Econometrics: Alchemy or Science? : Essays in Econometric Methodology

Econometrics: Alchemy or Science? : Essays in Econometric Methodology

Econometrics: Alchemy or Science? : Essays in Econometric Methodology

Econometrics: Alchemy or Science? : Essays in Econometric Methodology


Since the first edition of this book was published in 1993, David Hendry's work on econometric methodology has become increasingly influential. In this edition he presents a brand new paper which compellingly explains the logic of his general approach to econometric modelling and describes recent major advances in computer-automated modelling, which establish the success of the proposed strategy. Empirical studies of consumers' expenditure and money demands illustrate the methods in action. The breakthrough presented here will make econometric testing much easier.


A new edition entails a sustained interest in the subject — but then methodological disputes usually persist almost indefinitely, so perhaps the signal is rather noisy. Certainly, the seven years since the publication of the first edition of this book have witnessed spectacular advances. Even restricting consideration to parametric studies of discrete time series, major developments have taken place in modelling dynamic systems, in analysing non-stationary data, in understanding economic forecasting, in econometric computing, in methodology, and in computer automation: see, among many others, Doornik and Hendry (1997), Johansen (1995), Clements and Hendry (1999), Doornik (1999), Hendry (1995a), and Hoover and Perez (1999). Given such widely available treatments of the first four topics, I have concentrated on the last two, which anyway are the focus of this book.

The new Epilogue explains the recent advances in computer automation of model selection that justify its title: 'The Success of General-to-Specific Model Selection'. As so often in the history of econometrics, implementing an operational procedure rapidly leads to improvements in performance. Interacting closely with the computer program PcGets has radically altered my views on model selection, and on how successful it can be despite the extant theoretical analyses which present a somewhat unhopeful picture. As chapter 20 discusses, a new door has been opened on an old subject, revealing a gold mine of opportunities. After only a couple of years, such immense progress has been made that I no longer even attempt to beat PcGets in modelling: on both computer-generated and real data, on no occasion have I succeeded — nor to my knowledge has anyone else who has used $β$-test versions. Moreover, this is matched by its outstanding simulation characteristics. Yet the field is in its infancy — to quote from the final chapter:

The prospects for empirical econometric modelling are as exciting today as I have ever seen — the best imaginable note on which to conclude this book.

I am greatly indebted to Andrew Schuller of Oxford University Press for his help and encouragement in undertaking this new edition; to Jason Pearce for ensuring smooth production thereof; to Blackwell Publishers for permission to publish a new

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.