According to Gauss-Markov theorem
therefore, given the assumptions of the
classical linear regre ssion model, the least
squares estimators in the class of all linear
unbiased estimators, have minimum variance
(i.e. BLUE) (Gujarati 2003, p. 79).
Given that the satisfaction of the assumptions
of the classical linear regression is a
necessary condition for achieving BLUE, it
is therefore worthwhile to test for the satisfaction
of those assumptions because of
the following reasons: