8/2015 Dmitry Kulikov and Aleksei Netšunajev. Identifying shocks in structural VAR models via heteroskedasticity: a Bayesian approach

Working Papers of Eesti Pank 8/2015

This paper contributes to the literature on statistical identification of macroeconomic shocks by proposing a Bayesian VAR with time–varying volatility of the residuals that depends on a hidden Markov process, referred to as an MS-SVAR. With sufficient statistical information in the data and certain identifying conditions on the variance–covariance structure of the innovations, distinct volatility regimes of the reduced form residuals allow all structural SVAR matrices and impulse response functions to be estimated without the need for conventional a priori identifying restrictions. We give mathematical identification conditions and propose a novel combination of the Gibbs sampler and a Bayesian clustering algorithm for the posterior inference on MS-SVAR parameters. The new methodology is applied to US macroeconomic data on output, inflation, real money and policy rates, where the effects of two real and two nominal  shocks are clearly identified.

JEL Code: C11, C32, C54

DOI: 10.23656/25045520/82015/0010

Keywords: Markov switching models, Volatility regimes, Statistical identification, Bayesian inference, Clustering methods, SVAR analysis

Author’s e-mail address: dmitry.kulikov [at] eestipank.ee

in new window

The views expressed are those of the authors and do not necessarily represent the official views of the Bank.