Economic Modelling xxx (2017) 1–18

Contents lists available at ScienceDirect

Economic Modelling
journal homepage: www.elsevier.com/locate/econmod

The macroeconomic effects of monetary policy shocks under fiscal rules

constrained by public debt sustainability☆
Marco A.F.H. Cavalcanti a, c, Luciano Vereda b, *, Rebeca de B. Doctors c, Felipe C. Lima c,
Lucas Maynard c
a
Department of Macroeconomic Policies and Studies, Instituto de Pesquisa Econ^omica Aplicada (IPEA), Rio de Janeiro, RJ, Brazil
b
Department of Economics, Universidade Federal Fluminense (UFF), Niteroi, RJ, Brazil
c
Department of Economics, Pontifícia Universidade Catolica do Rio de Janeiro (PUC-Rio), Rio de Janeiro, RJ, Brazil

A R T I C L E I N F O A B S T R A C T

JEL: This paper studies the macroeconomic effects of monetary policy shocks when fiscal rules are constrained to
E32 ensure public debt sustainability. In such an economy, the rise in the interest rate following a monetary policy
E62 shock increases the cost of financing debt, thereby making a fiscal adjustment necessary to guarantee debt sus-
H30 tainability. The analysis is based on a DSGE model developed and calibrated to describe the Brazilian economy,
Keywords: where the effects of the interest rate on public debt service tends to be pronounced. The model incorporates a
Monetary policy detailed public sector capable of intervening in the economy through several channels. Our simulations show that
Fiscal rules the magnitude of the reduction in GDP following a monetary policy shock varies considerably depending on the
Public investment fiscal rule adopted. In particular, there is strong evidence that economic performance worsens when fiscal
Public debt adjustment relies on public investment cuts.
Sacrifice ratio

1. Introduction government debt increases, thereby leading to the adoption of restrictive

It is a well-established fact, both theoretically and empirically, that
under most circumstances a monetary policy shock that raises the interest
rate will decrease output and inflation.1 The transmission mechanisms
usually considered in the literature work through either (i) higher capital
costs, lower discounted present values of assets or a higher opportunity
cost of present consumption, which decrease investment and consump-
tion; (ii) domestic currency appreciation, which leads to lower net ex-
ports and overall demand; or (iii) bank capital and balance sheet effects,
which negatively affect the supply of credit and therefore drive invest-
ment and consumer spending down (Boivin et al. (2011)).
In this paper, we investigate an additional channel arising from the
interaction between monetary and fiscal policy: as the interest rate rises,
fiscal measures that reinforce the contractionary monetary shock. In fact,
we may expect public debt to be affected by a monetary policy shock in at
least two ways. First, the lower output brought about by the interest rate
shock translates into lower tax revenues. For a given level of public
expenditure, this implies a higher public sector primary deficit, which
needs to be financed through higher debt. Second, even if the fiscal policy
rule involves an endogenous response (e.g. higher tax rates or lower
expenditures) aimed at keeping a balanced primary budget, the rise in
interest rates may lead to higher interest payments on public debt, thus
giving rise to a higher public sector nominal deficit and higher debt. In
both cases, under an active monetary-passive fiscal policy regime,2 public
debt stabilization eventually requires a fiscal adjustment response
through either higher tax rates or lower public expenditures, which must

☆
We thank the Editors and an anonymous referee for their comments, which greatly improved the quality of this paper. This paper is the result of a research program sponsored by the
Brazilian government under the Subprogram of Research for National Development (PNPD). The authors would like to thank Instituto de Pesquisa Econ^ omica Aplicada (IPEA) for sup-
porting this research. Luciano Vereda would like to thank IPEA and the Brazilian government for the scholarship granted as part of the aforementioned research program.
* Corresponding author. Rua Prof. Marcos V. F. Reis, s/n - Bloco F, Campus do Gragoata (UFF), Niter oi, RJ, CEP: 24210-201, Brazil.
E-mail address: lvereda@gmail.com (L. Vereda).
1
See, inter alia, Christiano et al. (1999, 2005), Bernanke et al. (2005), Forni and Gambetti (2010) and Barakchian and Crowe (2013). Regarding emerging economies, Mallick and Sousa
(2012) show that a monetary policy contraction has a negative effect on output, whereas the aggregate price level either gradually falls (in the case of India, China, and South Africa) or
exhibits strong persistence (for Brazil and Russia).
2
As defined by Leeper (1991), some policy is said to be active if the corresponding authority is free to choose its actions independently of the public sector budget conditions, leaving to
the passive authority the task of generating enough revenue to balance the budget.

https://doi.org/10.1016/j.econmod.2017.12.010
Received 12 February 2017; Received in revised form 17 October 2017; Accepted 20 December 2017
Available online xxxx
0264-9993/© 2018 Elsevier B.V. All rights reserved.

Please cite this article in press as: Cavalcanti, M.A.F.H., et al., The macroeconomic effects of monetary policy shocks under fiscal rules constrained by
public debt sustainability, Economic Modelling (2017), https://doi.org/10.1016/j.econmod.2017.12.010
M.A.F.H. Cavalcanti et al.
drive output and inflation further down. Fiscal policy is therefore ex-
pected to amplify the initial contractionary effect caused by the monetary
policy shock.
Although the literature on the interaction between monetary and
fiscal policy rules has gained momentum recently,3 the channel
described earlier has not attracted much attention, being the works of
Canzoneri et al. (2006), Valli and Carvalho (2010) and Çebi (2012)
among the few to analyze its implications. Given the lack of studies
on this subject, we aim to contribute to the literature in four main
directions. First, we address this monetary-fiscal policy interaction
within a medium-sized small open economy model featuring a
detailed public sector, capable of intervening in the economy through
several policy instruments: taxation on consumption and income,
public investment, social transfers, public employees’ payroll and
government consumption. This allows us to investigate the effects of
monetary policy shocks under a range of fiscal rules that is much
wider than in previous studies. We are therefore able to identify
which fiscal instrument choices generate larger amplification effects
on output and inflation and lead to higher sacrifice ratios, defined as
the ratio between cumulative output loss and cumulative price
decrease following a monetary shock. Interestingly, some fiscal rules
may adversely affect production costs and cause inflation to rise in
the medium run, generating new rounds of interest rate increases that
may further depress output.
Second, we consider fiscal rules that may differ with respect to both
the timing of the policy response to the public debt and to the velocity
of the fiscal adjustment back to the steady state. We thus seek to inves-
tigate the macroeconomic consequences of, on the one hand, delaying
(anticipating) the fiscal adjustment required for debt stabilization; and,
on the other hand, responding more (less) aggressively in order to pull
the debt back to its equilibrium value more (less) quickly. By varying
both the timing of the response to the debt and the velocity of fiscal
adjustment, we may analyze how these parameters interact to generate
higher or lower sacrifice ratios.
Third, our model explicitly takes into account the presence of public
employment, besides other forms of public spending already considered
in previous studies. As noted by St€ahler and Thomas (2012), the explicit
consideration of public employment can significantly affect the analysis
of fiscal policy, given that the civil servants’ payroll usually accounts for
an important share of government consumption as measured by national
accounts.
Finally, by considering two types of individuals (Ricardian and non-
Ricardian agents) and two production sectors (tradables and non-
tradables), we are also able to infer possible distributional and sectoral
aspects of the interaction between monetary and fiscal policy. In
particular, we may rank each fiscal policy rule according to its effects on
the consumption of each type of individual and on the output of each
sector following a monetary policy shock.
The model is calibrated to represent the Brazilian economy, in which
the amplification mechanism we analyze may be particularly relevant for
a number of reasons. First, despite the undeniable evolution in macro-
economic fundamentals observed in recent years, Brazilian interest rates
remain high when compared to international standards. Second, public
debt as a proportion of GDP is also quite high compared to other
emerging economies.4 Third, there is a very close relationship between
the monetary policy rate and the cost of financing public debt, which
tend to move together (see Appendix 1). As a result, the interaction be-
tween monetary and fiscal policies occurring through the effect of in-
terest rates on public debt seems to be particularly relevant for the
country. As Brazil is significantly affected by developments in the world
3
See e.g. Schmitt-Groh
e and Uribe (2005, 2007), Leith and Wren-Lewis (2000, 2008),
Leeper et al. (2010a), Davig and Leeper (2011), Malik (2013) and Philippopoulos et al.
(2015).
4
See footnote 5.
2
Economic Modelling xxx (2017) 1–18
economy but not the opposite, the small open economy hypothesis seems
quite reasonable.
Regarding the main results of the paper, our simulations suggest that
the effects of monetary shocks in an economy similar to the Brazilian one
depend significantly on the fiscal policy rule adopted. In particular, rules
in which fiscal adjustment relies on public investment reductions to
stabilize the public debt tend to generate higher output losses and sac-
rifice ratios. The conclusion regarding public investment is robust, since
it remains valid even when public capital has a modest impact on total
factor productivity. Our results also indicate that more aggressive fiscal
policy will generally lead to better economic performance following a
monetary policy shock, and that most fiscal rules have clear distribu-
tional consequences, favoring Ricardian relative to non-Ricardian
individuals.
The dangers of cutting public investment to offset an increasing cost
of debt induced by a monetary squeeze is very relevant in the Brazilian
scenario. Actually, Brazil figures among the countries in which this kind
of policy has occasionally been put in place.5 The result that most fiscal
rules hurt non-Ricardian relative to Ricardian individuals is also impor-
tant in the Brazilian context, given that a significant proportion of the
population has limited access to financial services.6
Nevertheless, we must emphasize that fiscal adjustments carried out
by public investment cuts are not an exclusivity of the Brazilian economy.
Indeed, as emphasized by Orair (2016), this type of policy is widely
documented in the literature on developed and emerging economies.
Calderon et al. (2003), for example, show that almost half of fiscal ad-
justments put in place by Latin American countries in the 1990s was
achieved by cutting investments in infrastructure. We can also cite the
highly influential paper of Alesina and Perotti (1997), who identify
several episodes of fiscal adjustments in OECD countries, some of them
relying on tax increases and cuts in public investment. Therefore, we take
as relevant the message that ensuring public debt sustainability by means
of investment cuts may be regarded as one of the worst possible choices
in response to monetary policy shocks. In fact, this result could be added
to the literature regarding the theoretical and empirical importance of
public investment7,8.
We must also note that our analysis assumes perfect credibility by
monetary and fiscal authorities, so that we do not delve into questions
related to uncertainty on future policies, as in, inter alia, Feve and Pie-
trunti (2016) and Lemoine and Linde (2016). We also restrict our analysis
5
Graph 2 in the paper by Orair (2016) shows the behavior of public investment as a
proportion of GDP from the end of 1995 until the end of 2015. There was a decline in
public investment as a proportion of GDP since 2013, a decrease of 1 percentage point.
Throughout the same period, the monetary policy rate rose 7 percentage points (from
January 2013 to December 2015), reaching 10,25% per annum. The debt-to-GDP ratio
increased more than 10 percentage points over the same period, reaching 80% of GDP in
April 2017.
6
Castro et al. (2015) assume that 40% of the population has limited access to credit
markets, being essentially non-Ricardians. Kumar et al. (2005) find that 43 percent of
individuals in Brazil have a bank account, implying that approximately half of the pop-
ulation do not.
7
The strategic role of public investment has been extensively studied in the economic
literature since at least Aschauer (1989) and Barro (1991). Although empirical research
does not undoubtedly prove that public investment boosts economic growth, theoretical
reasons justifying its importance have been extensively discussed. Among them we
highlight the higher output and employment multipliers (especially during recessions), the
possibility of smoothing total investment by counterbalancing private investment down-
turns, the ability to break structural bottlenecks and the possibility of increasing the
productivity of the economy in the medium and long run - especially if the government
invests in infrastructure and human capital.
8
There is also another reason justifying the importance of public investment, which is
quite relevant in the context of emerging economies like Brazil. This reason was high-
lighted by Mallick (2006), who used a small macroeconomic policy-oriented model of the
Indian economy (focusing on the country's experience with the policy changes introduced
in 1991) to carry out optimal control exercises. The paper's main results are (i) traditional
IMF-supported adjustment programs are successful in improving the balance of payments
and controlling the inflation rate, but fail in avoiding a collapse in economic activity, and
(ii) output losses could be significantly reduced if the government boosted public in-
vestments in infrastructure.
M.A.F.H. Cavalcanti et al. Economic Modelling xxx (2017) 1–18

