Analysing monetary policy's transmission mechanisms through effective and expected interest rates: an application of MS-models, Bayesian VAR and cointegration approaches for Brazil

Abstract

This work applies Markov-switching models , a Bayesian vector autoregressive (BVAR) and Cointegration approach to verify the empirical relationships between expected and effective short-term interest rates in Brazil. The main results corroborate the theoretical notion that the Central Bank can smooth adjustments in effective short-term interest rates, as the latter affect expected short-term rates, thereby influencing long-term interest rates, which are fundamental for controlling output activity and price changes. Moreover, the MS-models show that these empirical relationships are more significant under a ‘higher response regime’. In turn, the BVAR test yields impulse-response functions showing that shocks in expected rates have more persistent impacts on effective rates than the latter have on the former. Finally, the Cointegration and Vector Error Correction approaches are used as a robustness test and confirm the idea of a co-movement between expected and effective interest rates in Brazil. These results support the notion that Brazilian monetary policy is transparent, predictable and efficient.