A Re View of "non-stationary Time Series Analysis and Cointegration" by Colin P. Harg Reaves

The book is addressed to both professional economists who be­ lieve in econometrics and graduate students with a taste for time series analysis. It discusses important features of the increasingly popular method of cointegration analysis and non-stationary time series blending the theoretical discussion with detailed implementa­ tions which appear quite helpful to practitioners. Common questions such as how to determine the lag length in a Johansen VAR, or what treatment to give to a second valid cointegratiJig vector are dealt with in this book, which is another volume of the excellent series Advanced Texts in Econometrics, edited by Granger and Mizon. In this volume, Collin Hargreaves gathers ten articles showing major developments in the econometric analysis of long-run relationships and model eval­ uation. The papers discuss in depth the problems involved with, and the new methods related to, the analysis of non-stationary time series and cointegration. The authors \;'110 contribute to the book not only address the technical details but also give a fair dimension of how the subject matter has so profoundly affected recent econometric analysis in general. The first chapter (Towards a Theory of Economic Forecasting) is authored by Michael P. Clements and David F. Hendry, and dis­ cusses recent developments in the theory of economic forecasting us­ ing econometric models. A comprehensive list of sources of forecast errors is analyzed, including paranleter nOll-constancy, estimation uncertainty, variable uncertainty, innovation nncertainty, and model misspecification. The authors also propose a theory of intercept ad­ justment to mitigate these errors and show the potential advantages