World Development Vol. 40, No. 9, pp.

17621783, 2012
2012 Elsevier Ltd. All rights reserved
0305-750X/$ - see front matter
www.elsevier.com/locate/worlddev

http://dx.doi.org/10.1016/j.worlddev.2012.04.024

Fiscal Sustainability: The Impact of Real Exchange Rate Shocks

on Debt Valuation, Interest Rates and GDP Growth I
CLAUDIA MARTINEZ CARRERA and RODRIGO VERGARA *
Central Bank of Chile, Santiago, Chile
Summary. This paper discusses the way in which the existence of foreign currency debt aects debt-sustainability analysis. We show
that a devaluation of the local currency can signicantly change the path of a sustainable scal policy. Our model expands previous
research as the adjustment comes not only through changes in the value of the foreign-currency-denominated debt, but also through
the eects of the devaluation on interest rates and economic growth. We nd that the scal adjustment required after a devaluation increases with the size of the devaluation, the length of the adjustment period, the eect on interest rates and growth, and the share of
public debt denominated in foreign currency.
2012 Elsevier Ltd. All rights reserved.
Key words public debt, valuation eects, interest rate, growth rate, Latin American countries, Chile

1. INTRODUCTION

knowledge that using domestic currency to borrow abroad

or to borrow long-term even domestically is a limited possibility for many developing countries (referred by Hausmann and
Panizza (2003) and Eichengreen and Hausmann (1999) as the
original sin). In fact, during the early 2000s there was a debate
regarding the causes of the inability of emerging economies to
borrow abroad in their own currency. Some authors attributed
this result to institutional weaknesses of emerging-market
economies that lead to weak and unreliable policies, while others associate the problem to the structure of global portfolios
and international nancial markets. More recently the discussion has focused on the reduction of the original sin. However,
Hausmann and Panizza (2010) nd that the original sin has
declined only marginally and in just a few selected countries.
Therefore domestic currency denominated debt is still a low
proportion of total debt in emerging economies, and the eect
of a real devaluation on scal sustainability is an important
concern to policy makers.
Our empirical model is similar to the approach of Penalver
and Thwaites (2006). The main dierence in our work is that
we not only include valuation eects in the sustainability analysis but also allow for a response from interest rates and the
growth rate to the increase in debt due to the valuation eect.
To accomplish this we rst estimate the impact of an increase
of public debt on interest rates, and on GDP growth rate. In
our estimations we considered a sample of ve Latin American
countries (Brazil, Chile, Colombia, Mexico, and Peru). 8 Using
quarterly data for the period 19992007 we estimate a dynamic panel. From here we are able to obtain the main parameters of the model. The last step previous to the simulation of
the sustainable path of scal policy is the estimation of the
time that a certain country takes to adapt to its sustainable
public debt level after an unexpected one time RER shock.

An economy is said to have achieved scal sustainability

when the ratio of public sector debt to GDP is stationary,
and consistent with the overall demandboth domestic and
foreignfor government securities. 1 An important by-product of public sector sustainability analysis is the computation
of the public sectors primary balance compatible with a sustainable and stable debt to GDP ratio. 2 This sustainable primary balance has become an increasingly important variable
in macroeconomic analyses and is now routinely included as a
disbursement condition in IMF programs.
Several studies have analyzed debt sustainability in both
developed and nondeveloped countries. A widely used criterion is the gap between the actual primary decit and the
one required to keep the debt to GDP ratio stable. The World
Bank and the IMF have studied external debt sustainability
using a present value constraint approach. 3 This approach
analyzes whether the net present value (PV) of the countrys
external debt stabilizes at its steady state level relative to
GDP, exports of goods, and services or government revenues. 4 It also provides several other debt burden indicators,
like the primary decit that stabilizes the debt to GDP ratio.
The latter only stabilizes the PV ratio in the year in question,
assuming that all previous years followed the path of a baseline scenario. Under this approach, a uniform discount rate
is used to calculate the PV of future external debt-service obligations. Also several scenarios and stress tests produce partial
bound tests. 5
This paper goes further than previous works and explicitly
discusses the way in which real exchange rate changesmore
specically, real exchange rate devaluationsaect scal sustainability not only through the change in the value of the foreign currency-denominated public debt, but also because a
higher debt-to-GDP ratio leads to a higher interest rate and
to a below potential GDP growth rate, changing the path of
primary balances consistent with scal sustainability. 6
Policy decisions must take this factor into account because
emerging economies frequently face terms of trade shocks or
other disturbances that require a real devaluation. One possibility, suggested by Hausmann (2004), 7 would be to issue
domestic currency-denominated debt. But it is common

