The RATS Software Forum

Hello,

I am relatively new to CATS, so I would like to know if one can use this package to estimate a smooth transition model
à la Franchi - Ordonez (2008) "Common smooth transition trend-stationarity in European unemployment" Economics Letters.

Here I attach the link to the article.

Thanks a lot!

http://www.sciencedirect.com/science?_o ... archtype=a

The link you provided doesn't work, so I can't comment on the specifics of the paper you were trying to reference.

However, the CATS procedure itself isn't set up to handle any sort of STAR model.

I suppose you could consider modifying the CATS code to do so, but that would be extraordinarily complicated. I think you'd be much better off just writing the appropriate code directly. We do discuss STAR models in the RATS User's Guide, and you'll find more discussions here on the forum.

Also, our structural breaks and switching models course covered threshold models. You can now purchase the materials from that course for $50. See:

http://www.estima.com/courses_completed.shtml

Regards,
Tom Maycock
Estima

Thanks alot, Tom: great!

I will first check the User's guide, then go to the course you mentioned.

Best and thanks again!

Hi,

I am using the code in the users guide to perform STAR models in RATS. I use as my transition variable the first lag of the dependent variable and I also include in the model an exogenous variable.
My question is when I add a second exogenous variable I obtain different values for gamma and c from the ones with one exogenous variable. Is that normal since these parameters are estimated form the same variable? Sorry if my question may seem unintelligent but I am a neophyte with STAR models and econometrics.

Thanks

This has nothing to do with CATS does it?

They're different regressions, so there's no reason to expect that the transition points would be the same. What happens when you fit a linear regression to the data with and without the extra variable?

Ok thank you for the information.