Model reduction methods for vector autoregressive processes. (Q1420347)

This monograph, based on the author's dissertation, deals with data based methods for reducing the number of VAR (vector autoregressive process) parameters by imposing parameter restrictions. The author proposes and investigates different strategies that are designated to find individual exclusion restrictions on the VAR coefficients. This is done for different VAR modeling classes that are popular in applied econometrics. The objective of the study is to find out whether these procedures can be used as practical methods for limiting the growth in the number of parameters, which is an unconvenient property of (unrestricted) VAR models. Chapter 2 (Model Reduction in VAR Models) considers a number of statistical model reduction methods for stationary VAR models and investigates their properties by Monte Carlo experiments. Chapter 3 (Model Reduction in Cointegrated VAR Models) deals with model reduction strategies specifically designed for cointegrated VAR models. Chapter 4 (Model Reduction and Structural Analysis) analyzes these procedures in the context of SVAR models (Structural VAR). Finally, Chapter 5 (Empirical Applications) presents two empirical case studies: macroeconomic effects of monetary policy shocks on the U.S. economy, and macroeconomic causes of unemployment in Germany.