to the case of an active monetary-passive fiscal policy regime. We which is paid by the government. They are also excluded from financial
therefore do not investigate the implications for equilibrium determinacy markets, so that:
or macroeconomic outcomes in general of the transition to alternative

regimes, in particular to a passive monetary-active fiscal policy regime, Pt Ci;tc ¼ 1  τwt Wtc Lci;t þ TRci;t (2)
on which there is a large literature.9
Besides this introduction, the paper contains three sections. Section 2 where Ci;tc
, Lci;t , Wtc and τwt denote the quantity of the final good consumed
describes the model and its calibration. In Section 3 we simulate the by the non-Ricardian worker, the quantity of non-Ricardian labor sup-
macroeconomic impacts of monetary policy shocks under different fiscal plied, the wage received by non-Ricardian workers in exchange for their
policy rules. Section 4 concludes the paper and points to future research labor supply and, finally, the tax rate on labor income, which is fixed by
directions. The paper also contains two appendices. The first one exam- the government. Non-Ricardian individuals supply labor services that are
ines the positive correlation between the basic interest rate and the cost homogeneous10.
of financing public debt in Brazil, while the second (entitled “A note on The preferences of non-Ricardian individuals exhibit habit formation,
calibration”) discusses the rationale behind the value attached to each that is, the utility derived from consumption depends on the difference
parameter. Due to its size, though, we opt to exclude Appendix 2 from the between each individual's consumption and per capita consumption of
paper and make it available as a supplementary material. non-ricardian individuals who take part in labor markets.11 The 1st order
condition that characterizes the optimal choice of Lci;t is given by:
2. The model  σ C

 Cc
  σ L
1  τwt wct Lci;t þ wmt  h t1 1  τwt wct  εLt Lci;t ¼0 (3)
We build a medium-sized dynamic general equilibrium model in the ζc
tradition of Smets and Wouters (2003) and Christiano et al. (2005),
extended to the open-economy case following Dib (2011) and Medina where Ct1 c
=ζc represents per capita consumption of non-Ricardian in-
and Soto (2006). The model includes typical features of emerging dividuals who work as measured at t  1, while wct and wm t stand for real
economies, such as non-Ricardian individuals, price and wage index- non-Ricardian wages and transfers, respectively. Since equation (3) de-
ation and a risk premium paid by domestically issued bonds. The fiscal pends only on variables that are the same for all non-Ricardian in-
policy apparatus is based on Forni et al. (2009), St€ahler and Thomas dividuals who take part in labor markets (except for Lci;t , which is specific
(2012), Castro et al. (2015) and Carvalho and Valli (2011). In this to each individual), it follows that all non-Ricardian individuals in this
framework, the government collects lump-sum and distortionary taxes group supply the same amount of labor.
to finance its expenses. Distortionary taxes are levied on consumer At period 0, the representative Ricardian individual makes his de-
spending, wages, capital income and imports of intermediate goods. cisions trying to maximize the expected value of current and future
Expenditures include public consumption, public investment, social utilities, which is given by:
transfers and wages paid to public employees. Allowing for the exis- 2 1σC  1þσ L 3
tence of public employment is an important feature of the model, since ∞
X 6 Ci;tl  hCt1
l
Lli;t þ Lgi;t 7
payments made to public employees constitute a substantial part of E0 ε
βt Bt 4 ε L
5 (4)
t¼0
1  σC t
1 þ σL
public consumption.

2.1. Individuals where σ C and σ L are positive constants and β is a parameter between
l l
0 and 1. Ci;t and Ct1 represent the i-th individual's consumption level at t
There are two types of individuals, Ricardians and non-Ricardians. and per capita consumption of Ricardian individuals at t  1, while Lli;t
Ricardian individuals supply labor, receive dividends (since they are and Lgi;t stand for the quantities of labor supplied to the private sector and
the ultimate shareholders of firms) and accumulate physical capital. They
the government at t. Each Ricardian individual supplies a special kind of
also have access to financial markets, so that they can smooth con-
labor, which is combined with the varieties provided by the other in-
sumption over time by borrowing and saving. Non-Ricardian individuals
dividuals to yield a basket of Ricardian labor. Ricardian preferences also
do not participate in financial markets, so that they are constrained to
exhibit habit formation and are affected by two shocks, εBt and εLt ; the first
spend all of the income earned in each period. Some of the non-Ricardian
shock affects the general level of utility achieved by the individual, while
individuals have government transfers as their only source of income,
the second one affects only his willingness to work.
while others are allowed to provide homogeneous labor in a competitive
Decisions taken by Ricardian individuals must satisfy the following
market. The measure of the Ricardian population is 1, while the measures
budget constraint:
of the working and non-working non-Ricardian populations are ζc and ζa ,
respectively. The share of Ricardian (non-Ricardian) individuals in the Pbt Bli;t et Pb;* l;*
Bli;t1 et Bl;* ~ t Ii;tl
P
t Bi;t
þ ¼ þ
i;t1
1
population is 1þζ ζ
(1þζ ), where ζ ¼ ζc þ ζa . þ Yi;tl  Ci;tl   Taxli;t (5)
Pt Pt Pt Pt Pt
Each non-Ricardian individual who does not work just lives on a
transfer TRai;t paid by the government. Since this individual does not have The variable Bli;t (Bl;*
i;t ) represents the quantity of one period bonds

any other source of income and is excluded from financial markets, it is issued by the local government (by foreigners) that the representative
true that:

Pt Ci;ta ¼ TRai;t (1) 10

a
where Pt and Ci;t denote the price of the final good and the quantity
consumed by the non-Ricardian individual who does not work, respec-
tively.
Non-Ricardian individuals who work receive a wage Wtc , which is
paid by firms in exchange for the labor they supply, and a transfer TRci;t ,
9
See, inter alia, Leeper (1991), Woodford (2001), Blanchard (2004), von Thadden
(2004), Leith and Wren-Lewis (2000, 2008) and Annicchiarico et al. (2008).
The superscript c comes from the fact that non-Ricardian individuals who work are
constrained to consume only from their current income. We use the superscript l for
variables related to Ricardian individuals because their consumption is determined by the
income they receive throughout their lives (lifetime income). The superscript a is used for
variables related to non-Ricardian individuals who do not work because their consump-
tion is sustained by public transfers or social assistance (hence the superscript a). We also
assume that transfers paid to all non-Ricardian individuals are the same, that is, TRci;t ¼
TRai;t ¼ Wtm for all i and t.
11
The instantaneous utility function that characterizes the preferences of non-Ricardian
individuals who take part of labor markets is almost identical to the one that is embedded
in Ricardian preferences (which is presented below). The only difference is the fact that
non-Ricardian individuals do not supply labor to the public sector.

3
M.A.F.H. Cavalcanti et al.
Ricardian individual holds in his portfolio at t. The market price of the
domestic bond is given by Pbt , while et Pb;*
t gives the market price of the
foreign bond converted to the local currency by means of the nominal
exchange rate et . The product Pbt Bli;t (et Pb;* l;*
t Bi;t ) measures the amount
invested in domestic (foreign) bonds at t, while the sum Bli;t1 þ et Bi;t1
l;*
measures the amount redeemed by the investor at t, period in which the
portfolio of one period bonds bought at t 1 comes due. All terms in the
budget constraint are expressed in real terms, that is, nominal figures are
divided by the price of the final good Pt .
The model follows Dib (2011) in that the price of the foreign bond in
terms of the foreign currency (Pb;* b;*
t ) obeys Pt ¼ θ R* , where Rt denotes
1 *
t t
the (exogenous) international one period gross interest rate and θt rep-
resents the risk premium required by foreign residents to invest in do-
 
e B*
mestic bonds. The risk premium is given by θt ¼ εθt exp  θ Ptt Ytt , where
εθt is a shock reflecting exogenous changes in foreign investors’ risk
appetite. The risk premium also depends on the ratio between the face
value of one period bonds issued abroad and owned by domestic resi-
dents (which is expressed in terms of the domestic currency) and the
nominal value of domestic output.
Since the government collects taxes on consumption, there is a wedge
between the price of the final good Pt and the price that producers
~ t ). Given the consumption tax rate τc , the relationship
actually receive (P t
~t I l
~ t . The term
between the two is such that Pt ¼ ð1 þ τct ÞP
P i;t
(which can be
Pt
written as 1þ1τc Ii;t
l
) denotes the investment in physical capital made by the
t l
i-th Ricardian individual; this figure is expressed in real terms and reflects
the hypothesis that no indirect taxes are paid on purchases of investment
goods, so that the price index of investment goods is the wholesale price
~t (Forni et al. (2009)). The variable Taxl denotes lump sum taxes paid by
P i;t
the i-th Ricardian individual.
The variable Yi;tl denotes real income received by the i-th Ricardian
individual at period t. It is given by:

 
  
