Granger Causality and Heteroskedasticity

[Magdalena Go]

Hello!

My name is Magdalena and I have the following problem; I have daily returns of 60 firms from the year 2000 to 2015. My aim is to test for granger causality between them and to see which firms are leaders or followers over different time periods.

I would like to test for Granger Causality with the var and vargranger command in Stata. However, I am not sure how to deal with heteroskedasticity. For normal regression Stata offers various options such as using Newey West standard errors. I am also thinking about applying the fitted arch variance.

Could someone tell me which Stata command I could use to filter out heteroskedasticity when examining Granger causality?

Thank you very much!

[Emad Shehata]

Check here: http://www.statalist.org/forums/foru...dasticity-test

[Magdalena Go]

Hello!

Thank you! But my problem is not only to test for heteroskedasticity but also to filter it out. Do you have some advice on which Stata command I could use to filter out heteroskedasticity when examining Granger causality?

Thanks a lot!

[Emad Shehata]

Put please your data and commands as you did in your analysis to check

[Emad Shehata]

This is Stata manual for VAR

http://www.stata.com/manuals13/tsvar.pdf

[Magdalena Go]

So far, I divided the return series by the ARCH fitted standard deviation to control for heteroscedasticity but I am not sure if this procedure is correct.

forvalues id = 1/60 {
2. arch ret if group==`id', arch(1) garch(1)
3. predict garch, variance
4. replace fittedva_n=garch if group==`id'
5. drop garch

gen fittedsd_n=sqrt(fittedva_n)
gen corret=ret/fittedsd

My data simply contains return series of several firms.

[Binyi Zhang]

Hello!

Thank you! But my problem is not only to test for heteroskedasticity but also to filter it out. Do you have some advice on which Stata command I could use to filter out heteroskedasticity when examining Granger causality?

Thanks a lot!

Hi Magdalena Go
Have you successfully solved the problem? Cuz I am facing the same problem for my research.