# Such contradictory results motivate Furlani et al

Such contradictory results motivate Furlani et al. (2008) to estimate a reaction function for the Brazilian Central Bank using the Bayesian method. The question to be answered was very direct. Does the Central Bank alter its conduct on monetary policy due to changes in the exchange rate? The reported results by the authors suggest that there is no change in the conduct of monetary policy due to changes in the exchange rate. Besides this, they confirmed the results reported in the literature that inflation targeting regime reacts strongly to the output gap variable.

Empirical methods: the proposed reaction function
The proposed COPOM\'s reaction function for the Brazilian economy is as follow.where i is the variation of the monthly Selic rate in log terms (Selic in log difference); LDESVIO is the variable that represents the deviation of market inflation expectations regarding inflation target for a given LDE225 Supplier t, which was composed monthly in order to compare it to the year inflation rate target; GAPPIB is the output gap in log terms; EXPGAP is the expectation of the output gap also in logarithm; and the ROTADIR variable is a dummy variable, whose goal is to capture the effect of turnover of at least one member of the Board of Governors of Monetary Policy Committee (COPOM in Brazil), with voting decision, in the period studied. An alternative approach is also utilized to measure turnover of COPOM directors. It is represented by ROTAPER, which is the percentage change in the number of directors from the previous meeting. Another important variable in the model is the log difference of the interest rate or its rate of growth change. This is used to make the model consistent with the new macroeconomic consensus described by our set of equations.
The empirical analysis is conducted using vector autoregression (VAR). In particular, structural VAR (SVAR) allows us to establish some degree of precedence among the variables based on the theoretical model result. In addition to the usual tests, we present the impulse-response functions. The format of the system of equations representing the SVAR model follows the one in Dias and Dias (2010).
Inflation rate (IPCA): Represents the change in the price level (CPI), which is the rate of inflation in 12 months (IBGE source), defined as the COPOM official price index of the inflation targeting regime.
Market inflation expectation (EXPIPCA): The variable inflation expectation is collected monthly by the Brazilian Central Bank from the main market players and made available through its Focus Report (Relatório Focus) since 2001.
Nominal interest rate (LSELIC): The nominal interest rate is the government bonds interest rate that is set by the COPOM. It is also known as the Selic rate.
Deviation of the inflation rate expectations (LDESVIO): Calculated as the difference between the inflation expectations of the market in relation to the inflation target for a given period t.
Output gap (GAPPIB): The output gap indicates the difference between current real output and potential output. The current and potential outputs are based on the Industrial Production Index – (IPI). The Hodrick–Prescott (HP) is applied to the current output series to obtain its tendency or trend series. The output gap is then the log of the current minus potential output.
Output gap expectations (EXPGAP): The expectation of the output gap refers to the expectation of the difference between current real output and potential output. The output gap expectation used in our calculation was obtained using the GDP expected growth rate predicted monthly by the market players. However, the growth rate of potential GDP is not available in the Brazilian macroeconomic databases. Thus, this variable was calculated based on the geometric interpolation methodology used by Gordon (2011).
ROTADIR: Dummy variable whose goal is to capture the effect of turnover (replacement or rotation) of at least one member of the COPOM – Board of Governors of the Monetary Policy Committee. So it is one (1) if there is change in the members of COPOM otherwise zero (0).