Title: The variance-frequency decomposition as an instrument for the identification of SVAR models
Authors: Yuliya Lovcha - Universitat Rovira i Virgili (Spain) [presenting]
Alejandro Perez Laborda - Univeridad Rovira i Virgili (Spain)
Abstract: A framework is proposed to study the identification of structural VAR models. The framework focuses on the contribution of the identified shock to the variance of the variables in the business cycle frequency range. The discussion is organized around the identification of technology shocks, since it has attracted considerable attention. First, we conduct a Monte-Carlo study to analyze within this framework the properties of a set of identification schemes for the technology shock. After that, we propose a new identification method, based on the variance-frequency decomposition, which delivers a reliable estimate of the response of hours. The empirical application of this scheme is illustrated in two datasets. Finally, we show that, aside from its use as a pure identification mechanism, the proposed method may be employed to evaluate the consistency between parameterized models and the data.