Multivariate Time Series Analysis

Multivariate Time Series is the result of more than 20 years of teaching courses at both the beginning-graduate level. The main motivation is to provide a broad coverage of the most fundamental aspects of multivariate time series analysis and its applications at an appropriate (sometimes challenging) level. As a consequence, the author makes every attempt to strike a balance between clarity of exposition and mathematical rigor. A very detailed, but approachable overview of both VAR and VARMA models is carefully woven throughout the contents. The text provides an updated coverage of several useful and newly-developed techniques such as methods for analyzing financial time series, cointegration, time-varying models, Bayesian methods, portfolio analysis, and linear dynamical systems, among others. The topics are systematically organized in a progressive manner so as to provide suitable continuity from beginning to end. Examples, exercise sets, and their corresponding solutions are plentiful. Theory is discussed when relevant. To facilitate the reading, the book is self-contained. Apart from assuming some elementary knowledge of calculus and linear algebra, all the more advanced statistical and mathematical concepts used in a given chapter have been previously defined in the text. In the introductory chapter, a brief overview of multivariate random variables and matrices is provided for ease of transition from univariate to multivariate concepts. A companion Web site is available for readers to access the relevant (but limited) R data sets that are used within the text.