EViews.com

Hi.

First of all thank you for letting me being a member of this forum.

I would like to know if we must work with STATIONARY data (making for instance the 1º difference and testing unit roots) whenever we are going to calculate the rolling correlation between two variables with the command : genr corr2=@movcor (x, y,200)


Or the rolling correlation calculations work with raw data, and there is no need for transformation?

Many Thanks

Kind Regards

i have 65 companies and 9 variables. How to choose the rolling window? depending on my objective? thanx
another question is can i perform PCA for panel data? as i tried but no luck. anyone can help?
Thankx