EViews.com

Hi there,

I'm completing a project that requires me to identify whether the weekend effect exists in a stock market. I understand why I need to use the GARCH model as opposed to OLS regression or ARCH, but the math and choosing the variables in EViews is beyond my current level of comprehension. I have the closing prices of an index over 8 years, which I have successfully imported into a 5 days a week structure. I am unsure of where to start though. I realise I have essentially gotten nowhere with this, but I have been scouring the library and the internet for resources and even guides.

Assuming I have already identified that the data shows heteroscedasticity, where do I begin with my base data (i.e. closing prices)? Shall I estimate an equation using dummy variables for the 5 days of the week using OLS- or do I go for it using ARMA/AR/MA?
The 'd' variables being dummies for their respective days of the week and the 'a' variables being their respective coefficients to be estimated.

I am up for working things out, but I have been thrown in at the deep end (having never heard of EViews or even econometrics prior to this task!) and am under the constraints of a deadline that is fast approaching. I have tried using the search function of this forum, and have found the guide on dummy variables which I will utilise if that is the way to proceed.

Any help is greatly appreciated,

Thank you!

P.S. I am sorry to ask so much when I have done so little in the way of tangible progress.

My thanks to the moderator who moved the post to the correct section.

Well, I've figured it out. I felt obliged to come back here and post how I did it so that others who come seeking assistance with a similar problem will find help in my thread.

So I got it completely wrong in my original post. With the data imported into a 5-day week structure and dummy variables created, go to Quick>Estimate Equation. A box will pop up, and you should change the drop down selection to ARCH. In the mean equation dependent and regressors box, add in:
There is no need to add anything into the variance regressors box, as the program sorts this out automatically. The output will give you the coefficients for each of the dummy variables as well as coefficients for the variables within the standard GARCH model variance equation. I hope this helps anyone that needs help.

It turns out it was a lot simpler than I was making it out to be. Where I was misunderstanding was that I thought the GARCH model was just the variance section, and that I needed to model the mean section using a different model. In actuality, the GARCH model consists of both a mean section and a variance section.