Yi;tl ¼ 1  τwt wli;t Lli;t þ wgt Lgt þ 1  τkt rtk zli;t Ki;t1
l
 Ψ zli;t Ki;t1
l
  
þ Divli;t  τkt Ψ zli;t Ki;t1
l
þ Ali;t
The variable Ali;t represents net income coming from an insurance
contract that protects the individual against undesired fluctuations in
total income. In practice, this contract makes all Ricardian individuals
face the same budget constraint, inducing them to make the same de-
cisions regarding consumption, labor and investment. The term
ð1  τwt Þðwli;t Lli;t þ wgt Lgt Þ represents real after-tax labor income, which
comes from labor supplied to the private and public sectors. The real
l
Wi;t
wage wli;t ¼ Pt
is specific to each Ricardian individual (hence the
subscript i), since each of them provides a special kind of labor. The other
g
Wt
real wage wgt ¼ Pt
is fixed as a proportion of the general level of the real
wage paid by the private sector to Ricardian individuals, that is,
Wtg ¼ υgt Wtl . We assume that every Ricardian individual supplies the same
amount of labor to the public sector (Lgi;t ¼ Lgt ) and that the government
freely decides the amount supplied by each Ricardian individual. There is
an implicit simplifying assumption that Ricardian individuals prefer to
work for the public sector, so that they are always willing to offer the
amount of labor the government demands.
Obviously, there is also the assumption that only Ricardian agents
provide labor to the government. Although this hypothesis could seem
ad-hoc at first glance, it is indeed quite reasonable in the Brazilian
context. In fact, the condition of being a civil servant benefit Ricardian
agents with a low probability of losing his/her job and/or experiencing
delays/interruptions in wage payments, allowing easy access to the
financial system. In other words, individuals who provide labor for the
Economic Modelling xxx (2017) 1–18
government are almost always Ricardian, given the availability of good
collaterals – such as a steady flow of income12,13.
The income flowing to Ricardian agents from their investments in
physical and financial capital is also taxed; the tax rate on capital income
is given by τkt . The variable Divli;t represents profits paid by firms; this
variable does not really depend on i because Ricardian individuals have
the same stake in all firms. The term rtk zi;t l l
Ki;t1  Ψðzi;t
l
ÞKi;t1
l
denotes
income received by the i-th Ricardian individual when he/she rents
physical capital. Ricardian individuals decide how much capital to
l
accumulate, the extent to which the available stock is used (zi;t ) and the
l
speed with which this stock grows (that depends on Ii;t ). The term
rtk zi;t
l l
Ki;t1 represents revenues achieved by renting physical capital,
which are equal to the product of the rate of return on capital (rtk ) and the
capital stock actually used by firms (zi;t
l l
Ki;t1 ). There is also a cost arising
l
from an abnormal use of the capital stock, which is paid whenever zi;t is
different from one; this cost is measured by Ψ ðzi;t
l
ÞKi;t1
l
. The evolution of
the capital stock obeys the following equation:
  
Ii;tl
Ki;tl ¼ Ki;t1
l
ð1  δÞ þ Ii;tl 1  S εIt l (7)
Ii;t1
where δ is the depreciation rate. The function S establishes an adjustment
cost, which is switched off in the absence of shocks that raise the cost of
accumulating physical capital or when investment does not change over
time. Investment shocks are captured by εIt ; if there are no investment
Ii;t
shocks, εIt ¼ 1. The ratio l
Ii;t1
is also equal to one when investment in
physical capital does not change over time. These circumstances make
the value of S become zero (i.e. Sð1Þ ¼ 0), allowing investment to be fully
converted into physical capital. There is also the assumption that the first
derivative of S with respect to its argument is zero in steady-state, which
makes the costs of adjusting physical capital depend only on the second
derivative of S. Another assumption is that expenses coming from the sub
(super) utilization of the stock of physical capital cannot be deducted
from the tax basis.
In summary, the representative Ricardian individual must choose the
paths of fBli;t ; Bl;*
i;t ; Ci;t ; Wi;t ; zi;t ; Ii;t ; Ki;t g in order to maximize the value of (4)
l l l l l
(6)
subject to (5), (6) and (7). The variable Wi;tl belongs to the set of variables
chosen by the Ricardian individual, since he/she is the only person
providing labor of type i in the entire economy, and therefore has monopoly
power over this labor type. When setting the value of the wage he/she wants
12
Evidences supporting this line of reasoning can be found in Kumar et al. (2005). Indeed,
the authors state that “… [financial] access can depend on the degree of client information
available, because of the impact on risks and hence costs … In the absence of full information,
especially on poorer persons with limited credit histories, institutions tend to look for proxies
to information on creditworthiness. Such proxy variables could include information on cash
flows (income) which could service loans, or assets (wealth), which could aid loan recovery. It
appears that banks also desire such information for opening an account, due to the need to
service accounts, or even to accept a deposit. Other characteristics of this form could include
position in household (heads versus dependents), gender (males versus females) or employ-
ment characteristics (pg. 29)”. When analyzing the impact of these characteristics on the
probability of having access to different financial services, the authors state: “The results …
indicate that information on such characteristics is very important in determining access to
financial services (pg. 30)”. It follows that the favorable conditions enjoyed by public em-
ployees make it easy for them to access financial markets.
13
The favorable conditions enjoyed by Ricardian individuals were magnified by the
support that the Brazilian government gave to the so-called payroll-deductible loans from
2003 onwards. This type of credit is offered to employees of the formal sector and/or
retirees and pensioners. Interest rates charged on these loans are significantly lower than
those charged on traditional personal loans, with interest and amortizations being dis-
counted directly from wages and/or pensions. The amount of payroll-deducted personal
loans provided by financial institutions grew rapidly since then, reaching almost 37% of
the total amount of nonearmarked loans granted to households in 2017. The share of these
loans directed to civil servants was always above 56% of total payroll-deductible loans,
reaching a peak of 62% in mid-2014. This evidence reinforces the notion that public
employees have good access to financial markets, being essentially Ricardians.
4
M.A.F.H. Cavalcanti et al. Economic Modelling xxx (2017) 1–18

to receive, the representative Ricardian individual takes into account the An important feature of the model is the fact that labor is differenti-
fact that the demand for labor of type i is a decreasing function of Wi;tl . ated along two dimensions: first, Ricardian labor is different from non-
Ricardian labor; second, each Ricardian individual supplies a special
variety of Ricardian labor, which firms mix with other varieties in order
2.2. Firms
to obtain the Ricardian labor basket. It is therefore interesting to discuss
in some detail how these different types of labor are combined to obtain
Firms can be divided into two sectors, wholesale and retail. Each firm of
the labor aggregate employed by intermediate goods firms.
the wholesale sector produces a differentiated intermediate good that is
As previously discussed, non-Ricardian individuals supply labor that
subsequently used by firms of the retail sector to produce the final good.
is homogeneous, so that the quantity of labor provided by the i-th non-
Firms of the final good (or retail) sector operate in a perfect competition
Ricardian individual to the j-th firm of sub-sector s at period t is simply
environment. They apply a technology involving two stages of production:
Lci;j;s;t ¼ Lcj;s;t =ζc , where Lcj;s;t represents total demand for non-Ricardian
in the first stage, firms combine a continuum of intermediate goods to form
two distinct baskets, which are called tradable (T) and non-tradable (NT); in labor coming from the j-th firm. Ricardian individuals, in turn, supply
the second stage, firms combine these two baskets to manufacture the final different types of labor that are demanded by each firm j of sub-sector s
good. A fraction of tradable baskets is exported before the second stage is according to the solution to the following optimization problem:
performed. The final good can be consumed by Ricardian, non-Ricardian 1
minfLi;j;s;t g ∫ 0 Wi;tl Lli;j;s;t di
individuals and the government; they can also be transformed into phys-
ical capital by Ricardian individuals or the government. 2 31þμw
 1þ1μ (10)
Firms in the intermediate goods (or wholesale) sector operate in a 6 1 7
sa 4∫ 0 Lli;j;s;t ¼ Llj;s;t
w
di5
monopolistic competition environment. They belong to two different
sub-sectors: the first one includes firms whose production can be
exported after being combined into a basket by firms of the final goods
sector, while the second one includes firms whose production will not be where Wi;tl is the wage charged by the i-th Ricardian individual at period
shipped abroad after manufacturing. This structure justifies the usage of t, Lli;j;s;t is the quantity of labor supplied by this individual to the j-th firm
the tradable and non-tradable terminology to identify each sub-sector, of sub-sector s at period t and Llj;s;t represents total demand for Ricardian
although the production of the first sub-sector (which is the tradable labor coming from this firm at the same time. The relative demands of
one) is not directly exported. Ricardian and non-Ricardian labor by firm j of sub-sector s at period t
Intermediate goods firms transform capital, imported inputs and a labor come from the solution to the following optimization problem:
aggregate into intermediate goods. Each intermediate good is produced by a
single firm. Capital may be public or private, with public capital being minfLl ;Lc g Wtl Llj;s;t þ Wtc Lcj;s;t
j;s;t j;s;t
exogenously determined by the government. Public capital, which is made (11)

1  1þa 1  1þa
1
available at zero cost to all firms, is assumed to be productivity-enhancing a
sa Lj;s;t 1þa ¼ ðφÞ1þa Llj;s;t
a
þ ð1  φÞ1þa Lcj;s;t
(as in Baxter and King (1993), Leeper et al. (2010b) and Carvalho and Valli
(2011)). The production function employed by intermediate goods firms is
where Wtl is the average wage paid to Ricardian individuals and Wtc is the
given by:
single wage paid to all non-Ricardian individuals who engage in working


 g ηsg ηsK ηsL 1ηsK ηsL activities. The combination of different types of Ricardian labor results in
Yj;s;t ¼ Ast v Lgt Kt1 ~ j;s;t Lj;s;t Qj;s;t
K (8)
the following demand function for the variety supplied by the i-th
Ricardian individual at period t:
where Yj;s;t represents the quantity of the j-th intermediate good pro-
duced by the corresponding firm of sub-sector s (s¼ T; NT), Ast is a pro- !1þμ μw
ductivity shock that is common to all firms of sub-sector s, Kt1 g
is the Wi;tl w

Lli;t ¼ Llt (12)

~ j;s;t is the stock of private
public capital stock available at period t, K Wtl
capital employed by the j-th firm of sub-sector s at period t According to (12), the quantity of labor supplied by the i-th ricardian
~ j;s;t ¼ zl K l
(K t j;s;t1 ), Lj;s;t is the quantity of the labor aggregate employed by individual (Lli;t ) depends on the total amount of ricardian labor demanded
this firm at period t and Qj;s;t denotes the quantity of baskets of imported by intermediate goods firms (Llt ) and on the relative cost of labor of type i,
goods that the j-th firm of sub-sector s uses at period t. The parameter ηsg 2 which is measured by the ratio between the wage charged by the i-th
½0; 1Þ measures by how much public capital enhances the productivity of ricardian individual (Wi;tl ) and the “average” wage charged by ricardian
intermediate goods firms. Public capital can be interpreted as infra-
individuals (Wtl ). Expressions analogous to (12) guide the choice between
structure provided by the government to firms; its quantity evolves ac-
ricardian and non-ricardian labor:
cording to the following law of movement:

 g
g  l 1þa
Ktg ¼ 1  ~δ Kt1
a
þ ~v Lgt Itn (9) Wt
Llt ¼ φ Lt (13)
Wt
Equation (9) implies that it takes n periods for public investment to
turn into capital and become fully productive. This formulation is simpler  c 1þa
Wt a
than the “time to build” approach used, among others, by Leeper et al. Lct ¼ ð1  φÞ Lt (14)
Wt
(2010b) and Mereb and Zilberman (2013), although it yields similar
results in the analysis of fiscal shocks. We also allow labor used by the According to (13), the quantity of Ricardian labor demanded by in-
public sector to affect the dynamics of Ktg , by exerting an amplifying ef- termediate goods firms depends on the total amount of the labor aggre-
fect on government investment, and to affect total factor productivity gate they intend to use (Lt ) and on the relative cost of Ricardian labor,
(see ~vðLgt Þ in (9) and vðLgt Þin (8), respectively).14 which is given by the ratio between the average Ricardian wage (Wtl ) and
the general wage level of the economy (Wt ). The parameter φ measures
the importance of Ricardian labor in the production of the labor aggre-
14
These effects are hard to estimate or calibrate. In most of our simulation exercises in gate Lt used by intermediate goods firms, while the parameter a quan-
Section 3 we ignore such effects by adopting the simplifying assumption that
tifies the degree of substitutability between Ricardian and non-Ricardian
vðLgt Þ ¼ ~vðLgt Þ ¼ 1. We perform a robustness analysis in Sub-section 3.3, though, and suppose
that public employment enhances total factor productivity and public capital accumulation. labor (recall the constraint of problem (11)). Equation (14) can be

