Markov Regime Switching & Smooth Transition Regression

[Concordia]

Hi,

I have two questions: (1) one related to Markov Switching Regression and (2) one related to Smooth Transition Regression.

(1) I would like to know if it is possible to get the mean of the dependent variable for each regime when using the Markov's switching regression model.
In the results I only get the overall mean.

(2) Can we perform Smooth Transition Regression in EViews?

Thank you very much

[EViews Glenn]

I'm not certain what you mean by the "mean of the independent variable in each regime". All that you get for each regime is a probability that you are in that regime in a given period. There is no certainty to the regime identity.

As to STAR, it is not built-in. I don't believe that anyone has an example, though there is one for TAR models

http://forums.eviews.com/viewtopic.php? ... old#p26248

[trubador]

Actually, the example is for Logistic STAR model, which uses the logistic function for smooth transition. Other forms can easily be formulated.
As for the Markov switching, I think what you are asking is the following:

Code: Select all

'get the regime probabilities my_eq.makergmprobs rp1 rp2 'assign to the regimes with relatively higher probability series r1 = (rp1>=.5) series r2 = (rp2>=.5) 'compute the number of periods in each regime scalar nr1= @sum(r1) scalar nr2 = @sum(r2) 'split the dependent variable with respect to the regimes r1 = r1*y r2 = r2*y 'compute mean value for each regime scalar mr1 = @sum(r1)/nr1 scalar mr2 = @sum(r2)/nr2

But, as Glenn pointed out, there is no certainty to the regime identity.

[EViews Glenn]

That's what I get for just glancing at the routine. I stand corrected. And thanks for the contribution, trubador.

[Concordia]

Thanks a lot Glenn and Trubador for your prompt responses.