Positive Serial Correlation = Standard Errors are too small Negative Serial Correlation = Standard Errors are too large Does the Hansen Method correct both (while correcting for heteroskedasticity) or is it only useful for positive serial correlation (which can also be corrected by robust standard errors)? If using the Hansen Method is not the proper way to correct for negative serial correlation, what is (GLS rings a bell but I may be way off).

I thought Hansen worked for either. GLS works when you have ARCH.

Figures, I was reading about random walks and ARCH yesterday.

I had to look it up. I am going to be pissed if we get questions like that…

GLS = Generalized Least Squares right?

Yeah.

i think hansen works for both…

thanks mwvt9, ARCH is the only one that I don’t really understand. I hope it won’t show up