[R-SIG-Finance] Markov Switching

Patrick Brandt patrick.t.brandt at gmail.com
Wed Sep 23 00:32:37 CEST 2009

If you look at the package list on CRAN you will see a package that
meets you needs for a frequentist MS VAR: MSVAR.  It is listed right
after (my) MSBVAR in the alphabetical listing of packages.

The MSBVAR vignette is in progress right now.  The real challenge here
is coding a general implementation of Bayesian MS BVARs that is useful
to the largest possible audience.

My intent is that the MSBVAR package is for BAYESIAN models.  At
present, the only frequentist model is a reduced form VAR.  And that
is only there for comparison purposes...

If you need general VAR models, check out the vars and urca packages.


Patrick Brandt
Assistant Professor
Political Science
School of Economic, Political and Policy Sciences
University of Texas at Dallas
Personal site: http://www.utdallas.edu/~pbrandt
MSBVAR site: http://yule.utdallas.edu

On Tue, Sep 22, 2009 at 7:53 AM, Angel Spassov <anspassov at googlemail.com> wrote:
> DeaR list,
> I am looking for a descent implementation of
> a Markov Switching Vector Autoregressive Model.
> Until now I found the following packages:
> 1) MSBVAR: It seems that this package estimates
> Markov-Switching VAR-models only from a Bayesian
> point of view? Correct me if I am wrong.
> I also need it from a frequentist point of view.
> Assumed, this package is the single option,
> I would greatly appreciate some basic
> code of how to estimate a single model with it.
> For example, how can I model the following
> bivariate time series:
> set.seed(1234)
> myts <- as.ts(data.frame(a=rnorm(100),b=rnorm(100) + 2))
> If I got the man pages correctly, I have to use both
> "msbvar" and "gibbs.msbvar". Mr. Brandt is
> pointing the user to a non-existing vignette
> (see ?msbvar, "Note" section) and no example are
> provided in the man pages.
> I think MSBVAR is a challenge for a new user.
> 2) fMarkovSwitching: This package is not compatible
> with the latest version of R and is seemingly
> suitable only for univariate models.
> Nevertheless, at least I succeeded to estimate
> my model and to interpret the results with this package.
> Any other suggestions?
> Angel.
> _______________________________________________
> 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