5
M.A.F.H. Cavalcanti et al. Economic Modelling xxx (2017) 1–18

interpreted in the same way. Equation (19) takes into account the fact that only Ricardian in-
The price of one unit of the basket of imported goods in the domestic dividuals accumulate physical capital. The stock of public capital at
~ * ),
currency (P ) is the product of the exogenous international price (P period t comes from investments made by the government (Itg ), which are
T;t T;t
the nominal exchange rate (et ) and the gross tax rate on imports (1 þ τ taken as exogenous. Government spending Gt is also an exogenous var-
t ):
iable. The last term of equation (19) takes into account the cost of moving

 *
P 
t ¼ 1 þ τ t et PT;t (15) away from full utilization of private capital. Equation (19) also reflects
the fact that the balance of payments of the domestic economy must be in
We also assume that foreign demand for the basket of tradable goods equilibrium at all periods (that is, its result is always zero).
produced by domestic firms (denoted by YT;t
*
) is given by the following
equation: 2.5. Monetary policy

!1þ*τ* The linearized monetary policy rule is given by:

~ T;t
P τ
*
YT;t ¼ Yt* (16)


 

~ *t
et P b t1 þ ð1  ϕR Þ ϕπ Et π tþp  π t þ ϕY Et Y
b t ¼ ϕR R b tþz þ bε mt
R (20)
In the expression above, P~T;t represents the price of the basket of This rule establishes that the one-period nominal interest rate de-
tradable goods in the domestic currency, Yt* is a measure of “international” pends on an inertial component, on the expected deviation of inflation
economic activity (taken as exogenous) and et P ~* is the foreign producer from a target chosen by monetary authorities and on the expected de-
t
price level (also taken as exogenous) converted to the domestic currency viation of output from its steady-state value, besides an exogenous i.i.d.
monetary policy shock bε t . Subscripts p and z are integer numbers that
m
by the nominal exchange rate et . This formulation, which follows Medina
and Soto (2006) and Dib (2011), reflects the fact that domestic producers may assume any value. When p ¼ q ¼ 0, the one-period nominal interest
are unable to charge different prices at home and abroad. rate reacts to the current deviation between the inflation rate and its
target, and to the contemporaneous deviation between output and its
2.3. Wage and price setting steady-state level. If p > 0 and q > 0 (p < 0 and q < 0), then the mone-
tary policy instrument depends on expected (past) deviations, thus being
Firms in the intermediate goods sector enjoy some degree of market forward-looking (backward-looking).
power, so that they are able to choose prices. We suppose that price ri-
gidity follows the Calvo (1983) hypothesis: at each period, a constant 2.6. Fiscal policy
proportion of intermediate goods firms (given by 1  αsP ; s ¼ T; NT) is
randomly selected to set new prices, while the other firms (a proportion Fiscal authorities manage ten instruments: lump-sum taxes paid by
αsP ) simply follow a rule of thumb by which prices are adjusted according Ricardian individuals (Taxtl ); tax rates on consumption, labor income,
to past inflation, as follows: capital income, and imports of intermediate goods (τct , τwt , τkt and τ
t ,
respectively); government consumption of goods and services (Gt );
 s
P~ t1 γP public investment (Itg ); transfers paid to non-Ricardian individuals
~ j;s;t ¼
P ~ j;s;t1
P (17)
P~ t2 (wm g
t ), public employment (Lt ) and the wage paid to public servants
(wgt ).15
where the ratio P ~t2 represents the gross producer inflation rate
~ t1 =P
The government budget constraint states that:
observed at t  1 and γ sP measures the degree of indexation to past
inflation prevailing among intermediate goods firms of sub-sector s. The Dlt1 Dlt
SPt ¼ Rt1  (21)
solution to the problem faced by intermediate goods firms who can Pt Pt
choose optimal prices at period t is such that all firms choose the same
~opt . As a result, the price level prevailing at sub-sector s evolves
price P where SPt represents the real primary surplus, Dlt (Dlt1 ) is the nom-
s;t
according to the following law of motion: inal value of outstanding public debt at period t (t  1) and Rt1 is
the gross nominal interest rate at period t  116 . If we define dlt
s
  γP (dlt1 ) as the real value of outstanding public debt at period t (t  1),
1
 opt μ1s 1 ~ t1  μs
P
~
P μs
s;t ¼ 1  αsP P~ s;t ~
þ αsP P μs
s;t1 ~ (18) we may write (21) as SPt ¼ Rπt1 l l 
 dt1  dt , where π t ¼ P
Pt
is the gross
Pt2 t t1

inflation rate observed between t  1 and t. The primary surplus SPt

Linearizing (18) around a zero-inflation steady state and combining it is given by:
with the linearized version of the expression for the optimal price P ~ opt
s;t
1  r *


yields two New-Keynesian Phillips curves, one for each sub-sector. SPt ¼ τt et ~pT;t Qt þ τwt wt Lt þwgt Lgt þ τkt rtk zlt Kt1
l
þDivlt þTaxlt þ…
Wage-setting by Ricardian individuals follows the same lines. At each 1þ τc;*
t

period, individuals may either be randomly selected to set new optimal τct
l c a  τct τct
 l
…þ c Ct þCt þCt þ c Gt þ Ψ zlt Kt1 …
wages, with probability αW , or simply adjust their previous wages ac- 1þ τt 1þ τt 1þ τct
cording to past inflation. The solution to the problem of choosing optimal 1 g
wages generates a law of motion analogous to (18); this equation gen- …Gt wgt Lgt ζwmt  I
1þ τct t
erates a wage Phillips Curve after linearization.
(22)
2.4. Resources constraint
15
This wage is managed through υgt .
At the macroeconomic level, the equality between resources and 16
The nominal value of outstanding public debt at period t is given by Dlt ¼ Pbt Blt , where
destinations can be derived from the budget constraints of individuals Pbt is the price of a one-period bond issued by the government at period t; such a bond
(Ricardian and non-Ricardian) and the government. This requirement promises to pay $1 at its redemption date. The variable Blt is the quantity of one-period
generates the following equilibrium condition: bonds purchased by Ricardian individuals at period t. The nominal value of public debt
at period t1 is given by Dlt1 ; this value is given by $1Blt1 , that is, the face value of a

 l
Yt ¼ Ctl þ Ctc þ Cta þ Itl þ Itg þ Gt þ Ψ zlt Kt1 (19) one-period bond ($1) multiplied by the quantity of bonds issued by the government at
period t1 (Blt1 ).

6
M.A.F.H. Cavalcanti et al.
All fiscal policy instruments, except one, evolve in the linearized
model according to an AR(1) process:
b t ¼ ρX X
X b t1 þ bε Xt
dl ; bI g ; b
b t ¼ f Tax g
τ ct ; bτ wt ; bτ kt ; bτ  b b m; w g
ε Xt is an i.i.d.
where X t t Lt ; b t ; Gt ; w t b t g and b
shock.
The remaining instrument is the variable used to balance the budget
and stabilize the public debt. It is implicitly determined by the following
condition in the linearized model:
bs t ¼ ρsbs t1 þ ϕ1xk b
d tk
where bs t and b
d t (b
dt ¼ b
l
d t ) are the primary balance and the public debt
deviations from steady-state at period t, ρs is a parameter between 0 and 1
and ϕ1xk is a positive parameter. Condition (24) states that, given any
shock that decreases output and consequently tax revenues, such as an
interest rate shock, the policy instrument is managed so as to keep the
primary budget balanced (at its steady-state level) during the first k  1
periods. The fiscal reaction may involve higher taxes or lower public
expenditures, depending on the adopted policy instrument. However,
this fiscal tightening does not prevent the public debt from rising, as the
higher interest rates translate into higher debt service, higher nominal
deficit and higher debt. From the k-th period onwards, the primary sur-
plus must therefore gradually rise in order to stabilize the public debt,
which requires a new round of fiscal tightening in response to the higher
debt level.17
For a given value of ρs , the size of the response coefficient ϕ1xk de-
termines how fast the debt is pulled back to its steady state level. We
define the corresponding velocity of debt convergence to the steady state
as m, the number of periods it takes for the public debt deviation from
steady state to reach 1% or less after a monetary policy shock.
Obviously, there are infinite possible combinations of parameter
values in the fiscal reaction function consistent with public debt
stabilization. In our benchmark, we adopt k ¼ 8 and select ϕ1xk (for
each possible fiscal instrument) so that m ¼ 40, i.e. the public debt
deviation from steady state reaches 1% or less 40 periods after a
monetary policy shock. We experiment with different values of k and
ϕ1xk (and therefore m) in order to investigate the consequences of, on
the one hand, delaying (anticipating) the fiscal adjustment required
for debt stabilization; and, on the other hand, responding more (less)
aggressively in order to pull the debt back to its equilibrium value
more (less) quickly.
As a robustness exercise, we also analyze the case in which fiscal
policy does not target a balanced primary budget after a shock. In this
case, the fiscal reaction function is not (24); instead, it is given by:
b t1 þ ϕ2xk b
b t ¼ ~ρX X
X d tk
According to this rule, primary deficits (or surpluses) are allowed in
the first k  1 periods following a shock, after which the fiscal instrument
b t is again managed in response to the public debt so as to guarantee its
X
stability.
2.7. Model calibration and solution
As in St€
ahler and Thomas (2012), the model is calibrated so as to
17
The possibility for fiscal policy instruments to react to fluctuations in primary surplus
and/or public debt is often taken into account in the DSGE literature; see, for example,
Castro et al. (2015), Forni et al. (2009), Leeper et al. (2010a, 2010b), Zubairy (2014) and
Philippopoulos at al. (2015). This possibility is also taken into account in the empirical
literature. Indeed, Jawadi et al. (2014) propose a structural VAR model suited to the BRICs
and in which government spending/revenues may depend on the average cost of financing
public debt.
Economic Modelling xxx (2017) 1–18
reproduce the main macroeconomic aggregates and ratios observed in
Brazil in recent years, which supposedly reflect a steady-state equilib-
rium. Values for some of the model parameters are taken from the
(23) relevant national and international literature; for a review of this
literature, see Cavalcanti and Vereda (2011) or Appendix 2 to this
paper, entitled “A note on calibration”, which is provided as a supple-
mentary material .18
Table 1 presents the basic parameterization used in our simulations.
Table 2 shows the resulting macroeconomic ratios valid in steady-state.
Monetary policy is conducted by means of a rule in which the nominal
interest rate reacts to its first lag, to the current deviation between the
(24) inflation rate and its target, and to the gap between observed output and
its steady-state level (that is, in equation (20) we have p ¼ q ¼ 0).
For fiscal policy simulations, one of the most important parameters of
the model is ηsg , which is the power of public capital in the Cobb-Douglas
production function of intermediate goods firms (see (8)). Leeper et al.
(2010b) propose two alternative values for this parameter, 0.05 and 0.10;
other authors adopt lower values - St€ahler and Thomas (2012), for
example, use 0:015. We fix the basic parameter value at 0.05, but we also
test 0.025 and 0.01 as alternatives.
We assume n ¼ 6 in equation (9), implying that public investment
takes 6 periods to become fully productive; in other words, there is a six-
period lag between public investment and its conversion into public
capital. Given the uncertainty about the shapes of ~vðLgt Þ and vðLgt Þ, our
benchmark simulations are based on the assumption that both functions
are equal to 1 at all feasible values of Lgt ; this means that public
employment does not have any direct impact on either capital accumu-
lation (see (9)) or total factor productivity (see (8)). Sub-section 3.3
shows the results obtained by relaxing this hypothesis, that is, by
allowing public employment to boost public capital accumulation
(through ~vðLgt Þ) or total factor productivity (through vðLgt Þ).
The parameter that measures the persistence of fiscal instruments (ρX ,
see) is specified at 0.89, which implies that fiscal policy shocks have a
half-life of 6 quarters. Given the nature of the shocks considered, which
involve public sector measures with effects that usually last more than
one year, this degree of persistence seems reasonable.
Another parameter that may have a significant impact on the
response of the economy to fiscal shocks is ζ, which controls the size of
the non-Ricardian population. Its value is set at 0.66, which implies
that 40% of the population has limited access to credit markets; this is
precisely the hypothesis adopted in the SAMBA model. In our model,
we must additionally specify the relative sizes of the working and non-
working fractions of the non-Ricardian population. We set the pro-
portion of the working fraction of the non-Ricardian population at
60%.
The process of solving the model follows techniques described in
Smets and Wouters (2003), Christiano et al. (2005) and Medina and Soto
(2006), among others. These techniques are implemented by the
(25)
DYNARE software, which embeds a set of routines developed by re-
searchers from CEPREMAP (Centre pour la Recherche Economique et ses
Applications); they are all executed by the MATLAB software.
3. Simulating the effects of monetary policy shocks
3.1. Benchmark simulation
We first analyze the effects of a monetary policy shock under five
different versions of fiscal rules, as specified below:
 Fiscal rule T: fiscal adjustment met by increasing tax rates on labour
and capital income.
18
This note explains how we calibrated the model's parameters, making it clear that our
values are in line with those found in the international and Brazilian literature. Due to
space limitations, the note is provided as a supplementary text to this paper.
7
M.A.F.H. Cavalcanti et al. Economic Modelling xxx (2017) 1–18

