1. Serial Correlation
a. A Durbin Watson statistic could indicate either the presence of “pure” or “impure” autocorrelation:
i. Describe a specific souce of “pure” and “impure” serial correlation.
ii. Which is more detrimental to a model? Why ?
b. Given a model “infected” with first-order autocorrelation:
i. Why are the t-statistics likely to be misleading?
ii. What danger does this pose to the researcher?
c. Given Yt= B0Xt +et and Yt-1 = B0+B1Xt-1+et-1,mathematically derive the Cochrane-Orcutt generalized difference equation whose parameters will be purged of first-order autocorrelation.