Structural vector autoregressions and information in moments beyond the variance

Abstract

This dissertation is concerned with the estimation of the simultaneous interaction in non- Gaussian SVAR models using generalized method of moments (GMM) estimators with higher- order moment conditions. The dissertation contributes to the literature by providing identification results using higher-order moment conditions derived from the assumption of independent structural shocks, by proposing modifications to the GMM estimation procedure to improve the small sample performance in the presence of higher-order moment conditions, and by developing a framework to combine traditional restriction based approaches with data-driven identification and estimation approaches.