, 2008 2:25 pm

Re: Markov switching regime probabilities

Postby startz » Fri Jul 08, 2016 10:20 am

further (smaller) bug: "regime" is misspelled when a regime probability series is created

Code: Select all

Smoothed probability of regieme 2, from equation SPAINAR

EViews Glenn
EViews Developer
Posts: 2620
Joined: Wed Oct 15, 2008 9:17 am

Re: Markov switching regime probabilities

Postby EViews Glenn » Fri Jul 08, 2016 4:53 pm

We'll fix the typo.

The estimated equation for Costa Rica is a simple switching model, not a Markov switching model so different period probabilities are unrelated.

startz
Non-normality and collinearity are NOT problems!
Posts: 3507
Joined: Wed Sep 17, 2008 2:25 pm

Re: Markov switching regime probabilities

Postby startz » Fri Jul 08, 2016 5:25 pm

Thanks for the substantive catch

EViews Glenn
EViews Developer
Posts: 2620
Joined: Wed Oct 15, 2008 9:17 am

Re: Markov switching regime probabilities

Postby EViews Glenn » Fri Jul 08, 2016 6:13 pm

From an interface point of view I thought that perhaps I should find a place to warn people of this possibility, but it wasn't obvious where to put in the graphical output. Perhaps we should simply disallow smoothing as an option in the simple switching case as I can see why this may be confusing.

startz
Non-normality and collinearity are NOT problems!
Posts: 3507
Joined: Wed Sep 17, 2008 2:25 pm

Re: Markov switching regime probabilities

Postby startz » Fri Jul 08, 2016 7:31 pm

Yeah, I think that would be a good idea. It's definitely weird to choose smoothed and get a graph labeled filtered.

EViews Glenn
EViews Developer
Posts: 2620
Joined: Wed Oct 15, 2008 9:17 am

Re: Markov switching regime probabilities

Postby EViews Glenn » Mon Jul 18, 2016 1:08 pm

It's a moderately big change to alter the interface to restrict this and I'm inclined not to do it at this time. What I'll do for now is to expand the labeling of the output so that it's clearer the model that was estimated. For example "Simple Switching Filtered/Smoothed" and "Markov Switching Filtered", "Markov Switching Smoothed", etc. I think that will make things obvious in situations like those described here.

startz
Non-normality and collinearity are NOT problems!
Posts: 3507
Joined: Wed Sep 17, 2008 2:25 pm

Re: Markov switching regime probabilities

Postby startz » Mon Jul 18, 2016 1:11 pm

Makes sense.


Return to “Bug Reports”

Who is online

Users browsing this forum: No registered users and 1 guest