This section estimates the monetary policy rule followed by the Central Bank of Brazil to determine
whether its main policy instrument, the Selic rate, responds to the inflation rate for the period between
November 2002 and December 2015. The same Markov-switching model used hitherto is employed, in
which regime alternance is determined by means of a Markov chain to model possible deviations from
a simple linear response function. As noted earlier, this procedure has the advantage of overcoming
uncertainty over the dates on which changes in the parameters occurred.
The discussion concerning the existence of a rule by which the central bank of the United States
(the Federal Reserve System) guides its monetary policy began with Taylor (1993 and 2000) and led
to the study of how monetary policy can be analysed by means of a response function. Taylor pointed
to a strong relationship between changes in the interest rate set by the Federal Reserve System in
response to variations in price levels and in output in the United States economy. In other words, the
policy instrument, mainly the base rate of interest, has risen in periods of rising inflation. The interest
rate also tends to rise when output is well above its potential.
This procedure commonly adopted by
central banks is aimed at avoiding future rises in the inflation rate.
In the case of Brazil, the literature on the topic computes the response function of the central
bank closely following the Taylor rule or a variant of it. There are many works that study the Taylor rule
for Brazil. Lima, Maka and Mendonça (2007) note that the main differences between these studies have
to do with the econometric methodology used for the estimation and the dependence of the policy rule
on current or expected inflation.
Minella and others (2002) estimate the central bank response function with data from July 1999
to June 2002 and show that the institution responded strongly to expectations of inflation and there
was a high degree of interest-rate smoothing. The authors discovered that neither the output gap nor
exchange-rate variation were statistically significant in the central bank response function. Holland (2005)
found that the central bank adopted an aggressive stance of inflation control starting with the adoption
of the targeting regime.
94 CEPAL Review N° 135
• December 2021
Fiscal and monetary policy rules in Brazil: empirical evidence of monetary and fiscal dominance
Salgado, Garcia and Medeiros (2005) use a threshold autoregressive (TAR) model) to explain the
movement of the nominal interest rate after the adoption of the Real Plan. They conclude that Brazilian
monetary policy underwent two different regimes after the Real Plan. The first was associated with
times of international turbulence, such as the Asian crisis and the Russian crisis, which affected Brazil
through the loss of international reserves. In the second regime, the central bank concerned itself with
the movement of the usual domestic variables.
Policano and Bueno (2006) estimate a policy rule for Brazil using a time-varying parameter (TVP)
model and conclude that, between 1995 and 2005, Brazilian monetary policy may be divided into
two regimes. In the first, associated with a fixed exchange rate, the interest rate responded strongly
to output and to international reserves. In the second, the establishment of the Selic rate was linked
more to inflation-targeting.
Teles and Zaidan (2010) use the state-space model to estimate a forward-looking central bank
response function. This study finds that rigorous central bank control of inflation occurs only from 2003
onward, when inflation expectations converge towards equilibrium.
Lima, Maka and Mendonça (2007) use a Markov model to estimate the central bank response
functions between July 1996 and June 2007. Their results show substantial differences in the conduct
of monetary policy before and after August 1999, which indicates that monetary was policy substantially
affected by the change in the exchange-rate regime with the migration to a currency float.
On the basis of a time-varying vector autoregressive (VAR) model, Balbino, Colla and Teles (2011)
attempt to identify differences in the conduct of monetary policy between the administrations of
Armínio Fraga and Henrique Meirelles. Their results show no significant differences between the two
administrations. Starting in 2003, the interest rate remained higher than was necessary for inflation
convergence. During the term of Armínio Fraga, inflation remained above target, owing to the effect of
exogenous shocks and not conflict with the stabilization rule in the crisis of 2002.
With a similar aim, Moreira, Souza and Ellery (2013) analyse the degree of tolerance to inflation by
the presidents of the Central Bank of Brazil in the period 2001–2012. Their results show that Henrique
Meirelles was the least tolerant to higher inflation, compared with Armínio Fraga. In turn, Alexandre
Tombini adopted a more tolerant stance than Armínio Fraga.
