CGARCH

[bubu]

hej hej
i would be really grateful in case somebody can help me with some hints about how can I plot the long run component and the short run component of a CGARCH.

Thank u all

[trubador]

Once you estimate the CGARCH model, you can save conditional variance and permanent (or long run) component via clicking "Proc/Make GARCH Variance Series". Short run component is simply the difference between conditional variance and the permanent component.

[bubu]

thanks a lot trubador, I succeed to make the conditional variance series, but I do not know how to do the permanent component :(

[trubador]

Unless you enter a name for the "Permanent Component" box, it will not save the series. Please see the picture attached.

[bubu]

unfortunatelly i do not have this screen when select "Make garch variance"...i think it's because of Eviews, I'm using Eviews 5.
Lots of thanks for u help.

[bubu]

trubador pls I need you help in this,
I'm analysing the volatility with 4 garch models and then VECM for long run relations between markets. What i do not understand is the connection between the first part of the article and the second part, I mean I should introduce the short run relations from CGARCH in VEC...or how else I can make connections between these?

Thank you for you time and u'r extreme helpful answers

[trubador]

Could you please cite the article you mention?

[bubu]

I think I manage to attached the file. The name of the article is
"Volatility and Long Term Relations in
Equity Markets: Empirical Evidence
from Germany, Switzerland, and the UK"
by Guidi, Francesco.

thank you so much for your support

[trubador]

It seems two parts are actually independent of each other empirically. Author preferred to analyse the volatility in markets in a univariate (but detailed) framework unlike the returns, where he built a vector error correction model. Of course, a multivariate GARCH analysis would also be helpful in analysing the interaction between the markets (simultaneously in terms of returns and conditional variances), author followed a more complementary approach and decided to analyse the relationship between returns in a longer term framework. So everything seems fine...

[bubu]

thank a lot for everything, it helps a lot
do u think makes sense to introduce the long-run component of CGARCH in VECM?

[trubador]

Joint estimation of conditional variances and returns in a multivariate framework will be better. As long as you are aware of the fact that you are following a two-step approach, it will not be a problem. You should take into account this difference when interpereting your results.

[bubu]

Lots of thanks for your help trubador