2.27 In Example 2.6(b), show that

(a) The bias of the ML estimator is 0 when ξ = u.

(b) At ξ = u, the ML estimator has smaller expected squared error than the UMVU estimator.

[Hint: In (b), note that u − X¯ is always closer to 0 than * n n−1 (u − X¯ ).]