EViews.com

hello,
I've been given TSP output for single equation estimation, and I'm duplicating it in EViews 6. The first equation is simple tsls, and I duplicated the coefficient estimates perfectly in EViews. However, the TSP Durbin Watson statistic is .74, and EViews is .69. Any idea what's going on there?

Then there's another equation that is the same as the first equation, but it is an AR(1) error specification, and many of the coefficients are off by ~1 or ~2 percent. Could this be a difference between EViews and TSP computation or am I looking for some other culprit?

Jason

Off the top of my head I have no ideas on the Durbin Watson.

For the AR coefficients, it is either that TSP uses a different calculation method, or that it is simply different starting values. If you use the TSP results as starting values in EViews, does it then give the same answers? Similarly, if you use the EViews answers as starting values in TSP, does it then give the EViews answers?

I used the TSP results as starting values in EViews and it still does not change the results. I only have access to EViews - the TSP output I'm looking at is just a printout of the code and output, I can't run the software. I believe it's TSP 4.5 if that helps.