1 True or false? Why? OLS estimator is not more efficient than IV estimator under Assumption MLR.1-MLR.4. 2. Under Assumption MLR.1-MLR.5 (homoskedasticity), WLS estimator has a smaller standard error than OLS estimator. 3. Under Assumption MLR.1-MLR.6 (u ~ N(0,0²)), we can construct valid confidence intervals by FGLS estimator. 4. 2SLS always has a smaller standard error than OLS if E[u|X₁,..., Xk] ≠ 0.