Markov-switching vector autoregressions. Modelling, statistical inference, and application to business cycle analysis (Q1366796)

This book on Markov-switching vector autoregressions is the author's Ph.D. thesis at the Humboldt University of Berlin. He presents an alternative approach to the analysis of the business cycle as an application of his rather fresh approach in this branch of research. Despite of standard econometric models business cycle analyses proceed along the lines of either nonlinear time series models or by the use of time invariant vector autoregressive models. Using Markov-switching vector autoregressions possible regime shifts and nonlinearities can be modelled more appropriately in the case of dynamic systems. Therefore, the first part of the book gives a comprehensive mathematical and statistical analysis of the Markov-switching vector autoregressive model. First, this class of models is introduced and its basic features are investigated. In addition, the relationships between the Markov-switching VAR model to the time invariant vector autoregressive model as well as against alternative nonlinear time series models are pointed out. These introductory thoughts are formalized in the state-space representation in chapter 2, which is the framework for analyzing the stochastic properties of Markov-switching-VAR processes and for developing statistical techniques for the specification and estimation of this class of models. Chapter 3 extends the work by introducing vector autoregressive moving average representation theorems for VAR models with Markov-switching means or intercepts. The following three chapters consider the statistical analysis of Markov-switching VAR models in the case that the parameters are known. Optimal predictors and an intensive discussion of the filtering and smoothing techniques follow. The main part of the study is devoted to the discussion of parameter estimation for this class of models. Topics include the estimation of parameters by maximum likelihood, model selection and model checking as well as the introduction of a multi-move Gibbs-Sampler for multiple time series subject to regime shifts (chapters 6 to eight). In chapter 9 further technical details of these estimation techniques are analyzed. Generalizations of the Markov-switching VAR model to open dynamic systems, endogenous regime selection and lag distributions of regime shift effects are discussed. The final part of the book applies these techniques and methodology to business cycle analysis. Besides of West Germany this application includes countries like the US, Japan, the UK, Canada and Australia. This book presents a further development in the theory of multiple time series analysis. To understand all the topics addressed in the book, the reader should be familiar with the basics of time series analysis as well as with multiple time series analysis.