Table 1 Table 1 (continued )

Calibrated parameter values.
ϕπ Sensitivity of the interest rate with respect to the 1.5
ζc Size of non-Ricardian working population 0.4 deviation between output and its steady-state level
ζa Size of non-Ricardian non-working population 0.26 in the monetary policy rule
σC Absolute value of the elasticity of the marginal 1.2 ϕY Sensitivity of the interest rate with respect to the 0.1
utility of consumption with respect to consumption deviation between the inflation rate and its target in
σL Elasticity of the marginal disutility of labor with 2 the monetary policy rule
respect to labor
β Intertemporal discount factor 0.9875
θ Sensitivity of the risk premium with respect to the 1.1
debt/output ratio Table 2
δ Depreciation rate of the private capital stock 0.02 Steady-state ratios for main macroeconomic aggregates.
~δ Depreciation rate of the public capital stock 0.02
ðC þ GÞ=PIB Ratio between total consumption and output 0.83
h Degree of habit formation of individuals in the 0.65 C=PIB Ratio between total private consumption and 0.62
economy output
!1
Measure of the convexity of function S, which 0.15 l Ratio between total Ricardian consumption and 0.35
ν¼ ∂2 S
C =PIB
∂ðargSÞ 2
imposes costs to the adjustment of the capital stock output
SS
ζC =PIB
c
Ratio between total non-Ricardian consumption 0.27
Parameter that characterizes function Ψ , which 0.169
d2 Ψ
2 =dΨ
dzl imposes a cost to the economy whenever the rate of
and output
dðztl Þ t g
SS SS
utilization of private capital departs from its steady- ðG þ wg L Þ=PIB Ratio between total government consumption 0.21
and output
state level
G=PIB Ratio between pure government spending 0.07
αTP Proportion of firms in the tradable goods sector that 0.5
(purchases of goods and services) and output
are unable to choose optimal prices in a given period
g g
γ TP Degree of price indexation prevailing at the tradable 0.469 w L =PIB Ratio between the real value of the payroll of 0.14
public servants and output
sector
I=PIB Ratio between aggregate investment and output 0.18
αNT
P
Proportion of firms in the non-tradable goods sector 0.5
l Ratio between private investment and output 0.16
that are unable to choose optimal prices in a given I =PIB
g
period I =PIB Ratio between public investment and output 0.02
γ NT
P
Degree of price indexation prevailing at the non- 0.469 ðY þ wg L Þ=PIB
g
Ration between total domestic absorption and 1.00
tradable sector output
αW Proportion of Ricardian individuals who are unable 0.6 ~pT Y T =PIB
* Ratio between the real value of exports and 0.10
to choose optimal wages in a given period output
γW Degree of wage indexation prevailing among 0.7 p Ratio between the real value of imports and 0.10
T Q=PIB
Ricardian individuals output
ηTK Exponent of capital in the Cobb-Douglas production 0.425 ðwL þ wg L Þ=PIB
g
Real labor income as a proportion of output 0.55
function used by intermediate goods firms wL=PIB Real labor income paid by the private sector as a 0.41
pertaining to the tradable sector of the economy proportion of output
ηTL Exponent of labor in the Cobb-Douglas production 0.4
wg L =PIB
g
Real labor income paid by the public sector as a 0.14
function used by intermediate goods firms proportion of output
pertaining to the tradable sector of the economy c
wc L =PIB Real labor income received by non-Ricardian 0.14
ηNT
K
Exponent of capital in the Cobb-Douglas production 0.25 workers as a proportion of output
function used by intermediate goods firms l g Real labor income received by Ricardian 0.40
ðwl L þ wg L Þ=PIB
pertaining to the non-tradable sector of the economy workers as a proportion of output
ηNT Exponent of labor in the Cobb-Douglas production 0.65  
L Real capital income as a proportion of output 0.27
function used by intermediate goods firms
Y
1þτc
 wL þ BC PIB
pertaining to the non-tradable sector of the economy  * Excise taxes þ import taxes as a proportion of 0.16
τc ðC þ GÞ þ 1þτ τc;* er ~pT Q
μw Parameter that measures the degree of substitution 0.5 1þτc
output
between the different kinds of labor supplied by
ζwm =PIB Social transfers received by non-Ricardian 0.16
Ricardian individuals individuals as a proportion of output
ηg Sensitivity of total factor productivity of 0.05 c c
τw ðwL þ w L Þ=PIB Taxes levied on labor income as a proportion of 0.12
intermediate goods firms with respect to public output
capital  
Taxes levied on capital income as a proportion 0.06
a Parameter that measures the degree of substitution 5 τk Y
1þτc
 wL þ BC PIB
of output
between the labor aggregates supplied by Ricardian
T=PIB Lump-sum taxes as a proportion of output 0.09
and non-Ricardian individuals T NT
Y =Y Ratio between total productions of the tradable 0.51
γ Parameter that measures the importance of tradables 0.4
and non-tradable sectors
goods in the production function used by firms
K=PIB Capital stock as a proportion of output 2.19
manufacturing the final good
η Parameter that measures the degree of substitution 4
between tradable and non-tradable aggregates in the
production of the final good
φ Parameter that measures the importance of 0.65
Ricardian labor in the labor aggregate used by  Fiscal rule G: fiscal adjustment met by reducing public consumption
intermediate goods firms
μT Parameter that measures the degree of substitution 0.1
of goods and services.
between the various kinds of tradable goods used in  Fiscal rule Ig : fiscal adjustment met by reducing public investment.
the production of the tradable aggregate  Fiscal rule Lg : fiscal adjustment met by reducing public employment.
μNT Parameter that measures the degree of substitution 0.2  Fiscal rule wg : fiscal adjustment met by reducing public wages.
between the various kinds of non-tradable goods
used in the production of the non-tradable aggregate
τ* Sensitivity of the foreign demand for tradable goods 10 In our benchmark, we adopt k ¼ 8 in the fiscal rule and select ϕ1xk (for
manufactured in the domestic economy with respect each possible fiscal instrument) so that m ¼ 40. This means that fiscal
to its relative price policy begins to react to the public debt 8 periods after a monetary shock
ρX Degree of persistence of fiscal policy instruments 0.89
and that it takes 40 periods for the public debt deviation from steady state
ϕR Sensitivity of the interest rate with respect to its own 0.9
lag in the monetary policy rule to reach 1% or less.
Fig. 1 shows how the public debt reacts to the monetary policy shock
under each of these fiscal rules, as well as the path of the active fiscal

