A COMPANION TO Theoretical Econometrics
Uses of Vector Autoregressive Models
When an adequate model for the DGP of a system of variables has been found it may be used for forecasting and economic analysis. Different tools have been proposed for the latter purpose. For instance, there has been an extensive discussion of how to analyze causal relations between the variables of a system of interest. In this section forecasting VAR processes will be discussed first. Forecasting in more general terms is discussed in Chapter 27 by Stock in this volume. In subsection 5.2 the concept of Granger-causality will be introduced which is based on forecast performance. It has received considerable attention in the theoretical and empirical literature. In subsection 5.3 impulse responses are considered. They may also be regarded as instruments for analyzing causal relations between variables. Finally, forecast error variance decompositions and policy analysis are discussed in subsections 5.4 and 5.5, respectively.
