Abstract:
This study considers the model of Miranda-Agrippino and Rey (2020) implementing the framework of Plagborg-Moller and Wolf (2021), but with various Bayesian configurations. Specifically, we use a Bayesian Structural Vector Autoregression (SVAR) approach to analyze the causal relationships between our variables. To do this, we utilize Impulse Response Functions and Forecast Error Variance Decompositions. To delve deeper into our analysis, we build a hierarchical model that incorporates both a Stochastic Volatility model and a Markov Switching model. We then proceed to compare the results obtained from these three different model specifications.