8
M.A.F.H. Cavalcanti et al.
instrument under each rule. Under all rules, a monetary policy shock
initially raises the public debt as the higher real interest rate drives debt
service and the public deficit up. Given the fall in tax receipts caused by
lower output and income, the active fiscal instrument under each rule
must be managed so as to keep the primary budget balanced (at its
steady-state level) right after the shock. We thus see that the income tax
rate immediately goes up under fiscal rule T, whereas public consump-
tion, investment, employment and wages go down under each of the
other four rules. As the monetary shock subsides and economic activity
and tax revenues recover, the fiscal instruments move back to their
steady state values. Around the 8th period after the shock, however, a
new round of fiscal tightening begins in response to the higher public
debt. As tax rates increase or public expenditures decrease, the public
Fig. 1. Public debt and fiscal instruments responses to a monetary policy shock.
9
Economic Modelling xxx (2017) 1–18
debt gradually converges back to steady state.
Fig. 2 shows impulse responses to the monetary policy shock under
each fiscal rule for the following variables: GDP, inflation, nominal in-
terest rate, real exchange rate, private investment and the relative output
between the tradable and non-tradable sectors. Under all rules, a mon-
etary policy shock initially drives GDP below steady-state, as consump-
tion, investment and net exports fall. As the effect of the shock subsides,
GDP gradually recovers, moving back towards its steady-state level - and
even overshooting it for a while. However, as a new round of fiscal
tightening begins in response to the higher public debt, GDP goes down
again. It is interesting to note that the magnitude of this reduction in GDP
varies considerably across fiscal rules: under rules G and Lg , GDP falls
M.A.F.H. Cavalcanti et al.
Fig. 2. Responses of main macroeconomic variables to a monetary policy shock.
only slightly and quickly converges back to steady-state; under rules T
and wg , GDP falls considerably more than in the previous cases and
converges back to steady-state at a much lower speed; finally, under rule
Ig , the reduction in GDP is relatively moderate in the medium run, but
very pronounced and persistent in the longer term.
The fact that GDP falls more under rules T and wg than under rules G
and Lg may be partly explained by the relatively more adverse impact
that the former rules impinge on private investment. On the one hand,
rule T requires large increases in taxation on both labour and capital, thus
providing both less resources and less incentives to invest. On the other
hand, rule wg requires public sector wages to decrease in order to pro-
10
Economic Modelling xxx (2017) 1–18
mote the desired fiscal adjustment. This may be interpreted as a tax on
the Ricardian population, who are the only individuals who supply labor
to the public sector. They are also the individuals who save and invest in
the economy, so that investment must go down.
The basic reason for the very different GDP response under rule Ig
relates to the negative effect on total factor productivity of intermediate
goods firms brought about by the reduction in public investment. In fact,
as public capital goes down, total factor productivity in the intermediate
goods sector diminishes, which causes marginal costs to expand and
output to contract. Less profits translate into less dividends distributed to
Ricardian individuals and less resources available for private investment.
Moreover, the upward pressure on costs keeps inflation above its steady-
M.A.F.H. Cavalcanti et al.
state level for a long time, thus driving interest rates also up. As a result,
investment is driven below steady-state for many years, thus leading to
persistently lower output and consumption levels.
The real exchange rate response to the monetary policy shock also
helps to explain differences in the GDP response across fiscal rules. The
monetary policy shock forces the real exchange rate to appreciate
immediately, but soon thereafter (between the 3rd and the 5th period
following the shock) one can notice some overshooting. Rules Ig and T
lead to a sizeable and persistent appreciation of the real exchange rate in
the medium run, which drives net exports down. Under rule Ig , the real
exchange rate remains below its steady-state level even in the very long
run. Under rules G and Lg , though, the real exchange rate shows a slight
depreciation, which provides some stimulus to net exports, thus
contributing to relatively higher GDP levels.
The lower right panel in Fig. 2 shows the ratio between output in the
tradables and non-tradables sectors. In the very short run, relative
output in the tradables sector falls under all fiscal rules, basically as a
result of exchange rate appreciation. In the medium run, performance
in the tradables sector is particularly poor under rule T, again as a result
of real exchange rate appreciation, coupled with higher tax rates on
capital that impose a heavy burden on the cost of capital, thus penal-
izing the more capital-intensive tradables sector. Under rule Lg , relative
output in the tradables sector also falls in the medium run, but for
different reasons. Basically, as public employment falls in order to
stabilize the public debt, Ricardian agents tend to increase their labor
supply to the private sector; this pushes wages downwards, making
labour relatively cheap and benefiting the more labour-intensive non-
tradables sector. In the longer run, as these effects subside, relative
output in the tradables sector recovers under all rules, except rule Ig . As
previously discussed, under this rule public investment must fall in
order to drive the public debt back to its steady state level; this
adversely affects total factor productivity in the intermediate goods
sector, leading marginal costs, inflation and the interest rate to increase
and investment to decrease, thus again penalizing the more capital-
intensive tradables sector.
Inflation falls in the short run in response to the monetary policy
shock, then recovers under all fiscal rules. In the longer run, responses
vary. Under rules T, wg and G, inflation converges rapidly to its steady-
state level, eventually exhibiting a mild oscillation around it. Under
rule Ig , inflation is kept above steady-state for a long time due to the
negative productivity effects arising from lower public investment, as
discussed above. Conversely, under rule Lg , inflation is kept below
steady-state for many years. This is due to the impact of lower public
employment in the labour market: the reduction in public employment
induces Ricardian agents to increase their labor supply to the private
sector, pushing their wages downwards; this causes production costs to
decrease, which leads to lower inflation.
The relatively good performance of GDP and inflation under rule Lg
means that the sacrifice ratio, defined as the ratio between cumulative
output loss and cumulative price decrease following a monetary policy
shock, must be relatively low under this rule. This is confirmed by
Table 3, which shows sacrifice ratios under each rule for four different
time horizons. For all horizons, the Lg rule generates the lowest sacrifice
ratios, as one might expect. Conversely, the worst sacrifice ratios arise
under rules T, wg and Ig , which generate large output losses and smaller
benefits in terms of lower prices, as seen above. Rules Ig and wg are
particularly detrimental to economic performance, due to the negative
effects on aggregate productivity (under rule Ig ) and investment (under
both rules).
Fig. 3 shows cumulative consumption gains (and losses) for the
Ricardian (R) and non-Ricardian (NR) populations following a monetary
policy shock under each fiscal rule. The monetary shock and the ensuing
fiscal tightening have clear distributional consequences, as shown by
Fig. 3. Under rules T , G and Ig , Ricardian consumption falls on impact,
but later recovers and presents gains between 3% and 10% of steady-state
11
Economic Modelling xxx (2017) 1–18
Table 3
Sacrifice ratios under different fiscal rules.
Fiscal Rule Sacrifice ratio
1 year 3 years 5 years 10 years
T 2.25 2.35 3.88 5.64
G 2.11 2.12 2.50 2.65
Ig 2.12 2.47 3.52 10.60
Lg 2.02 1.76 1.78 1.52
wg 2.29 2.84 4.98 7.70
(*) The sacrifice ratio is defined as the ratio between cumulative output loss and cumulative
price decrease following a monetary policy shock.
consumption; whereas non-Ricardian consumption presents either long
run losses (of up to 10% of its steady-state level) or a relatively modest
gain (of less than 2.5% of steady state). Therefore, all rules benefit the
Ricardian population, relative to the non-Ricardian population. This is
basically due to the fact that the fiscal burden required to stabilize the
public debt weighs fully on the non-Ricardian population, who are con-
strained to consume out of current income. Ricardian individuals, on the
other hand, are able to adjust to the expected future fiscal measures
following the monetary shock, shifting consumption and resources across
time.
The wg rule is the only one under which Ricardian consumption falls,
not only in absolute terms, but also relative to non-Ricardian consump-
tion. As discussed above, this rule requires public sector wages to
decrease in order to promote the desired fiscal adjustment, which may be
interpreted as a tax on the Ricardian population, who are the only in-
dividuals who supply labor to the public sector.
One interesting point to note is that rule Lg is the only one that gen-
erates long run consumption gains for both Ricardian and non-Ricardian
individuals. This effect is easy to understand, as public employment
generates no output gains under our basic calibration (recall that we
assume vðLgt Þ and ~vðLgt Þ to equal to 1 at all feasible values of Lgt , thus ruling
out the possibility of direct impacts of public employment on either
capital accumulation or total factor productivity).19 As the reduction in
public employment leads to higher labour supply by Ricardian in-
dividuals to the private sector, we shift resources from nonproductive to
productive activities, thus making it possible to increase overall con-
sumption. Moreover, as Ricardian labour supply increases relative to
non-Ricardian labour in the private sector, Ricardian wages fall relative
to non-Ricardian wages, which provides room for non-Ricardian con-
sumption to also increase.
3.2. Varying k and m
We now consider how the sacrifice ratio changes as we vary both the
timing of the policy response to the public debt (parameter k in the fiscal
rule) and the velocity of the fiscal adjustment back to the steady-state
(parameter m, related to parameter ϕ1xk in the fiscal rule). We thus seek
to investigate the macroeconomic consequences of, on the one hand,
delaying (anticipating) the fiscal adjustment required for debt stabiliza-
tion; and, on the other hand, responding more (less) aggressively in order
to pull the debt back to its equilibrium value more (less) quickly. For
reasons of space, we restrict attention to the three best-performing rules
according to Table 3; we are seeking to uncover recommendations on
how to properly conduct fiscal adjustments, and rules Ig and wg have
already shown to be unsatisfactory.
Fig. 4 presents, for each fiscal rule, indifference areas along which
sacrifice ratios (calculated for a 100-period horizon) are roughly constant
19
This outcome changes when public employment affects public capital accumulation,
for example. Indeed, if the government increases public employment, exerting a positive
influence on public capital, then non-Ricardian consumption may even fall in the medium
and long run. Aiming at saving space, though, we do not follow this thread in the paper.
Simulations pertaining to it can be sent upon request.
M.A.F.H. Cavalcanti et al.
Fig. 3. Cumulative consumption gains following a monetary policy shock.
as we vary both k (horizontal axis) and m (vertical axis). In each panel in
the figure, the red lines delimit an area in which the sacrifice ratio is
relatively high, while the blue lines show an area with relatively low
sacrifice ratios. The upper left panel in the figure, for instance, shows that
under rule T, when k ¼ 8 and m ¼ 32, the sacrifice ratio is in the range
between 5.4 and 5.9, while when k ¼ 8 and m ¼ 45, the sacrifice ratio is
in the range 6.1–6.4, therefore higher.
For all three rules, the basic pattern is the same: for a given value of
k, the sacrifice ratio increases as we increase m. In other words, given
the timing of the policy response to the public debt, the sacrifice ratio is
higher when fiscal policy is less aggressive, thus taking more time to
12
Economic Modelling xxx (2017) 1–18
pull the debt back to its equilibrium level. The policy recommendation
is clear: it is better to promote an aggressive fiscal adjustment, as
opposed to a gradual approach, which would tend to keep investment
and output below steady-state levels for a longer than necessary time
period.
3.3. Sensitivity with respect to the effects of public capital and public
employment on output
In this section, we analyze how our results depend on the assumptions
about the impact of public capital on total factor productivity (see (8))
M.A.F.H. Cavalcanti et al. Economic Modelling xxx (2017) 1–18

by 10% would increase total factor productivity by 0.5%, also with

respect to its steady-state level. We also try ηsg ¼ 0:01 and ηsg ¼ 0:1. In the
first alternative, a 10% increase in the public capital stock in relation to
its steady-state level causes total factor productivity to change by only
0.1%, while in the second case this variation rises to 1%.
The sacrifice ratios which result from this first robustness analysis are
displayed in Table 4. Simulations are performed under our benchmark
case: rule (24), public investment is cut to abide to it, k ¼ 8 and ϕ1xk is
fixed in such a way that m ¼ 40 (meaning that it takes 40 periods for the
deviation of public debt with respect to its steady-state to reach 1% or
less). We can see that sacrifice ratios tend to rise when ηsg increases; this
result is valid regardless of the time horizon (1, 3, 5 or 10 years). The
figures in Table 4 suggest that the conclusion of our benchmark simu-
lation, according to which fiscal adjustments based on public investment
cuts generate relatively high sacrifice ratios (especially in the long run)
remains valid even when public capital exerts a modest impact on the
productivity of intermediate goods firms.
We now consider how the sacrifice ratio changes when we relax the
hypothesis that public employment does not affect total factor produc-
tivity. We try two additional scenarios: in the first one, a 10% increase in
public employment above its steady-state level causes total factor pro-
ductivity to increase by only 0.5%, also above its steady-state level. In the
second one we try a more optimistic elasticity of 0.1 – meaning that a
10% increase drives productivity up by 1%. Results of this second
robustness check are shown in Table 5 . In all cases we assume that fiscal
authorities follow rule (24), public employment is cut to abide to it, k ¼ 8
and ϕ1xk is fixed to induce m ¼ 40. We can see that sacrifice ratios tend to
increase when the link between public employment and total factor
productivity is strengthened; this result is valid regardless the time ho-
rizon (1, 3, 5 or 10 years). Therefore, despite the fact that cutting public
employment brings about the lowest sacrifice ratios under our

Table 4
Sacrifice ratios under different values of ηg.

ηg Sacrifice ratio

1 year 3 years 5 years 10 years

0.01 2.08 2.24 2.93 4.32

0.05 2.12 2.47 3.52 10.60
0.1 2.18 2.90 4.60 16.38

(*) The sacrifice ratio is defined as the ratio between cumulative output loss and cumulative
price decrease following a monetary policy shock.

