Here’s a leg-up on bootstrap

A decade ago, the authors of this book published Estimation and Inference in Econometrics , which achieved considerable success in defining the then-current state of econometric theory. I recall reviewing it and concluding that it represented an essential and engrossing read for teachers and practitioners but feeling that it was not especially student friendly - the initial chapters were pitched at a dauntingly high level and there were no exercises.

Ten years on is about the time many authors might contemplate a revised edition of such a book. Instead, the authors have given us a completely new one, this time aimed squarely at the graduate student market.

The book starts fairly gently, with a careful extended treatment of the regression model, and it is well endowed with exercises. It is not elementary, and it does not pull its punches in the use of mathematical notation and advanced concepts. However, given its explicit brief as a graduate text, it succeeds admirably and can be recommended for students who have cut their teeth on basic texts such as G.S. Maddala’s Econometrics or Jack Johnston and John DiNardo’s Econometric Methods .

It covers the standard topics in estimation and testing theory, followed by chapters on cross-section and time series methods. As the subject progresses, it becomes increasingly difficult to do equal justice to both areas in a single volume at this level, but this book succeeds more than adequately for a two-semester graduate course. It is up to date, and especially welcome is the extensive treatment of bootstrap and simulation methods. This is probably the first such book to consider at any length this important topic, which is bound to dominate econometric practice in years to come.

Everyone has their pet approach to "intuiting" econometric ideas. I still feel, as I did with the earlier book, that the dimensions we typically deal in are too large for geometric intuitions about regression to be useful. It is better to keep the statistical model at the forefront of the discussion.

But while there are inevitably a few things one might do differently, the authority and clarity of this book is unquestionable. Many will want to adopt it as a course text, and no reading list should omit it.


James Davidson is professor of econometrics, Exeter University.