EViews.com

Hi,

I'm new to EViews 10 and saw that there is a new regression method called "Least Squares with Breakpoints" which splits the regression results based on bai-perron test for structural breaks. I have two questions on this:

1- What happens when one of the regressors is non-significant in only 1 bracket (sub-period), and significant in the other, can we keep the variable in the model?

2-If the variables have unit root, but are cointegrated, will my inferences based on the associated statistics be correct?

If anyone has any documentation on this method please share!

Thanks.

1. EViews doesn't allow you to have a variable in only one regime so the variable will be included, but in both regimes.

2. I don't recall which results go through.

http://www.eviews.com/help/helpintro.ht ... ect_header