Table 5
Sacrifice ratios under different values of the elasticity of the TFP of intermediate goods
firms with respect to public employment.

elast. Sacrifice ratio

1 year 3 years 5 years 10 years

0 2.02 1.76 1.78 1.52

0.05 2.54 2.25 3.33 4.26
0.1 3.28 2.81 5.63 8.58

(*) The sacrifice ratio is defined as the ratio between cumulative output loss and cumulative
price decrease following a monetary policy shock.

Fig. 4. Sacrifice ratios under selected fiscal rules and various combinations of Table 6
parameters k (horizontal axis) and m (vertical axis). Sacrifice ratios under different values of the elasticity of public capital with respect to
public employment.

elast. Sacrifice ratio

and the effects of public employment on the same variable (see (8) 1 year 3 years 5 years 10 years
again), as well as its impact on public capital accumulation (see (9)). The
0 2.02 1.76 1.78 1.52
first exercise is to assign different values to the parameter ηsg , that mea- 0.01 2.02 1.78 1.83 1.77
sures by how much public capital enhances the productivity of inter- 0.05 2.02 1.87 2.03 2.83
mediate goods firms. In our benchmark case ηsg ¼ 0:05 (s ¼ T; NT), which 0.1 2.02 1.99 2.31 4.37

means that increasing public capital with respect to its steady-state level (*) The sacrifice ratio is defined as the ratio between cumulative output loss and cumulative
price decrease following a monetary policy shock.

13
M.A.F.H. Cavalcanti et al. Economic Modelling xxx (2017) 1–18

Table 7
Sacrifice ratios under different values of αSP (proportion of intermediate goods firms that is randomly selected to set new prices at period t).

Fiscal Rule Sacrifice ratios

3 years 5 years 10 years

αSP ¼ 0.25 αSP ¼ 0.5 αSP ¼ 0.75 αSP ¼ 0.25 αSP ¼ 0.5 αSP ¼ 0.75 αSP ¼ 0.25 αSP ¼ 0.5 αSP ¼ 0.75
T 1.75 2.35 5.36 3.00 3.88 8.10 4.62 5.64 11.02
G 1.39 2.12 5.23 1.71 2.50 5.97 1.82 2.65 6.41
Ig 1.68 2.47 5.87 2.53 3.52 7.94 7.94 10.60 24.31
Lg 1.19 1.76 4.17 1.25 1.78 3.93 1.08 1.52 3.29
wg 2.08 2.84 6.04 3.94 4.98 9.18 6.34 7.70 13.07

(*) The sacrifice ratio is defined as the ratio between cumulative output loss and cumulative price decrease following a monetary policy shock.

benchmark case, we cannot say that this result is robust. Actually, if
public employment has a sizeable (and positive) effect on total factor
productivity, then sacrifice ratios become quite high.
We close our robustness checks by considering how the sacrifice ratio
changes when we relax the hypothesis that public employment does not
affect public capital accumulation (through ~vðLgt Þ, see (9)). We try three
alternatives besides the benchmark. In the first one, a 10% increase in
public employment with respect to its steady-state level causes public
capital to increase by 0.1%, again with respect to its steady state level,
after 6 periods. In the second scenario, public capital increases by 0.5%.
In the third scenario (which is the most optimistic), the same 10% in-
crease in public employment drives public capital up by 1%.
Results of this final check are shown in Table 6. The benchmark case
assumes there is no positive effect of public employment on public capital
accumulation. Simulations are performed under the assumptions that
fiscal authorities follow rule (24), public employment is cut to abide to it,
k ¼ 8 and ϕ1xk is fixed to induce m ¼ 40. We can see that sacrifice ratios
tend to increase when elasticities go up, being this effect very pro-
nounced in the long run. Apparently, the strength of the link between
public employment and public capital accumulation does have a sizeable
impact on sacrifice ratios.
The results of this subsection point to the need of studying in more
detail how to model and calibrate the impact of the various fiscal policy
instruments on the economy. Nevertheless, one of the main conclusions
of this paper (i.e., that the type of fiscal response matters for the response
of a small open economy to a monetary policy shock) remains valid for a
wide range of specifications of the link between fiscal instruments and
economic activity.
3.4. Sensitivity with respect to different values of the Calvo parameter (αsP )
In this section, we analyze how our results depend on parameters
related to the price adjustment process, namely, the Calvo parameters αTP
and αNT
P . These parameters are important in determining the response of

Fig. 5. Responses of selected variables to a monetary policy shock under rules T and Td .

14
M.A.F.H. Cavalcanti et al.
output to monetary policy shocks and thus could have a bearing on our
previous results. Therefore, it is important to verify whether our main
conclusions depend on specific choices for these parameters.
Table 7 presents the results of this additional robustness exercise, in
which we study the behavior of sacrifice ratios under three different
values of αsP (s ¼ T; NT). Besides our benchmark value (0.5, which im-
plies that a firm changes its price twice a year, on average) we consider
two additional values: 0.25 and 0.75, which make the average lifetime of
prices chosen by firms to be slightly above one quarter and exactly one
year, respectively. Simulations are performed under our benchmark case:
rule (24), k ¼ 8 and ϕ1xk fixed in such a way that m ¼ 40(meaning that it
takes 40 periods for the deviation of public debt with respect to its
steady-state value to reach 1% or less). The exercise is performed under
each of the five fiscal rules previously considered: T, G , Ig , Lg and wg .
We can see that, for all fiscal rules, sacrifice ratios tend to increase
with the value of αsP . For example, under fiscal rule G, in which gov-
ernment consumption on goods and services is reduced in order to bring
public debt back to its steady-state value after 40 periods, the sacrifice
ratios (for a 10-year horizon) are 1.82 when αsP ¼ 0:25, 2.65 when αsP ¼
0:5 and 6.41 when αsP ¼ 0:75. Such behavior does not come as a surprise,
since the real effects of monetary policy shocks tend to be more pro-
nounced under higher degrees of nominal price rigidity.
The important point to note, however, is that the ranking of our five
fiscal rules in terms of sacrifice ratios is unaffected relative to our pre-
vious results; in particular, we confirm the results presented in our
benchmark simulation, Table 3, according to which the worst sacrifice
ratios arise under rules T, wg and Ig .
Fig. 6. Responses of selected variables to a monetary policy shock under rules G and Gd .
15
Economic Modelling xxx (2017) 1–18
3.5. Fiscal rules without a balanced primary budget
As another robustness exercise, we analyze the case in which fiscal
policy does not target a balanced primary budget after a shock. In this
case, the fiscal reaction function is (25) instead of (24): primary deficits
(or surpluses) are allowed in the first k  1 periods following a shock,
after which the fiscal instrument is again managed in response to the
public debt so as to guarantee its stability.
For reasons of space, we mainly focus on the implications of moving
from rule (24) to rule (25) when the active fiscal instrument used to
stabilize the public debt is either the income tax rate or public con-
sumption. When the active fiscal instrument is the income tax rate, we
label rule (24) as T and rule (25) as Td ; when the active fiscal instrument
is public consumption, these rules are labeled G and Gd , respectively.
Fig. 5 shows impulse responses to the monetary policy shock under
rules T and Td for the following variables: GDP, inflation, public debt
and the fiscal instrument. Fig. 6 depicts the same impulse response
functions for rules G and Gd . As expected, under rules Td and Gd the
public debt rises more than under rules T and G. In fact, as the former
rules allow a primary deficit to arise in response to the monetary policy
shock, there is now an additional factor leading to the increase in debt
(besides higher debt service payments, which was the only factor
causing the debt to rise under rules T and G). This implies that the fiscal
adjustment required to stabilize the debt is necessarily larger, as we can
see from the bottom right panels in Figs. 5 and 6: the tax rate must rise
even more, and public consumption must decrease even more, under
rules Td and Gd than under rules T and G, respectively. As a result, the
fall in GDP is larger under rules Td and Gd relative to rules T and G in the
M.A.F.H. Cavalcanti et al. Economic Modelling xxx (2017) 1–18

medium run. This partially compensates for the fact that the initial fall
in GDP is larger under rules T and G as a result of the fiscal tightening
required to keep a balanced primary budget. Cumulative (undis-
counted) output losses 100 periods after the shock are therefore rela-
tively similar: under rules T and Td , this loss amounts to 11% and 12%
of steady-state GDP respectively; under rules G and Gd , losses are 2.2%
and 1.6% respectively.
With regard to inflation, rules T and Td lead again to very similar
results in terms of cumulative price decreases after 100 periods. As a
consequence, the corresponding sacrifice ratio under both rules are very
similar. Rule Gd , however, leads to a small price decrease (0:05%)
relative to rule G (0:24%). Consequently, despite the smaller output
loss under rule Gd , this rule delivers a higher sacrifice ratio.
4. Conclusions
In this paper we quantified and compared the macroeconomic effects
of monetary policy shocks under different fiscal policy rules. We analyzed
the case in which the rise in the interest rate following a monetary policy
shock increases public debt interest payments, thereby making a fiscal
adjustment necessary to guarantee debt sustainability. The analysis was
based on a medium-sized DSGE model developed and calibrated to
represent the Brazilian economy, in which the effect of the interest rate
on the public debt service tends to be pronounced. The model in-
corporates a detailed public sector capable of intervening in the economy
through several channels, which allowed us to investigate the impact of
various fiscal policy rules on the size of the sacrifice ratio generated by
the monetary policy shock.
Our simulation results show that the magnitude of the reduction in
GDP following a monetary policy shock varies considerably depending
on the fiscal rule adopted. In particular, rules in which fiscal adjustment
relies on public investment reductions to stabilize the public debt tend to
generate higher output losses and sacrifice ratios. The use of public in-
vestment as a fiscal adjustment tool is particularly detrimental to eco-
nomic performance, as it not only generates the largest output losses, but
may also cause inflation to rise as it leads to lower total factor produc-
tivity in the private sector and higher production costs. As a result, sac-
rifice ratios under this rule are particularly perverse. This conclusion
remains valid even when the strength of the link between public capital
and firms’ productivity is relatively modest.
The fiscal rules that generate the lowest sacrifice ratios adopt public
employment as the fiscal adjustment tool. These rules not only lead to
relatively low output losses, but also to relatively low inflation, due to the
impact of lower public employment in the labour market: the reduction
in public employment induces Ricardian agents to increase their labor
supply to the private sector, pushing their wages downwards; this causes
production costs and prices to decrease. Nevertheless, this conclusion
depends crucially on the assumption that public employment does not
affect total factor productivity or public capital accumulation. Indeed, we
have shown that sacrifice ratios can increase dramatically in the medium
and long run if public employment exerts a positive and sizeable effect on
firms’ productivity.
Our results indicate that more aggressive fiscal policy will generally
lead to better economic performance following a monetary policy shock:
given the timing of the policy response to the public debt, the sacrifice
ratio is generally lower when fiscal policy is more aggressive, thus taking
less time to pull the debt back to its equilibrium level.
Finally, our exercise also shows that the monetary shock and the
ensuing fiscal tightening have clear distributional consequences, with
most fiscal rules favoring Ricardian individuals relative to non-Ricardian
ones. The only rule that potentially generates long run consumption gains
for both Ricardian and non-Ricardian agents is again the one that adopts
public employment as the fiscal adjustment instrument. As the reduction
in public employment leads to higher labour supply by Ricardian in-
dividuals to the private sector, we shift resources from nonproductive to
productive activities, thus making it possible to increase overall con-
sumption. Moreover, as Ricardian labour supply increases relative to
non-Ricardian labour in the private sector, Ricardian wages fall relative
to non-Ricardian wages, which provides room for non-Ricardian con-
sumption to also increase. Again, this result depends on the hypothesis
that public employment does not directly impact either capital accu-
mulation or total factor productivity.
Indeed, some conclusions may depend on our assumptions
regarding the impact of fiscal policy instruments, notably public in-
vestment, government employment and public consumption, on the
economy. Further studies on the impact of government spending on
individuals' utility (especially on the marginal utility of consumption),
the correct evaluation of the impact of public capital and employment
on firms' productivity, and a careful measurement of the effects of
public employment on public capital accumulation should therefore be
tackled in more depth, given their importance on the behavior of
macroeconomic variables and society's welfare, both throughout eco-
nomic cycles and in the long run.

