A percepção de incerteza na comunicação do Banco Central do Brasil e seus efeitos na transmissão da política monetária

Abstract

This study aims to investigate the effects of uncertainty perception in the communication of the Central Bank of Brazil on the transmission of monetary policy. Uncertainty variables were constructed using machine learning methods combined with textual and sentiment analysis applied to the minutes of the Monetary Policy Committee (Copom) of the Central Bank of Brazil.In the empirical analysis, a Threshold Vector Autoregression (TVAR) model was employed to examine how the responses of the output gap, inflation, inflation expectations, and the DI x Pré swap differ in reaction to positive shocks in the economy’s benchmark interest rate, considering different highand low-uncertainty regimes. The full sample exhibited non-significant thresholds, ocassioning the division of the data into two subsamples.The results from the first subsample indicated negative responses of the output gap, inflation, inflation expectations, and the DI x Pré swap to interest rate shocks, with stronger and more statistically significant effects in the low-uncertainty regime compared to the highuncertainty regime. The second subsample did not show significant regimes in the TVAR model. Consequently, this subsample was analyzed using a Bayesian Vector Autoregression (BVAR) approach.