* We are grateful to Sebastian Edwards for valuable suggestions and to

participants in a seminar at Universidad Catolica de Chile. We would also
like to thank two anonymous referees for providing useful comments and
suggestions that improved signicantly the quality of the paper. Final
revision accepted: December 12, 2011.
q
The views and conclusions presented in this paper do not necessarily
reect the position of the Central Bank of Chile.
1762

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

To address this we use vector autoregressions for each country

in our sample in order to compute the impulse response functions which give us an estimate of the time that takes for a given country to go back to its sustainable public debt level.
Finally we are able to simulate the sustainable primary balance and debt dynamics of a country facing an initial real
depreciation.
Our results suggest that the eects on this sustainable path
can be substantial. Indeed, assuming that the equilibrium
debt-to-GDP ratio for a country is 35%, that in this country
foreign currency denominated public debt accounts for half
the total public debt, and that it has three years to return to
the equilibrium public debt-to-GDP ratio after a 25% real
devaluation, the required primary balance adjustment after
the devaluation is about 1% of GDP for each of those three
years. This is a very important adjustment in terms of magnitude for emerging economies that usually have diculties
using domestic currency to borrow abroad and with scal
positions that are often very tight.
The paper is organized as follows. In Section 2, we discuss
recent policy debate on scal sustainability analysis and review
the recent literature. In Section 3, we present the sustainability
model, which includes the valuation eects. In Section 4, we
estimate the key parameters of the model. With this estimation
and other assumptions that are common among developing
countries, 9 we perform, in Section 5, some simulations to
determine the sustainable path of the scal balance after a signicant devaluation of the domestic currency. Section 6 concludes.
2. LITERATURE REVIEW
Since 1980s there has been an important amount of studies
concerning debt sustainability. This section will focus on the
literature that is most relevant to the theme being investigated
in this paper.
There are several theories regarding inability of emerging
economies to borrow using its domestic currency. There is a debate of whether in the last years there has been a redemption
of the original sin, in the sense that many emerging market
countries can now issue external debt in their own currency. 10
However, as mentioned earlier, Hausmann and Panizza (2010)
nd that there has been a modest increase in the past decade in
the number of countries able and willing to issue local-currency
debt in international markets. In a sample of 65 developing
countries, only nine ever managed to issue at least 15% of their
debt in their own currency and only 18 would have ever been
able to swap at least 25% of their international debt securities.
They conclude that the original sin continues to make nancial
globalization unattractive, and developing countries have
opted for abstinence because foreign currency debt is too risky
given the mismatches it generates. 11
Related to this literature, Calvo and Reinhart (2002) argue
that devaluations are more contractionary in developing countries than in industrial countries. The basic explanation is that
the former have larger currency mismatches than the latter.
These currency mismatches give rise to what they call the fear
of oating. Hausmann, Panizza, and Stein (2001) assert that
the greater the dependence of a developing economy on foreign currency borrowing, the greater its fear of oating. They
show that the currency mismatches explain the cross-country
dierences in the fear of oating better than the cross-country
dierences in the pass-through coecient.
On the other hand, Mendoza and Ostry (2008) study scal
solvency and public debt sustainability. They use the

