Are specifying the 5 IVs in both mean and Variance equations?  What is your hypothesis?

You may look at using the GARCH-BEKK Model

You'd better make it clear if you want single equation or multiple equation GARCH model.  Then we can go ahead with single or multiple structure. Eviews supply with excellent manual for both cases. 

as mentioned, it is not clear what your question is - what are you trying to do with your model - at the moment you have a fairly simple mean equation with a GARCH extension, what is unclear?

Dear Malik

You can use the following book to learn How to use effectively Eviews

Time_Series_Data_Analysis_Using_EViews__Statistics_in_Practice_

Dear Ihtishem,

As mentioned in your preliminary results, you use yearly data (23 observations) to explain the "China-variable" by five other regional variables. In that case, I think that you don't need to add the GARCH equation.

In the small data, the existence of GARCH effet (In Eviews -> correlogram of Residuals Squared) mean that the mean-equation is not well specified. I suggest you use an dynamic relationship :

In Eviews, use for example the "Estimation Command" :

===============================

STEPLS(NVARS=7) CHINA C @ JAPAN(0 TO -2) MALAYSIA(0 TO -2) SINGAPORE(0 TO -2) SOUTH_KOREA(0 TO -2) THAILAND(0 TO -2)

You can control the number of determinant 'NVARS=7' to get better estimation.

Otherwise, If you whant to the volatility relationship you should use the monthly, weekly or daily data if it's possible.

Best regards.

Dear Malik,

I am including a PPT to explain how to model any GARCH type model in Eviews. I hope it will be beneficiary for you. Take care.

Mustafa Ozer

I highly recommend you to study the following textbook written by Brooks (2002).

Brooks, C., 2002, Introductory Econometrics for Finance, 2nd edition, Cambridge University Press.

The above mentioned book is very easy to follow.  Sample instructions and output from   EViews will enable you to implement models themselves and understand how to interpret results

http://rum.prf.jcu.cz/public/mecirova/eng_ekonomka/Chris_Brooks_Introductory_Econometrics_for_Fina.pdf

Good luck

 Thanks amira

Dear Mustaafa Özer 

For estimating Turkish reel effective exchange rate volatility do u think is it more usefull to add garch models  dummy which include 2000q3 2001q1 and 2008q3 crisses ? I think it is Garchx model  also in eviews are we adding it in varriance regressors.

I added conditional varriance graphs of to estimation garch 8 with dummy garc7 without dummy

Dear Yılmaz,

I tried to examine your results. Plot of the conditional variance seem to explain the volatility of ER in Turkey. regarding the your question, You did a very good job additing dummy variables as variance regresoors. But, there are some issues that I would like to be sure: First of did you beli,eve that you correctly specify your mean and variance equation and time series properties of RER. Secondly, have you already checked the stationarity and non-negativity conditions of estimated volatility models. Finally, Did you have significant ARCH and/or GARCH coefficients. I hope my inquires will help you out! In the meantime, if you have any fyrther questions, don't hesitate to contct with me. Good luck!

one excellent document for GARCH Models

I am of the opinion that the  independent variables should be modeled in the variance equation and not the mean equation if you are to check for volatility spillovers, If u are using eviews, the Independent variables have to be displayed as variance regressors in the variance equation section .

It's not clear from your pdf's what you are trying to do. You have contemporaneous country data (stock returns??), which is fine. But this doesn't tell you anything about spillovers (if that's what you want?). In terms of EViews you don't appear to have done anything wrong. However, your biggest problem the number of observations. You cannot estimate a GARCH model with 23 observations - it just won't work - ideally you need several hundred. This explains your strange results (negative values in the variance equation). You need to decide what is your research question and then model appropriately.