[R-SIG-Finance] threshold autoregression

Matthieu Stigler stigler3 at etu.unige.ch
Thu Jan 24 11:32:28 CET 2008


Hello Alex and Sylvain

Alex, do you want to do univariate or multivariate analysis?

For univariate analysis, SETAR models are available in R in package TSA 
(with diagnostics tests, forecasts, simulation... the package is based 
on examples of a book) and in package BAYSTAR for Bayesian estimation.

STAR models are more rare and you can just find some basic features in 
packages tsDyn. You can find also STAR with time varying coefficients in 
package timsac. For others software, I don't know about Ox but have a 
look on Jmulti, free software based on GAUSS code with Terasvitra code 
for STAR. If you are generally interested in TAR models, have also a 
look on the website of Bruce Hansen, who wrote many important articles 
and makes available these papers and the code used, for Matlab (which 
you should able to run from octave, open source software) and GAUSS.

If you are interested in multivariate analysis, there is for now nothing 
in R to my knowledge. I'm working on my part on multivariate series with 
thresholds (TAR and TVECM for threshold cointegration) and I can send to 
you my code if interest.

I' plan to write some functions (which can be applied to univariate 
analysis, like Hansen linearity test for SETAR (1999) or Lundbergh et al 
(2003) test for parameter constancy), so if you or anyone has interest 
in writing some code and extending the actual features tell me!

Regards

Sylvain BARTHELEMY a écrit :
> Hi A
>
> If you want to play with SETAR and more generally with Markov switching
> models, my view is that you should use another software than R. 
>
> There is a great pack written by Hanz Martin Krozlig. But as soon as I know,
> it only works on Ox and it has not been ported to R. I think that it would
> be a great idea to rewrite the pack for R. I initially planned to do one,
> but it takes time...
>
> There is also a pack written by Franck Arnaud, that was written for R and is
> available here: http://arnaud.ensae.net/Rressources/Rressources.html
> But it seems to be a beta version and less powerful than the Krozlig one.
>
> Regards
>
> ---
> Sylvain Barthélémy
> Research Director, TAC
> www.tac-financial.com | www.sylbarth.com
>
>
> -----Message d'origine-----
> De : r-sig-finance-bounces at stat.math.ethz.ch
> [mailto:r-sig-finance-bounces at stat.math.ethz.ch] De la part de Alexander
> Moreno
> Envoyé : dimanche 20 janvier 2008 23:47
> À : r-sig-finance at stat.math.ethz.ch
> Objet : [R-SIG-Finance] threshold autoregression
>
> Hi,
>
> Does anyone know of anyone packages to fit tar and setar time series models?
>
> Thanks,
> Alex
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-SIG-Finance at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only. 
> -- If you want to post, subscribe first.
>
> No virus found in this incoming message.
>
>
>
> 14:15
>
>
>
>
>
>
> 20:12
>
> _______________________________________________
> R-SIG-Finance at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only. 
> -- If you want to post, subscribe first.
>



More information about the R-SIG-Finance mailing list