2. (Review) Consider the following regression model y = x + U where a, has length K, E(uz) = 0, and rank (E(r,2)) 2 K. Let Y = (Un)', X = (,...,n)', and Z= (1,...,n)'. Also define Pz = 2(Z'Z)-2', X = PzX, and = PY. (a) Show that the two stage least squares estimator's GMM objective function is (Y Xb) Pz(Y XB) = ( XB)'(Y X) - and that it corresponds to using the efficient weight matrix when the variance covariance matrix of uizi, E(utziz), is conditional homoskedastic, i.e. E(uz) = 0