Previous econometrics texts have a "Losing sight of the forest for the trees" sort of feel to them. Ruud's text, however, works like the old drill Seargent in the Kipling poem who explained his teaching method as "Firsts I tells 'em what I'ms goings to tells em; then I tells 'em; and then I tells 'em what's I tolds 'em." Ruud does this by first building up the fundamental concept of matrix projection. Then he demonstrates how that can be used to explain Ordinary Least Squares regression. Then he adds onto that all the common assumptions: independent, identically distributed errors; normality of the errors, etc. He builds things up one assumption at a time. And all the while he tells you what he's doing and why the content of each chapter matters and how it is related to what has come before and to what will come afterwards.
But, then--in a master stroke of pedagogy--he tears it all down. He starts taking away, one at a time, all the assumptions like normality that he just spent chapters building up and shows how econometricians deal with matters when they *do* in fact remove the standard assumptions. In this way he can introduce consistent estimators, non-linear regression, latent variables, and so on as what they were historically: practical solutions needed when the assumptions of the classical model fail to hold.
By systematically showing which assumptions imply which results and then showing how to deal with things when a given assumption fails to hold, Ruud's book produces a better econometrician. Too often have previous books left previous readers unable to really understand the art of data analysis, which involves taking a data set, seeing what assumptions can be fairly made about it, and then analyzing it given those fairly made assumptions.
Professor Ruud deserves many plaudits for writing what will surely become the standard text for the next generation of graduate students.
So, if you want more than just a recipe book and actually understand econometrics, read this book!
