Model Reduction in Cointegrated VAR Models

Abstract

Cointegrated vector autoregressive models have become a standard modeling tool in applied econometric time series analysis during the last decade. Therefore, in this chapter we explore the possibilities to extend the model selection strategies suggested in Chapter 2 to cointegrated VAR models. We start in Section 3.1 with a brief description of cointegrated VAR models and introduce the related vector error correction model (VECM). Section 3.2 provides a discussion of the modeling sequence usually applied in empirical studies and suggests where statistical model selection can be usefully applied. Based on this discussion we describe in Section 3.3 how different model selection strategies can be applied to overcome the problem of overparameterization. This section also includes different proposals for weak exogeneity tests which may be interpreted as a special case of model reduction. We evaluate the performance of different approaches by means of Monte Carlo simulation experiments in Section 3.4. Finally, Section 3.5 summarizes the results.