Finding:
The correlation between regression errors across time is very close to 1.
ρε(t-1),ε(t) = 1
Indicating the presence of significant positive serial correlation. Positive serial correlation causes the standard errors to be too small, which then causes the t-statistics to be too large (biased upward).
Thursday, March 10, 2011
Subscribe to:
Post Comments (Atom)
0 comments:
Post a Comment