Jeroen van der Heide
2007-Jul-19 20:53 UTC
[R] Questions regarding R and fitting GARCH models
Dear all, I've recently switched from EViews to R with RMetrics/fSeries (newest version of july 10) for my analysis because of the much bigger flexibility it offers. So far my experiences had been great -prior I had already worked extensively with S-Plus so was already kind of familiar with the language- until I got to the fSeries package. My problem with the documentation of fSeries is that it is pretty sparse; therefore I don't whether I am doing something wrong or if my problem is related to elementary statistics (though I hold an MSc in Econometrics, it's been quite awhile :o). Next to that I have two questions related to the syntaxis of the R language itself; I've been searching for a good couple of hours but couldn't find the answers. Hope you can help me out.>From the descriptive statistics of my series, I had determined that myGARCH error term should follow a student's t distribution, preferrably skewed; the sstdFit function returns nu=2.002 and xi=1.012. I know that for this distribution 2 degrees of freedom means the second, third and fourth moment are not defined; however, the QQ plot looks good -had to use the skwt package though, because rsstd didn't work- so I decided to give it a try. 1) That didn't go all to well. garchFit returns alpha1 and beta1 of 0.1 and 0.8 respectively (and beta1 has a standard error of NaN) - which are not consistent at all with what EViews reported. How is this possible? Is it because GARCH estimation with a student's t distribution with 2 degrees of freedom is impossible? What are my options in this case? Or do I need to use garchSpec, of which I have absolutely *no* idea what its use is (i.e. the entire idea is to obtain estimates for a given order of AR-GARCH, right)? 2) R/RGui crashes frequently - also if I "just" want to estimate a GARCH(1,1) with a student's t distributed error term and 4 or 5 degrees of freedom. I think it just gets into an endless calculation loop; is there a way to abort this? 3) Is there a way to extract the data from the garchFit objects? So if "fit" is my object, how can I extract specific data like the fit$coef[1] I can use with other objects? Ultimately, fSeries comes with a really nice plotting function (i.e., plot(fit, which=2)) but I want to extract the series it actually plots from there - is that possible like the hist function? Thanks a lot for your time. Best regards, Jeroen van der Heide
As for (3) below, we are also interested in the xi parameter from the gpdFit function from fExtremes. It seems the fit classes returned by the associated functions don't have the full properties as finMetrics has for its classes. My solution was to modify the appropriate fMetrics code, build a new package, and install it. I haven't completed that yet. Are there any other things I can try? Dale Smith Vicis Capital, LLC -----Original Message----- From: r-help-bounces at stat.math.ethz.ch [mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Jeroen van der Heide Sent: Thursday, July 19, 2007 4:54 PM To: r-help at stat.math.ethz.ch Subject: [R] Questions regarding R and fitting GARCH models Dear all, I've recently switched from EViews to R with RMetrics/fSeries (newest version of july 10) for my analysis because of the much bigger flexibility it offers. So far my experiences had been great -prior I had already worked extensively with S-Plus so was already kind of familiar with the language- until I got to the fSeries package. My problem with the documentation of fSeries is that it is pretty sparse; therefore I don't whether I am doing something wrong or if my problem is related to elementary statistics (though I hold an MSc in Econometrics, it's been quite awhile :o). Next to that I have two questions related to the syntaxis of the R language itself; I've been searching for a good couple of hours but couldn't find the answers. Hope you can help me out.>From the descriptive statistics of my series, I had determined that myGARCH error term should follow a student's t distribution, preferrably skewed; the sstdFit function returns nu=2.002 and xi=1.012. I know that for this distribution 2 degrees of freedom means the second, third and fourth moment are not defined; however, the QQ plot looks good -had to use the skwt package though, because rsstd didn't work- so I decided to give it a try. 1) That didn't go all to well. garchFit returns alpha1 and beta1 of 0.1 and 0.8 respectively (and beta1 has a standard error of NaN) - which are not consistent at all with what EViews reported. How is this possible? Is it because GARCH estimation with a student's t distribution with 2 degrees of freedom is impossible? What are my options in this case? Or do I need to use garchSpec, of which I have absolutely *no* idea what its use is (i.e. the entire idea is to obtain estimates for a given order of AR-GARCH, right)? 2) R/RGui crashes frequently - also if I "just" want to estimate a GARCH(1,1) with a student's t distributed error term and 4 or 5 degrees of freedom. I think it just gets into an endless calculation loop; is there a way to abort this? 3) Is there a way to extract the data from the garchFit objects? So if "fit" is my object, how can I extract specific data like the fit$coef[1] I can use with other objects? Ultimately, fSeries comes with a really nice plotting function (i.e., plot(fit, which=2)) but I want to extract the series it actually plots from there - is that possible like the hist function? Thanks a lot for your time. Best regards, Jeroen van der Heide ______________________________________________ R-help at stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. All e-mail sent to or from this address will be received or otherwise recorded by Vicis Capital, LLC and is subject to archival, monitoring and/or review, by and/or disclosure to, someone other than the recipient. This message is intended only for the use of the person(s) ("intended recipient") to whom it is addressed. It may contain information that is privileged and confidential. If you are not the intended recipient, please contact the sender as soon as possible and delete the message without reading it or making a copy. Any dissemination, distribution, copying, or other use of this message or any of its content by any person other than the intended recipient is strictly prohibited. Vicis Capital, LLC only transacts business in states where it is properly registered or notice filed, or excluded or exempted from registration or notice filing requirements.