1. We start from the model Y = X + u for which what we have called hypotheses are fulfilled ideals. Some authors, suspecting the presence of multicollinearity, have transformed the model by dividing each observation of all variables by the corresponding to one of them. a) Show that the random perturbations of the transformed model are heteroscedastic. b) Show that applying an efficient method to this reformulated model leads to some estimators that coincide with the OLS applied to the original model c) Show that the LS applied to the original model are more efficient than the OLS applied to the reformulated model.