Measuring The Dynamic Effects Of Monetary Policy Shocks: A Bayesian Favar Approach With Sign Restriction

We estimate the effects of monetary policy shocks in a Bayesian factor-augmented vector autoregression (BFAVAR). We propose a novel identication strategy of imposing sign restrictions directly on the impulse responses of large sets of variables. The novel feature and key strength of our approach is the additional "bite" due to the differences in factor loadings across sets of time series representing, say, "prices" or "output". Furthermore, our procedure does not require a structural interpretation of the factors themselves or adding observables to the list of factors. We impose the conventional wisdom regarding the responses of prices, monetary aggregates, spreads and interest rates. Our results are robust across different subsamples and avoid anomalies arising in Cholesky identications.