TVAR model

Discussion of models with structural breaks or endogenous switching.
Scalper
Posts: 2
Joined: Thu Oct 13, 2011 2:02 pm

TVAR model

Unread post by Scalper »

Hello,

I am estimating a TVAR model for my PHd Thesis, but i have some doubts, mainly, in apply the commands in RATS. I´ve never used it before.
Firstly, how can i know which variable is the threshold variable?! Through a Test Tsay?
Secondly, for the calculating of threshold Value, we use which method : Tsay Test or Hansen test or both?!

Then we know that, is it possible to do the Granger Causality for each Regime?

Yours Sincerely,

Joaquim Ferreira
TomDoan
Posts: 7814
Joined: Wed Nov 01, 2006 4:36 pm

Re: TVAR model

Unread post by TomDoan »

The Tsay JASA 1998 examples are the best source for that, although the published results were wrong.

Given the threshold, you can run Granger causality tests on each branch since, conditional on the threshold, you can just do OLS estimation of the VAR equations.
Scalper
Posts: 2
Joined: Thu Oct 13, 2011 2:02 pm

Re: TVAR model

Unread post by Scalper »

thank you TomDoan. But I though that the TSAY test it was only to determinate the delay paremeter.

how can i determinate the threshold Variable=??


Therefore, i can´t import my xls files to RATS and run the codes either. I dont know why. could you help me?

Thank you very much

yours sincerely,
TomDoan
Posts: 7814
Joined: Wed Nov 01, 2006 4:36 pm

Re: TVAR model

Unread post by TomDoan »

Scalper wrote:thank you TomDoan. But I though that the TSAY test it was only to determinate the delay paremeter.

how can i determinate the threshold Variable=??
The largest LM test combination of delay and threshold variable would be the best choice.
Scalper wrote:Therefore, i can´t import my xls files to RATS and run the codes either. I dont know why. could you help me?
If you're having trouble reading your data set, you should contact support@estima.com.
Post Reply