Appendix A. Supplementary data

Supplementary data related to this article can be found at https://doi.org/10.1016/j.econmod.2017.12.010.

Appendix 1. The link between the policy rate and public debt service

In our model, we assume that there is a close relationship between the monetary policy rate and the cost of financing public debt, which strengthens
the link between monetary and fiscal policies. The information displayed in Fig. 7 below gives support to this hypothesis. The solid line shows the
trajectory of the Brazilian policy rate, the so-called SELIC rate, over the past 15 years, while the dashed line shows the interest rate on public debt (both
series represented in 12-month moving averages). The latter is calculated as follows: we use Intt , which is the cost of servicing the public debt at period t
(or the amount of interest payments observed in that month), and the value of outstanding public debt in the previous month (Dt1 ) to calculate the ratio
Dt1 , and subsequently smooth the final result by means of a 12-month moving average. The chart shows that the two series move very close; in
Intt

quantitative terms, the correlation coefficient between them exceeds 0.9. This result suggests that the aforementioned hypothesis is quite reasonable in
the Brazilian case.

16
M.A.F.H. Cavalcanti et al. Economic Modelling xxx (2017) 1–18

Fig. 7. Comparing the SELIC rate (left scale) and the cost of financing public debt in Brazil (right scale).

One of the main reasons for this correlation is the importance of the so-called LFT's amongst the securities usually issued by the Brazilian Treasury.
The LFT's are floating rate bonds that pay the investor their face value plus the accrued daily interest between their issue and maturity dates, with the
accrued interest calculated as a function of the SELIC rates observed over the period. The share of LFT's in the Brazilian federal domestic public debt
stock has fluctuated around 40% between January 2000 and August 2017, having reached almost 70% in April 2003 and being above 30% more
recently20. The importance of this type of security in the domestic public debt stock makes its rollover cost quite sensitive to the SELIC rate, which helps
to explain the behavior illustrated in Fig. 7.

References Christiano, L.J., Eichenbaum, M., Evans, C.L., 1999. Monetary policy shocks: what have
we learned and to what end?. In: Taylor, J.B., Woodford, M. (Eds.), Handbook of
Macroeconomics, 1A. Elsevier, , Amsterdam.
Alesina, A., Perotti, R., 1997. Fiscal adjustments in OECD countries: composition and
Christiano, L., Eichenbaum, M., Evans, C., 2005. Nominal rigidities and the dynamic
macroeconomic effects. IMF 44 (2), 210–248.
effects to a shock of monetary policy. J. Polit. Econ. 113 (1), 1–45.
Aschauer, D., 1989. Is public expenditure productive? J. Monetary Econ. 23, 177–200.
Davig, T., Leeper, E.M., 2011. Monetary-fiscal policy interactions and fiscal stimulus. Eur.
Annicchiarico, B., Marini, G., Piergallini, A., 2008. Monetary policy and fiscal rules. B E J.
Econ. Rev. 55, 211–227.
Macroecon. 8 (1). Contributions.
Dib, A., 2011. Monetary policy in estimated models of small open and closed economies.
Barakchian, S.M., Crowe, C., 2013. Monetary policy matters: evidence from new shocks
Open Econ. Rev. 22 (5), 769–796.
data. J. Monetary Econ. 60, 950–966.
Feve, P., Pietrunti, M., 2016. Noisy fiscal policy. Eur. Econ. Rev. 85, 144–164.
Barro, R.J., 1991. Economic growth in a cross section of countries. Q. J. Econ. 106 (2),
Forni, M., Gambetti, L., 2010. The dynamic effects of monetary policy: a structural factor
407–443.
model approach. J. Monetary Econ. 57, 203–216.
Baxter, M., King, R., 1993. Fiscal policy in general equilibrium. Am. Econ. Rev. 83,
Forni, L., Monteforte, L., Sessa, L., 2009. The general equilibrium effects of fiscal policy:
315–334.
estimates for the euro area. J. Publ. Econ. 93 (3–4), 559–585.
Bernanke, B., Boivin, J., Eliasz, P., 2005. Measuring the effects of monetary policy: a
Jawadi, F., Mallick, S., Sousa, R., 2014. Fiscal policy in the BRICs. Stud. Nonlinear
factor-augmented vector autoregressive (FAVAR) approach. Q. J. Econ. 120,
Dynam. Econom. 18 (2), 201–215.
387–422.
Kumar, A., Beck, T., Campos, C., Chattopadhyay, S., 2005. Assessing Financial Access in
Blanchard, O., 2004. Fiscal dominance and inflation targeting: lessons from Brazil. In:
Brazil. The World Bank. World Bank Working Paper n. 50.
Giavazzi, F., Goldfajn, I., Herrera, S. (Eds.), “Inflation Targeting, Debt and the
Leeper, E., 1991. Equilibria under active and passive monetary and fiscal policies.
Brazilian Experience, 1999 to 2003. MIT Press.
J. Monetary Econ. 27, 129–147.
Boivin, J., Kiley, M.T., Mishkin, F.S., 2011. How has the monetary transmission
Leeper, E.M., Plante, M., Traum, N., 2010a. Dynamics of fiscal financing in the United
mechanism evolved over time? In: Friedman, B., Woodford, M. (Eds.), Handbook of
States. J. Econom. 156 (2), 304–321.
Macroeconomics, Volume 3A. Elsevier.
Leeper, E., Walker, T., Yang, S., 2010b. Government investment and fiscal stimulus.
Calderon, C., Easterly, W., Serven, L., 2003. Latin America's infrastructure in the era of
J. Monetary Econ. 57, 1000–1012.
macroeconomic crises. In: Easterly, W., Serven, L. (Eds.), The Limits of Stabilization:
Leith, C., Wren-Lewis, S., 2000. Interactions between monetary and fiscal policy rules.
Infrastructure, Public Deficits, and Growth in Latin America. Stanford University
Econ. J. 110 (462), C93–C108. Conference Papers.
Press, Stanford, pp. 21–94, 2003.
Leith, C., Wren-Lewis, S., 2008. Interactions between monetary and fiscal policy under
Calvo, G.A., 1983. Staggered price setting in a utility-maximizing framework. J. Monetary
flexible exchange rates. J. Econ. Dynam. Contr. 32, 2854–2882.
Econ. 12, 383–398.
Lemoine, M., Linde, J., 2016. Fiscal consolidation under imperfect credibility. Eur. Econ.
Canzoneri, M., Cumby, R., Diba, B., 2006. How do monetary and fiscal policy interact in
Rev. 88, 108–141.
the european monetary union? In: Clarida, R., et al. (Eds.), NBER International
Malik, A.K., 2013. The effects of fiscal spending shocks on the performance of simple
Seminar on Macroeconomics 2004. MIT Press.
monetary policy rules. Econ. Modell. 30, 643–662.
Castro, M., Gouvea, S., Minella, A., Santos, R., Souza-Sobrinho, N., 2015. SAMBA:
Mallick, S.K., 2006. Policy instruments to avoid output collapse: an optimal control model
stochastic analytical model with a bayesian approach”. Brazilian Review of
for India. Appl. Financ. Econ. 16, 761–776.
Econometrics 35 (2), 103–170.
Mallick, S.K., Sousa, R.M., 2012. Real effects of monetary policy in large emerging
Carvalho, F., Valli, M., 2011. Fiscal Policy in Brazil through the Lens of an Estimated
economies. Macroecon. Dyn. 16 (S2), 190–212.
DSGE Model. Working Papers Series 240, Central Bank of Brazil, Research
Medina, J., Soto, C., 2006. Model for analysis and simulations: a small open economy
Department.
DSGE for Chile. Central Bank of Chile 2006.
Cavalcanti, M., Vereda, L., 2011. Propriedades Din^amicas de um Modelo DSGE com
Mereb, J., Zilberman, E., 2013. O Programa de Aceleraç~ao do Crescimento Acelera o
Parametrizaç~ oes Alternativas para o Brasil. Working Paper n. 1588, Instituto de
Crescimento? Texto para Discuss~ao 613, Departamento de Economia, PUC-Rio.
Pesquisa Econ^ omica Aplicada.
Orair, R.O., 2016. Investimento Público no Brasil: Trajet oria Recente e Relaç~oes com
Çebi, C., 2012. The interaction between monetary and fiscal policies in Turkey: an
Ciclo Econ^ omico e Regime Fiscal. Working Paper n. 2215, Institute for Applied
estimated new keynesian DSGE model. Econ. Modell. 29, 1258–1267.
Economic Research (IPEA).

20
Source: Time Series Management System (SGS), Central Bank of Brazil, public finance indicators, series 4177 (floating rate debt securities indexed by the Over/SELIC rate – share in the
portfolio of bonds issued by the Brazilian federal government).

17
M.A.F.H. Cavalcanti et al.
Philippopoulos, A., Varthalitis, P., Vassilatos, V., 2015. “Optimal fiscal and monetary
policy action in a closed economy”. Econ. Modell. 48, 175–188.
Schmitt-Groh e, S., Uribe, M., 2005. Optimal fiscal and monetary policy in a medium-scale
macroeconomic model. In: Gertler, M., Rogoff, K. (Eds.), NBER Macroeconomics
Annual 2005.
Schmitt-Groh e, S., Uribe, M., 2007. Optimal simple and implementable monetary and
fiscal rules. J. Monetary Econ. 54, 1702–1725.
Smets, F., Wouters, R., 2003. An estimated dynamic stochastic general equilibrium model
of the euro area. J. Eur. Econ. Assoc. 1 (5), 1123–1175.
ahler, N., Thomas, C., 2012. FiMod - a DSGE model for fiscal policy simulations. Econ.
St€
Modell. 29 (2), 239–261.
18
Economic Modelling xxx (2017) 1–18
Valli, M., Carvalho, F., 2010. Fiscal and Monetary Policy Interaction: a Simulation Based
Analysis of a Two-country New Keynesian DSGE Model with Heterogeneous
Households. Working Papers Series 204, Central Bank of Brazil, Research
Department.
von Thadden, L., 2004. Active monetary policy, passive fiscal policy and the value of
public debt: some further monetarist arithmetic. J. Macroecon. 26, 223–251.
Woodford, M., 2001. “Fiscal requirements for price stability. J. Money Credit Bank. 33
(3), 669–728.
Zubairy, Sarah, 2014. On fiscal multipliers: estimates from a medium scale DSGE model.
Int. Econ. Rev. 55 (1), 169–195.