Title: Stargazing with structural VARs: Shock identification via independent component analysis
Authors: Aleksei Netsunajev - Tallinn University of Technology (Estonia) [presenting]
Dmitry Kulikov - Eesti Pank (Estonia)
Abstract: A new ICA-based statistical identification procedure is introduced for Bayesian SVAR models with independent non-Gaussian structural innovations. Additional statistical information, available in non-normally distributed shocks, allows us to estimate a fully general structural VAR model without use of any strong conventional a priori identifying restrictions. The new procedure is validated using the US macroeconomic data series, where the nature of four different shocks is empirically examined. In particular, we statistically identify the short-run impacts and impulse responses of four structural shocks, which we label a monetary policy shock, a money demand shock, an aggregate demand shock and an aggregate supply shock. We find a robust and well-pronounced price puzzle in response to our monetary policy shock, while the statistically identified money demand shock induces a strong reaction in the US real output and prices. In addition, we also document a new real output puzzle in response to the Federal Reserve monetary policy action.