1

. Econometric results

Clarida, Galí and Gertler (2000) estimate the monetary response function with a forward-looking

specification, where current policy actions depend on expected inflation in the future. As seen in the

previous section, several studies treat expected inflation as a monetary policy target in Brazil. Expected

inflation could also thus reasonably be treated as a variable to consider in computing the central bank

response function. However, according to Sims (1999 and 2001), forward-looking specifications have

a backward-looking equivalent. The following specification is thus adopted as a functional manner of

computing the monetary response function:

SELICt

= b0(st ) + b1(st )INFLA12t–1 + b2(st )TXPIB12t-1

+ σ(St )εt (9)

where SELIC is the annualized base rate of interest set by the Central Bank of Brazil. The other variables

were defined earlier, in section IV.

CEPAL Review N° 135 • December 2021 95

Tito Belchior S. Moreira, Mario Jorge Mendonça and Adolfo Sachsida

Below, we examine the results of monetary response function computation also estimated

using the Markov-switching model. The results are shown in table 2. Taking into account the different

specification tests, the three-regime model was the best fit for the data.21 In effect, application of

the likelihood-ratio (LR) test rejected the null hypothesis of linearity (LR = 241.45,X2(11) = [0.000]** e

X2(12) = [0.000]**

).22 The graphics of smoothed probability that illustrate the chronology of regimes are

presented in annex A2.

Table 3 shows that, whatever the regime, the Selic rate is set taking into account not only inflation

but also output growth. In all cases, the coefficients of the variables are significant, and the signs are

as expected.

According to the Taylor rule, the central bank should raise the interest rate by more than one

unit for a given rise in inflation (or expected inflation), in order to ensure stability of singularity and of

equilibrium. Thus, in keeping with the Taylor rule, monetary policy is active or restrictive if the coefficient

of inflation from equation 9 is equal to or greater than 1, and passive or accommodative if the coefficient

is less than 1 (Woodford, 2003).

Following the Taylor rule, table 3 verifies the existence of two regimes of lesser tolerance to inflation

(regimes 2 and 3), while a third regime of central bank stance on monetary policy is accommodative

(regime 1). It should be noted, however that regime 2 refers only to sporadic moments of monetary policy

in Brazil. Conversely, regime 1 is long-lasting, as it ran from late 2007 to December 2014, and mostly

coincided with the administration of Alexandre Tombini, who headed the Central Bank of Brazil from

December 2010. It is interesting to note that, although the inflation rate has eased on several occasions

since then, it shows a structural uptrend. Only after the second half of 2014 did the Central Bank of

Brazil begin to respond more strongly by gradually raising the Selic rate. Throughout 2015, the central

bank took a tight monetary policy stance, according to the analysis of regime 3. As will be seen in the

following section, the reason why inflation has continued to rise is related to the lack of control on the

part of fiscal policy and also to tariff shocks occurring just after the presidential elections of late 2014.

Table 3

Model MS(3)-AIH(1)

Dependent variable: Selic rate

Regime 1 Regime 2 Regime 3

Constant 0.025 (0.000) 0.029 (0.000) 0.169 (0.008)

INFLA12(-1) 0.955 (0.000) 1.295 (0.000) 1.780 (0.000)

TXPIB12(-1) 0.600 (0.000) 0.651 (0.000) 0.811 (0.000)

Standard deviation 0.013 (0.000) 0.001 (0.000) 0.011 (0.000)

Observations 156

Likelihood 462.137

Source: Prepared by the authors, on the basis of data from the Brazilian Institute of Geography and Statistics (IBGE) and the

Getulio Vargas Foundation.

Note: p-value in brackets.

Before Alexandre Tombini, Henrique Meirelles chaired the Central Bank of Brazil from 2003, with

a restrictive monetary policy that marked a major difference with respect to his predecessor. In addition,

during the term of Meirelles the Treasury’s fiscal policy was compatible with debt sustainability.

21 See note 12.

22 See note 13.

96 CEPAL Review N° 135 • December 2021

Fiscal and monetary policy rules in Brazil: empirical evidence of monetary and fiscal dominance

Table 4 shows the matrix of transition probabilities between states assumed by the monetary rule.

Given that regime 2 occurs only occasionally, for simplicity’s sake, the table shows the probabilities of

transition between regimes 1 and 3. A point to remark is that the probability of switching from state 1 to

state 3, and vice versa, is zero, while the sum of probabilities in each column of the matrix of transition

is less than 1. This raises the question of whether the transition between these two states will not occur

without a shock in the Selic rate.

Table 4

Transition probabilities

Pr(St = 1|St–1= 1)

0.979

(0.000)

Pr(St = 1|St–1= 3) =

0.000

(0.000)

Pr(St = 3|St–1= 1) =

0.000

(0.000)

Pr(St = 1|St–1= 3) =

0.962

(0.000)

Source: Prepared by the authors.

Note: p-value in brackets.

