Consider the same setup of the previous question, where the model is:

$y_i=x_i+u_i$

and $E(u_i^2)={}_i^2,$ with ${}_i^2={}^2{z}_i^2.$ Obtain the WLS estimator and its variance for the cases:

$($i$){}_i^2=1$

$($ii$){}_i^2=x_i.$

$($iii$).$

$($iv$){}_i^2=x_i^2.$