EViews.com

I love EViews Recursive Regression capability. However, I am attempting to do a Recursive Regression that has a constant number of observations. Let's say you have a time series of 60 observations. The Recursive Regression typically cuts out at least a third of the data and regresses the coefficients based let's say on the first 40 observations. Then, in the second regression it grabs the first 41 observations, etc. I would like it to grab on the second regression from the 2nd to the 41st observations. So, the regression would have a constant number of observations (40 in this case). Is there an easy way to do that. I think this is called Rolling Regression. I have seen much written on the subject within the forums. But, I could not extract any clear actionable information. Thanks in advance for the help.

Try the Roll add-in.

What is the Roll add-in? Does it come in packaged with EViews? How do you get to it?

Go to Add-in menu then Download add-ins.

Thanks much. I'll try it.