VI. Determination of fiscal and monetary dominance

Tables 1 and 3 showed the parameters estimated for the fiscal and monetary response functions,

respectively, to compute the absolute values of the roots of the Leeper model (1991). As was seen in

section II, characterization of the roots of the system indicates when monetary or fiscal policy behaves

actively or passively. In this context, assuming an intertemporal discount rate β = 0.95, the four situations

presented in the Leeper model (1991) can be identified, once the computed values of α and γ are known

for each of the policy (monetary and fiscal) response functions, considering also the respective regimes.

To determine the period corresponding to each of the four possible combinations of active and

passive policies, we compare the graphs shown in annexes A1 and A2. For each pair of policy rules

(fiscal and monetary) seen in table 5, we observe the intersection between the shaded areas that relate

a given monetary authority function with a given regime of the fiscal authority function.

For example, consider regime 1 of the Treasury response function and regime 3 of the central

bank response function, where γ = 0.000, α = 1.781, considering β = 0.95, which is naturally the same

for all cases. It may be observed that |αβ| = 1.691 and |β-1 – γ | = 1.052. On the basis of the Leeper

model (1991), considering the parameters computed, the results show that in this case both monetary

policy and fiscal policy were active, that is, monetary policy prioritized pursuit of the inflation target, but

fiscal policy did not prioritize pursuit of a primary surplus in alignment with public debt sustainability.

But in what period did this situation occur? Comparison of the shaded areas of annex figure A1.1 with

those of annex figure A2.3 shows that the two policies were active only in fiscal year 2015.

Table 5 helps to explain why the inflation rate continued to rise in 2015, even when the central

bank took an active monetary policy stance. The fact is that, even though monetary policy was restrictive,

fiscal policy also took an active position instead of accommodating by seeking budget equilibrium. This

is an explosive situation, in which agents will demand higher and higher interest to take on government

securities and ever-rising interest rates will drive up expectations of inflation, putting inflation control in

jeopardy. Thus, in 2015 both monetary policy and fiscal policy are seen to be active.

CEPAL Review N° 135 • December 2021 97

Tito Belchior S. Moreira, Mario Jorge Mendonça and Adolfo Sachsida

Table 5

Brazil: definition of policies as active or passive on the basis of Leeper (1991)

Parameters

Central bank response function

Parameters

Treasury response function

Regime 1

α = 0.955

Regime 2

α = 1.295

Regime 3

α = 1.781

Regime 1

γ = 0.000

|αβ| = 0.907

|β-1 – γ| = 1.052

FD

Periods: 2010; 2013 and 2014

|αβ| = 1.225

|β-1 – γ| = 1.052

|αβ| = 1.691

|β-1 – γ| = 1.052

Active fiscal and monetary

policies: 2015

Regime 2

γ = 0.299

|αβ| = 0.907

|β-1 – γ| = 0.721

Passive fiscal and monetary

policies: end-2003, 2004,

2008, 2009, 2011 and 2012

|αβ| = 1.225

|β-1 – γ| = 0.721

MD

|αβ| = 1.691

|β-1 – γ| = 0.721

MD

Periods: most of 2003,

2005–2007

Source: Prepared by the authors, on the basis of E. Leeper, “Equilibria under ‘active’ and ‘passive’ monetary and fiscal policies”,

Journal of Monetary Economics, vol. 27, No. 1, Amsterdam,

Elsevier, 1991.

Note: β = 0.95; FD: fiscal dominance, MD: monetary dominance.

This may explain to some extent why the term of Tombini was marked by an accommodative

position, even when inflation approached the ceiling of the band, indicating that it could “get out of

control”. Should the Central Bank of Brazil have taken a more active stance at that point, increasing

interest rates more steeply?

As noted by Sargent and Wallace (1981), in a situation of loose fiscal policy, the adoption of

tight monetary policy leads to an increase in the money supply and, thus, higher inflation in the future.

The question about how the central bank should administer monetary policy must therefore take into

account how fiscal policy is being conducted. The action of the monetary authority can thus be seriously

compromised if fiscal policy does not act to ensure public debt sustainability, as appears to be the case

in Brazil’s fiscal policy since 2013.

The two policies are also seen to have acted passively at the end of fiscal 2003, at the end of

fiscal 2004 and in the period 2008–2012 (except 2010).

That period is obtained from the intersection

between the shaded areas of the graph for regime 2 with respect to the fiscal policy response function

and of regime 1 with respect to the monetary policy response function. This means that fiscal policy

followed a sustainable path in relation to public debt. In the same period, however, the Central Bank

of Brazil did not respond adequately to the increases in the inflation rate. On the basis of table 5 and

observing the graphs of the chronology of regimes in annexes A1 and A2, it is apparent that there was

fiscal dominance in 2010 and between 2013 and 2014. Monetary dominance obtained for much of

2003 and in the period 2005–2007.

As shown in figure 3, in late 2014 the Brazilian economy began to run primary deficits. This is

strongly characteristic of an active fiscal policy and tighter monetary policy in the same period, since

the Selic rate23 rose from 10.92% in October 2014 to 14.15% in December 2015. The results show

empirical evidence of active conduct of both monetary policy and fiscal policy in 2015.