on 21 March 2011
This is definitely a graduate level text and should not be considerd by any stretch as a book for those without a good grounding in linear univarite time series methods. Nor, would it be possible to cover the main chapters 1-12 in a semester - chapters 1-8 maybe!
Nevertheless it is an excellent book, probably the best book covering VAR and VECM models. The early chapters of the book cover the VAR model really well including causality, parameter estimation and impulse response. Then there are excellent chapters on the VECM model and cointegration and estimation, though a lot of other stuff such as martingale differences and brownian motion are added to the mix to complicate the picture.
Chapter 10 is somewhat weak as most real world VAR models will probably be VARX models. More information could be given on linking estimated GLS and 2 and 3 stage LS - this leaves a gap with the traditional approach where it must be assumed that endogenous variables are the basis of all VARX models.
In this regard chapters 10 and 18 need more work as the Kalman filter could be used a lot more effectively to estimate more parsimonous models than the VECM structure.