Fiscal response function
Although relatively small compared to what is termed the “central bank response function” there is an
analogous literature concerned with estimated the “fiscal response function” of the National Treasury
(Bohn, 1998; Taylor, 2000; Galí and Perotti, 2003; Thams, 2007, among others).
Accurate estimation of the fiscal response function is important for analysis of macroeconomic
policy for at least two reasons. The first concerns the sustainability of public debt. Here, it is important
to ascertain whether the primary surplus reacts to variations in the public debt to GDP ratio in a manner
that keeps that ratio at sustainable levels (Bohn, 1998). The second reason is that estimation of the
fiscal response function serves to investigate whether fiscal policy pursues some other objective, such
as supporting aggregate demand or helping the monetary authority to control inflation.
Given that we are using the Leeper model to test policy rules for Brazil, it is important to be aware
of the specificities, legal aspects and, above all, the calculation methodologies of fiscal indicators. In
this contest, Carvalho and Feijó (2015)
perform a detailed study of the “below the line” methodologies
used by the economic department of the central bank to calculate public sector financing needs, i.e.
the primary and nominal balances and the net and gross debt of the public sector. In that work, the
authors also describe the main characteristics of the method employed to calculate the fiscal balances
used to renegotiate debt with federal bodies, developed on the basis of Act No. 9496 of 1997, and
the implicit calculation methodology of the fiscal balance established in the Fiscal Responsibility
Act – Complementary Law No. 101 of 4 May 2000.
For the case of Brazil, Mello (2005) estimates the fiscal response function according to various
definitions of “public sector” using monthly data for the period 1995–2004 and observes that in all cases
the primary surplus shows a strong positive response in the event of a rise in public sector net debt.5
Mello also finds that output is weakly but positively correlated with several definitions of primary surplus,
which suggests that Brazilian fiscal policy was non-cyclical or acyclical during this period. Mello (2005)
acknowledges the possibility that structural breaks may occur in the series used and proposes to
address these by working with different subsamples. However, this yields significant variations in relevant
parameters, in particular a weakening of the primary surplus response to public sector net debt after 2002.
In order to address the uncertainty over the regime changes that may have occurred, Mendonça,
Santos and Sachsida (2009) calculate the fiscal response function using the Markov-switching model,
with monthly data from January 1995 to December 2007. Their results strongly suggest that fiscal
policy in Brazil underwent two different regimes after the Real Plan, and that late 2000 is the most likely
moment of transition between the two. The regime after 2000 shows a limited (or even nil) response
by the primary surplus to variations in public sector net debt. By contrast, in the regime prior to 2000
(with greater volatility) the primary surplus showed quite an evident response to variations in public
sector net debt. It was also seen that in both regimes the primary surplus seemed to respond positively
to variations in output and that the government did not appear to have explicitly used fiscal policy to
control inflation in either regime.
4 In general, this method is robust when the starting values are chosen arbitrarily or inefficiently.
5 The broadest concept of the public sector is the “consolidated public sector”, which includes the union, the states, the municipalities
and State enterprises. Mello also works with the primary surplus of the “union” (i.e. the federal public administration) and of the
“regional governments” (i.e. all the state and municipal public administrations taken together).
88 CEPAL Review N° 135 • December 2021
Fiscal and monetary policy rules in Brazil: empirical evidence of monetary and fiscal dominance
1. Empirical analysis
The purpose of econometric analysis of the fiscal response function is to test the hypothesis that the
primary surplus adjusts in response to variations in debt to ensure debt sustainability over the long
term, or that fiscal policy is used as a tool to stabilize output or inflation. Thus, in accordance with the
empirical literature on the subject (Bohn, 1998; Galí and Perotti, 2003; Thams, 2007), we propose to
estimate the fiscal response function using a Markov-switching model, specified as follows:
PRIMt
= b0(st
) + b1 (st )DLSPt–1 + b2 (st
)INFLA12t–1 + b3(st
)TXPIB12t–1 + σ(St )εt (8)
where, in this research, we use monthly data from November 2002 to December 2015. The variables
used in this study (whose graphics appear below) are described as follows:6
PRIM: primary balance of the consolidated public sector, not including 12-month cumulative
exchange-rate appreciation divided by GDP (also 12-month cumulative).7
DLSP: ratio between the monthly value of the consolidated public sector net debt and GDP (12-month
cumulative GDP adjusted by the general price index-domestic supply (IGP-DI)).8
INFLA12: rate of inflation measured by the broad national consumer price index (IPCA) over a
12-month period.9
TXPIB12: real GDP growth rate over the past 12 months.10
Figure 1 shows the evolution of the inflation rate and the real GDP growth rate, both in 12-month
cumulative figures. It also shows the behaviour of economic growth in the Brazilian economy in the
period between November 2002 and December 2015.
Instead of the primary surplus, some studies analysing policies of fiscal or monetary dominance
make use of other variables as fiscal policy tools. Leeper (1991 and 2005) uses direct taxation revenue,
as do Moreira, Souza and Almeida (2007) in a study for Brazil. Meanwhile, Davig and Leeper (2011)
use net government revenues.13 Like Mendonça, Santos and Sachsida (2009), the present work treats
the primary surplus as the most suitable variable to reflect fiscal policy. Leeper’s theoretical model
(1991 and 2005) uses real debt, while the empirical studies generally use the debt to GDP ratio, as
we do here.
A number of remarks are called for before moving on to the econometric results. A positive
relationship is expected between the primary balance, PRIM, and net debt (i.e. that b1 > 0)
, given that
when debt rises it is prudent to increase revenues or the primary surplus in order to ensure a sustainable
debt trajectory.
On the basis of Leeper (1991 and 2005), if fiscal policy is committed to maintain a sustainable
debt trajectory in public debt, the policymaker must act passively, that is, concern itself only with
adjusting the ratio between the primary surplus and GDP to give a strong enough positive response
to a rise in the public debt to GDP ratio to keep that ratio under control. In this context, a passive
fiscal policy, in line with a monetary dominance model, should not respond to other variables such as
inflation or output growth. Fiscal policy should adjust passively to monetary policy decisions. If this
does not happen, it will mean that the fiscal authority is adopting an active stance, in other words, it is
concerned more with controlling aggregate demand to tackle unemployment, for example, than with
controlling the trajectory of public debt. By reducing the ratio between the primary surplus and GDP to
raise economic activity levels, an active fiscal policy generates two unwanted side-effects: inflationary
pressure and a rise in the public debt to GDP ratio. These side-effects undermine the effectiveness of
the central bank’s response function based on the Taylor rule
