SVAR Identification with High-Frequency Macroeconomic Data

Abstract

Starting from the theoretical observation that the identification problem of SVAR models arises from the presence of contemporaneous dependence among the macroeconomic variables, we propose a new identification method based on nowcasted macroeconomic data for which such contemporaneous relations tend to vanish as the observation frequency increases. We show that by increasing the sampling frequency of the variables is possible to identify all the structural shocks with no further information. An illustrative empirical analysis is presented with a focus on the 1990-2020 monetary policies. Contrary to standard SVAR analysis, our methodology is robust to permutation in the orderings and to the inclusion of additional variables. Our empirical results are in line with the recent findings that interest rate hikes positively impact output and prices, although they do not seem to support the widespread view of a private information content conveyed by Central Bank's interventions.

Keywords: Structural vector autoregressive model, Identification, High-frequency data, Temporal aggregation, Heterogeneous VAR model.

JEL Classification: C10, C32, C51, E52