Regression with ARMA errors and Student T innovations

You can do this with the rgarch package on r-forge which at present can
be downloaded from: rgarch.r-forge.r-project.org/download.html.

The functions you are likely to need are 'arfimaspec' and 'arfimafit'.
There are also methods for filtering, forecasting, simulation and
rolling estimation (see the documentation).

Extensive examples can be found in 'unit.test13.R' In the
'inst/src/rgarch.tests' folder of the source package or the
'rgarch.tests' folder of the binary package.

Regards,

Alexios Ghalanos

Is there a package that can do this?
And if not would you recommend coding the joint likelihood in R and then use
optim()?
The MA-part scares me a bit because of computation time.

I am not a C coder so I would like to avoid reading through C code of ARMA
regression + Normal innovations (like gls() ) and adapt that to Student T.

Thanks a lot for your help,

Arthur