garch-bekk

[marta]

HI!
I am working on my thesis about the hedging effectiveness in the european electricity market. It's the first time I work with eviews (version 5) and I have to perform the hedge ratio with different estimation methods: I have already used the OLS estimation (static and dynamic) and now I want to use the multivariate garch ones. Could anyone help me with these models?
In another discussion I have found this code that I am attaching but it doesn't seem to work. an error message displays: "c:\users\windowsxp\desktop\eviewsstime\eex.wf1 not found on disk in "LOAD EEX SPOT.WF1".
I have also attached the file wf1 on where I worked.

[theologos]

Hello,

Open both the file and the code at the same work file and replace: load eex spot.wf1 with : ‘load eex spot.wf1


Regards

[marta]

Hello,

Open both the file and the code at the same work file and replace: load eex spot.wf1 with : ‘load eex spot.wf1


Regards

thank you.
now another error mesage dispays: missing values in @LOGL series at current coefficients at observation 07/02/2002 in "DO_BVGARCH.ML(SHOWPOTS, M=100, C=1E-5)".
What does it mean? And How can I figure out the hedge ratio with these results?
Thanks in advance!

[theologos]

Hi,

Simply re-size your estimation sample. Find within the code

sample s1 07/01/2002 07/04/2008
and replace it by:
sample s1 09/01/2002 07/04/2008


I think now it works!

Regards

[marta]

Hi,

Simply re-size your estimation sample. Find within the code

sample s1 07/01/2002 07/04/2008
and replace it by:
sample s1 09/01/2002 07/04/2008


I think now it works!

Regards

Thanks.
Do you think I have to re-size also sample s0?
and a last question: how can I compute the hedge ratio? h= cov(y1,y2)/var(y2)? is it right?
I need this for my thesis (undergraduate student)

[marta]

Hi,

Simply re-size your estimation sample. Find within the code

sample s1 07/01/2002 07/04/2008
and replace it by:
sample s1 09/01/2002 07/04/2008


I think now it works!

Regards

Thanks.
Do you think I have to re-size also sample s0?
and a last question: how can I compute the hedge ratio? h= cov(y1,y2)/var(y2)? is it right?
I need this for my thesis (undergraduate student)

Now I have adjusted the results: s0 07/01/2002 and s1 07/03/2008 and it seems to work, but now another message displays: "positive or non-negative argument to function expected in "SCALAR LR_PVAL=1 - @CCHSQ(LR,1)"".
What's this again?

[theologos]

Hi,

Correctly you have adjusted the sample, dropping just two observations and not two whole months. You do not need to change the first sample. The new message is associated to the LR statistic for selecting between univariate versus bivariate specifications. It receives a negative value something which is non-expected. Change the starting value for omega(2) from 0 to 0.1. The error disappears.

By inspecting your file I realized that one of the variables involved in your bivariate specification is stationary in levels, why do you calculate the growth rate? Is something like that supported by theory? Have you performed preliminary testing for your variables (eg cointegration, Arch effects, serial correlation)? Are you sure that the specification for the two mean equations fits your data optimally? These are questions that you need to answer before you proceed to the estimation.


Regards