This paper contributes to the literature in several ways. Unlike all previous work that estimated the BCB’s reaction function, this study uses real-time economic activity data, that is, data that the monetary authority had available when it made interest rate decisions. According to Qin and Enders (2008), not using real time variables might generate different results in the estimated parameters, because the data are often subject to later revisions.
Secondly, the work uses a smooth transition regression (STR) model to represent the nonlinear response of the BCB in relation to inflation and economic activity. Cukierman and Muscatelli (2008) argue that this class of models is more appropriate for modeling nonlinear reaction functions of the monetary authority.
Thirdly, through the empirical findings, the study analyzes the BCB’s preferences during this period, questioning whether the Central Bank of Brazil can be characterized as inflation or recession avoidance preferences, that is, whether it is more responsive to deviations of inflation from its target or the output from its potential.
Finally, we can address the common critique about the behavior of the monetary authorities under an inflation targeting regime. According some authors inflation targeting might lead to an excessive focus on inflation inducing the monetary authority to disregard the economic activity with the objective to achieve its target on inflation. Therefore, we analyze whether a very concerned and well-succeed central Bank in the reduction and stabilization of inflation like the Central Bank of Brazil has preferences with relative weight to output.
The results reject the linearity hypothesis of the BCB’s reaction function during this period in favor of a nonlinear model. The nonlinear estimated coefficients indicate that the BCB’s reaction function is concave in relation to expected inflation, which, given the New Keynesian economic structure, is consistent with the existence of recession-avoidance preferences (RAP). Furthermore, there is no evidence that the reaction function is nonlinear with respect to the output gap, because the nonlinear coefficient associated with the output gap is not statistically significant.
The fact that the BCB’s reaction function is concave in relation to expected inflation means that it has reacted more strongly in reducing interest rates when expected inflation is below target, than when it is above. Despite this, the total response of the monetary authority to inflation in the nonlinear model indicates that for most of the time, the BCB respected the Taylor principle, strongly responding to changes in the expected inflation.