EViews.com

My data has an unbalanced panel structure with 50 cross-sections (i) and 67 periods (t), totaling 3316 observations. I estimated this equation using the "cross-section SUR" weights option:

y(i,t) = c(1) + c(2)*x(i,t)

The results, including coefficients, standard errors, R-squared, etc., are very different between two "builds" of Eviews 6, namely the 2008 build vs the latest (Apr 2009) build. I want to know which build is correct? I need to make sure this inconsistency is not a new bug created unintentionally by the latest patch. Thanks

SK

It is probably due to one of the fixes we made to panel estimation in March of this year.

Feel free to send your workfile to support@eviews.com if you'd like us to take a closer look, though.