smooth transition regression model

[vasco27]

Hi:
I am trying to run a smooth transition regression model on time-series data. I understand that eviews does this through non-linear squares (which requires entering the actual equation) and LogL object (which I believe is log likelihood models). Since I’m at the initial stage of learning how to actually run this, the questions I have are: (i) how do I define and introduce the transition function? (ii) how is the threshold parameter determined? (iii) how is the time invariant regressor introduced into the equation? Does it refer to the starting value of the dependent variable? I also have an older version of eviews (4.0)
Thanks in advance for your help

[lgm]

Hi there,

I was wondering if you've got help on your questions. I have similar questions and wonder if you can help.

Thanks

[vasco27]

No, unfortunately I've had no replies. Will be happy to share them if I ever receive them though

[lgm]

Hi there,

Thanks for your reply.

I've found this discussion which I thought might be of help. Please find the link below.

http://forums.eviews.com/viewtopic.php?f=4&t=3712

To select whether the STR is LSTR or ESTR, may be good to read "Applied Time Series Econometrics", a book by Luetkepohl, Helmut Kraetzig, Markus Phillips, Peter C. B. or check this http://www.diva-portal.org/smash/get/di ... TEXT01.pdf

Kind regards,

lgm

[vasco27]

Thank you for the link, much appreciated

[lgm]

Thanks. All the very best.