[R-SIG-Finance] ARMAX, t-GARCH estimation(RUGARCH package)

[Alexios Ghalanos]

It means that the standard errors could not be calculated as a result of not being able to invert the hessian during the post-estimation phase. As explained in the manual and vignette, general failures in the final optimization/post-estimation can generally be approached by changing the solver and/or tweaking the solver.
Other suggestions:
- DO use model 'sGARCH' rather than the 'GARCH' model of 'fGARCH' for estimation.
- Because you include regressors in the variance equation, try to provide a starting parameter for those and also try eGARCH where positivity is guaranteed  (possibly the source of the problem).

Regards,
Alexios


On Feb 11, 2012, at 10:44, ecsniffer林娟 <ecsniffer at hotmail.com> wrote:

> 
> I use ARMA(4,0), t-GARCH(1,1) model to estimate the volatility series(y_t). In the model, I add rate ( r_t ) and lagged rate( r_{t -1} ) to the conditional mean function and conditional variance function as the exogenous variables.
> 
> Here is my code:
> 
> data<-read.csv("...")
> volatility<-data$volatility;
> volatility.lag<-embed(volatility,2);
> volatility<-as.matrix(volatility.lag[,1]);
> rate<-data$diff_rate;
> rate.lag<-embed(rate,2);
> spec=ugarchspec(variance.model=list(model="fGARCH",garchOrder=c(1,1),submodel="GARCH", 
> external.regressors=as.matrix(rate.lag), variance.targeting = FALSE),
> mean.model = list(armaOrder = c(4,0), include.mean = TRUE, garchInMean = FALSE,
> inMeanType = 1, arfima = FALSE, external.regressors =as.matrix(rate.lag)), distribution.model = "std",
> start.pars = list(), fixed.pars = list())
> fit=ugarchfit(spec,data=volatility)
> fit
> 
> I got some errors and can't figure out the reason. Thanks!
> 
> Iter: 1 fn: 2986.4841    Pars:   -0.211418971918   0.097905377888  -0.084512900004  -0.053448303895  -0.071041833756 -22.798457638632  16.379426108878   1.626051893520   0.206719227655   0.747514215831   0.000000009905   0.000000009905   4.008451660273
> Iter: 2 fn: 2986.4841    Pars:   -0.211418971919   0.097906631452  -0.084513745760  -0.053447443169  -0.071041878730 -22.797651046488  16.377961083304   1.626059793075   0.206723362776   0.747509647714   0.000000009905   0.000000009905   4.008479595008
> solnp--> Completed in 2 iterations
> Warning:
> In .makefitmodel(garchmodel = "fGARCH", f = .fgarchLLH, data = data,  : 
> rugarch-->warning: failed to invert hessian
> 
> Errors in t.default(grad) : Parameters are not the matrix.                         
>    [[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.