[R-SIG-Finance] rmgarch: dccforecast() and mregfor

[Alexios Ghalanos]

If you have a multivariate conditional mean specification (i.e. VAR) then you cannot have a univariate conditional mean specification (arma model)...they are mutually exclusive. In short, do not enter anything for mean.model in ugarchspec (include.mean is automatically set to FALSE if VAR is selected).

-Alexios

On 2 Sep 2013, at 11:44 AM, Cedric Membrez <cedric.membrez at gmail.com> wrote:

> Dear all,
> 
> I'm a first time user of this list. Please let me know if I need to provide
> more info,etc.
> I have a question regarding dccforecast() and the number of external
> regressors needed.
> 
> I'm using the dccforecast() function from the package "rmgarch" and use the
> simple DCC model and a AR(1)-GARCH(1,1). The forecast works fine when I do
> NOT include external regressors. However, I have an error message from the
> dccforecast() function when I include the external regressors in the
> external.forecasts=list(mregfor = ...) option.
> 
> "Error: dccforecast-->error: wrong number of external regressors!..."
> 
>> From the documentation, I understand that I need to have unconstraint
> forecast of my ext. regr and these regressors must be the same as in the
> ugarchspec(). I googled my problem, but nothing helpful came up. Anyone see
> the mistake here ? I appreciate your help :) thanks!
> 
> Here are the main steps I took and my code is below.
> 
> step (1):  univariate GARCH specification with ARMA(1,0), GARCH(1,1) and a
> 842x17 matrix of external regressors. Then, I create a multispec object.
> step (2):  varxfit( ... data matrix of 758x6,  and regressors of 758x17
> dimensions)
> step (3): dccspec( multispec, ext.regr of 842x17  dimensions)
>              dccfit( dccspec, data of 842x6  and out.sample=84)
> step (4) dccforecast ( dccfit,  n.ahead = 1 , n.roll = 5,  mregfor of
> dimension 842.17 )
> ==>  and I get the embarrassing message with "wrong number of ext.
> regressors!..."
> 
> My code:
> ###########################################################
> uniGarchSpec <- ugarchspec(mean.model = list(armaOrder=c(1,0),
>                                             external.regressors =
> lag(extRegr)),
> 
> variance.model=list(model="sGARCH",
>                                             garchOrder=c(1,1)))
> multiGarchSpec <- multispec(replicate(6, uniGarchSpec))
> # VAR
> varFit <- varxfit(X=data.agriFact[-outSample.ind,c(5:9, 23)],
>                                p=1, postpad='constant', robust=FALSE,
> ns=200,
>                                exogen=extRegr[-outSample.ind,])
> # DCC
> dccSpec <- dccspec(multiGarchSpec, VAR=TRUE, robust=FALSE,
>                             lag=1, lag.max=NULL,
>                             external.regressors=lag(extRegr),
>                             robust.control = list("gamma" = 0.25,
>                                                   "delta" = 0.01, "nc" =
> 10, "ns" = 250),
>                             dccOrder=c(1,1), model=c("DCC"),
> distribution=c("mvt"),
>                             start.pars=list(dcca1=0.05, dccb1=0.816))
> dccFit <- dccfit(dccSpec, data=lag(data.agriFact[,c(5:9, 23)]),
>                                  VAR.fit = varFit, out.sample=84)
> # Forecast
> dccForecast <- dccforecast(dccFit, n.ahead=1,n.roll=5,
>                           external.forecasts=list(mregfor=lag(extRegr)))
> ###########################################################
> 
>    [[alternative HTML version deleted]]
> 
> _______________________________________________
> R-SIG-Finance at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only. If you want to post, subscribe first.
> -- Also note that this is not the r-help list where general R questions should go.
>