This is a very good book. It consistently reviews econometrics without assuming homoscedasticity. It states theorems very accurately throughout, once in simple form, then more generally, and finally, in some cases, using matrix algebra.
But it fails in two ways, most surprising given these authors' pedigrees: first, the book gets to time-series very late and and achieves very little in time-series. Second, the exercises are such that a student can study the book without doing a single piece of applied econometrics.
A better book (at a slightly higher level) is Jeffrey M. Wooldridge's Introductory Econometrics.