Topics in Structural VAR Econometrics

This monograph deals with the so-called Structural Vector Autoregressive (SVAR) approach, the most recent development of vector autoregressive econometric modeling. Three different types of models, which encompass all the models used so far in the SVAR applied literature, are analysed using a full-information likelihood-based set up and linear constraints of the more general form. Identification analysis andestimation of these models are carried out using compact formulae coming from an application of some new tools in matrix differential analysis. Using approximation theorems of mathematical statistics, the asymptotic distributions of impulse response and forecast error variance decomposition functions are analytically derived, avoiding the use of bootstrapping and Monte Carlo integration techniques. The monograph also contains a qualitative discussion of the results of an exercise on Italian data and two rats procedures implementing identification, estimation and simulation phases according to the proposed approach