Skip to content

Seasonal GARCH

6 messages · ihernan at stat.Berkeley.EDU, Spencer Graves, Eric Zivot +1 more

Original
ihernan at stat.Berkeley.EDU
# 
I am trying to use the library(fGarch) and fit a GARCH model but I am
interested in fitting a ARCH for the volatility.
If I use ~GARCH(5,0) then 5 autoregressive parameters are fitted but I am
just interested in the a_{t-5}^2 parameter. Is there any way I could
obtain this model using the function library(fGarch).

Thank you Irma
Spencer Graves
# 
The last I checked, garchFit could not estimate a model with zero 
for either of the garch lag parameters. 

      The expert on current and planned garchFit capabilities is Yohan 
Chalabi, and I've copied him on this reply.  Unless you hear otherwise 
from him, I think it is best to assume that you can fit any garch(i, j) 
model you want as long as both i and j are strictly positive. 

      I'm sorry I couldn't be more helpful. 
      Spencer
ihernan at stat.berkeley.edu wrote:
Spencer Graves
# 
The 'garch' function in the 'tseries' package can estimate a 
garch(0, 5) or garch(5, 0) model. 

      Hope this helps. 
      Spencer
Spencer Graves wrote:
Eric Zivot
# 
You must be careful here. A garch(0,5) model is not identified. If all of the ARCH coefficients are zero then the model is not a conditional heteroskedastic model. Volatility is constant in this case.

****************************************************************
*  Eric Zivot                  			               *
*  Professor and Gary Waterman Distinguished Scholar           *
*  Department of Economics                                     *
*  Box 353330                  email:  ezivot at u.washington.edu *
*  University of Washington    phone:  206-543-6715            *
*  Seattle, WA 98195-3330                                      *                                                           *
*  www:  http://faculty.washington.edu/ezivot                  *
****************************************************************
On Sun, 6 Apr 2008, Spencer Graves wrote:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      chalabi at phys.ethz.ch
    

    

      
      #
    

  


  

      
Hi Irma,

After reading the previous post, I think there is some confusion about
what parameters you want to estimate. So I hope I will not add more
confusion out there.

As far as I understand your question, you want to estimate a GARCH(5,0)
and keep all alpha parameters fixed in the optimisation except
alpha_{t-5}.

Currently in garchFit you can only fix the parameters of the conditional
distribution (include.skew, include.shape), the mean equation
(include.mean) and the delta of an APARCH model (include.delta).

We plan to add the ability to fix any parameters in the optimisation for
a future release.

best regards,
Yohan

 
Currently it is not possible to fix parameters

Spencer Graves <spencer.graves at pdf.com> writes:

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      chalabi at phys.ethz.ch
    

    

      
      #
    

  


  

          

      
      
      
    

        
As far as fGarch is concerned, I refer you to my previous post on this
matter (https://stat.ethz.ch/pipermail/r-sig-finance/2008q1/002241.html).

regards,
Yohan

Spencer Graves <spencer.graves at pdf.com> writes: