Let the regression model between Y and X be given by i i i y = ????x +???? , where i i E(???? |x ) = 0, the form of heteroscedasticity is 2 2 i i i Var(???? |x ) = ???? x , i = 1, 2,....,n, and i j Cov(???? ,???? ) = 0, ????i ???? j. Here, the variables y and x are measured in the deviation form from their mean.

(i) Given a sample of observations on y and x, what is the most efficient estimator of ????? What is its variance?
(ii) What is the OLS estimator of ???? and what is the variance of the OLS estimator?
(iii) Show that the OLS estimator of ???? is never as efficient as the GLS estimator.