[R-SIG-Finance] GJR-GARCH

[alexios ghalanos]

Change the bounds for alpha (which default to positive and based on the
[1,1] model):

>setbounds(STOXX50.tgarch21.spec)<-list(alpha1=c(-1,1), alpha2=c(-1,1))

-Alexios

PS TGARCH is for the standard deviation (and hence does not nest the
vanilla GARCH) whilst GJR is for the variance (subtle little difference
which appears to escape many).

On 12/08/2014 11:00, Urs Gröpl wrote:
> 
> Dear all,
> 
> I really enjoy working with the rugarch package and now my first question came up.
> I have a problem regarding the GJR-GARCH model implementation in the rugarch package in R.
> 
> I want to fit it to the EURO STOXX 50 index return series, but after fitting it, there is still significant autocorrelation at lag 1 and the Sign Bias tests show still evidence for asymmetric effects.
> There is no such problem with the GARCH(2,1) and EGARCH(2,1), only with the GJR-GARCH(2,1). I checked several other orders of the model, but it didn’t help.
> There is no error message, the GJR model just fails to incorporate the asymmetric effects and the alpha estimates are zero.
> I tried all solvers, changed the starting value, and increased the iteration numbers. The GJR-GARCH nested in the APARCH or fGARCH function yield the same problem.
> 
> I was asking myself, whether the failure is just due to the model itself or if there could be numerical problems, even there is no error message.
> 
> 
> Here is my code for the model specification and fitting:
> STOXX50.tgarch21.spec=ugarchspec(variance.model=list(model="gjrGARCH",garchOrder=c(2,1)),mean.model=list(armaOrder=c(5,0)),fixed.pars=list(ar1=0,ar2=0,ar4=0),distribution.model="sstd")
> STOXX50.tgarch21.st.fit=ugarchfit(spec=STOXX50.tgarch21.spec,rtnSTOXX50,solver="hybrid")
> STOXX50.tgarch21.st.fit
> 
> 
> I retrieved the data from the yahoo finance website:
> symbol.vec = c("^STOXX50E")
> getSymbols(symbol.vec,from="2000-01-03",to="2013-12-31")
> STOXX50=STOXX50E[,"STOXX50E.Adjusted",drop=F]
> length(STOXX50)
> rtnSTOXX50=CalculateReturns(STOXX50, method="log")
> rtnSTOXX50=rtnSTOXX50 [-1,]
> colnames(rtnSTOXX50)="STOXX50"
> 
> 
> I would be deeply grateful for any advice.
> Thank you very much in advance for your answer.
> 
> Regards,
> Urs
>  
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 	[[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.
>