About DCC-garch model...

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

About DCC-garch model...

windseav
Hi, everyone,

I currently run into a problem about DCC-Garch model. I use the package cc-garch and the function dcc.estimation. One of the output of this function is DCC matrix, which shows conditional correlation matrix at every time period you gives. However, I cannot figue out how the function calculate the conditional correlation matrix at the first time period, since there is no data to be conditioned... How do we usually deal with the first period data of GARCH model? Could anyone suggest any paper about the DCC GARCH model, including how to deal with the first time period data?

Thank you all guys in advance!!!!!!  

Reply | Threaded
Open this post in threaded view
|

Re: About DCC-garch model...

Arun.stat
Dear Windseav, I found that it is quite subjective because the effect of initial value will dilute after couple of time periods, hence whatever value you put there never matters. However I found that common practice is to put the unconditional variance/covariance/correlation for the first period. I can recall (sometimes back I probably checked) that fgarch package (or may be something related, I cant recall now) put average sum-square of the observations as the estimated variance for the 1st period.

Thanks,
_____________________________________________________

Arun Kumar Saha, FRM
QUANTITATIVE RISK AND HEDGE CONSULTING SPECIALIST
Visit me at: http://in.linkedin.com/in/ArunFRM
_____________________________________________________
Reply | Threaded
Open this post in threaded view
|

Re: About DCC-garch model...

windseav
Hi Arun, thank you so much for your reply.

I have tried to use cor() function in R to calculate the unconditional correlation matrix of my time series, but it is not the same as the calculated first period Dynamic Conditional Correlation matrix by the function dcc.estimation...I don't know why....

Besides, is this function very slow, because of the optimization methods? I tried 30 time series with 771 time periods, and the function runned for one and half hours under the windows and even cannot get a result...

Thank you!!  
Reply | Threaded
Open this post in threaded view
|

Re: About DCC-garch model...

Marcin P?�ciennik
In reply to this post by windseav
Hi,
I thought that a common practice is just to ommit the first period data
since it does not have much influence on further results / calculations.

Cheers
Marcin



2011/6/7 windseav <[hidden email]>

> Hi, everyone,
>
> I currently run into a problem about DCC-Garch model. I use the package
> cc-garch and the function dcc.estimation. One of the output of this
> function
> is DCC matrix, which shows conditional correlation matrix at every time
> period you gives. However, I cannot figue out how the function calculate
> the
> conditional correlation matrix at the first time period, since there is no
> data to be conditioned... How do we usually deal with the first period data
> of GARCH model? Could anyone suggest any paper about the DCC GARCH model,
> including how to deal with the first time period data?
>
> Thank you all guys in advance!!!!!!
>
>
>
> --
> View this message in context:
> http://r.789695.n4.nabble.com/About-DCC-garch-model-tp3579140p3579140.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> [hidden email] 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.
>

        [[alternative HTML version deleted]]

______________________________________________
[hidden email] 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.
Reply | Threaded
Open this post in threaded view
|

Re: About DCC-garch model...

Arun.stat
Hi Marcin, I do not think you can just ignore the past period's estimate (or I misunderstood your statement?)(M)GARCH estimation is essentially an iterative procedure, therefore you need to have something as the starting value.

Thanks,
_____________________________________________________

Arun Kumar Saha, FRM
QUANTITATIVE RISK AND HEDGE CONSULTING SPECIALIST
Visit me at: http://in.linkedin.com/in/ArunFRM
_____________________________________________________
Reply | Threaded
Open this post in threaded view
|

Re: About DCC-garch model...

windseav
Thank you all guys!!

I have some other questions....For garch model, for example, we have 10 time periods, and the function use MLE to get the parameters based on these 10 time periods. Then, the function calculates covariance matrix at each time period based on the estimated parameters. Is this right??

Moreover, does anyone know that whether the correlation matrix calculated by GARCH model is semidefinite positive?

Thanks a lot!!!