1763

Model Based Sustainability (MBS) approach proposed

by Bohn (1998, 2005) to test whether the scal authority reacts to positive changes in the public debt ratio by systematically raising the primary surplus to GDP ratio.
Their main nding is that both in emerging market economies and in advanced economies there is a positive conditional
response of primary scal balances to changes in government
debt. This response is estimated to be stronger in emerging
markets than in advanced countries, because the riskier scal
and nancial environment in the former group of countries requires a stronger response to maintain scal solvency, meaning that emerging markets can sustain lower mean public
debt ratios in the long run. Thus, the stronger response is
not an indicator of more scal discipline. Moreover, for
emerging market economies they nd a nonlinear relationship
between scal policy and debt. Specically, the primary balances responsiveness to debt diminishes above a threshold level of debt.
Penalver and Thwaites (2006) estimate an econometric model of the determinants of public debt dynamics on Brazilian
data and use this model to simulate the eect of dierent scal
policy rules for future paths of debt. They then derive the set
of scal policy rules which stabilize public debt dynamics.
Their empirical estimates show that the interaction of shocks
to the determinants of debt sustainability with feedback from
the debt-GDP ratio can be quantitatively important for a representative EME.
Valuation eects have been studied by Calvo, Izquierdo, and
Talvi (2003), among others (mainly IMF studies). Calvo et al.
consider the eects of a 50% real depreciation of the domestic
currency on scal sustainability in dierent Latin American
countries, assuming that interest rates and GDP growth remain unchanged. This last assumption is very common when
analyzing scal sustainability. For the case of Argentina, these
authors nd that the 50% permanent depreciation requires a
permanent primary surplus of 1.6% of GDP so as to keep
the ratio of debt to GDP constant. The PV of the dierence
between the required primary surplus before and after the
RER depreciation would be equivalent to 14.3% of GDP.
Garcia and Rigobon (2004) in an attempt to endogenize the
interest rate, propose a vector autoregression to estimate the
correlation pattern of the macro variables of the Brazilian
economy and use it to implement Monte-Carlo simulations.
From these simulations they compute risk probabilities,
i.e., probabilities that the simulated Debt to GDP ratio exceeds a given threshold deemed risky (say, 75% of GDP).
The time-series of such probabilities is then used to investigate
whether or not it is correlated with the market risk assessment,
measured by the spread on sovereign dollar denominated debt.
They show that properties of the debt dynamics are closely related to the EMBI Plus Brazil spread. Hostland and Karam
(2006) used stochastic simulation methods to assess debt sustainability in emerging market economies. According to these
authors, the vulnerability of public debt to adverse shocks is
determined by four inter-related factors, namely the volatility
of output, nancial fragility (reliance on short-term, foreign
currency borrowing), the endogenous response of the risk premium, and sudden stops in private capital ows; the vulnerability of external debt is sensitive to the determination of the
exchange rate and to the pricing of traded goods. Their model
is calibrated such that a 10% point increase in the debt burden
raises the risk premium by 250 basis points. 12 Endogenizing
risk premium into the model has a large eect on the primary
scal balance but relatively little eect on the public debt,
external debt and trade balance. In our analysis, we assume
that both interest rates and GDP are aected, at least in the

1764

WORLD DEVELOPMENT

short run, and we also introduce dynamic aspects, such as

countries having some time to adjust to their sustainable debt
level after the devaluation.
3. THE MODEL
Total public debt (measured in domestic currency) is calculated as:
Dt P t e t F t ;

where Dt is total debt, Pt is domestic currency-denominated

debt, Ft is dollar-denominated debt, and et is the nominal exchange rate (domestic currency units per dollar). Subscript t
denotes time.
We assume that this country starts from a position of scal
sustainability, in the sense that its total public debt as a fraction of GDP (Dt/Yt) is consistent with the demand for government debt.
From (1) it follows that,
DDt DP t et DF t F t Det :

The rst term on the right-hand side (DPt + etDFt) represents

new debt being issued, and in that regard it captures fresh
resources. The second term (FtDet) is the valuation eect.
Eqn. (2) clearly shows that the total debt measured in domestic
currency can increase for one of two reasons: new debt may be
issued or the local currency-value of the old debt may increase
due to a nominal devaluation. Naturally, there may be a combination of these two.
According to the public sector budget constraint, net new
debt issuedin domestic and foreign currencyhas to be
equal to the gap between expenditures and revenues. Expenditures can be broken down into two components: primary
expenditures and interest payments. Revenues, in turn, are
equal to seigniorage (St) and other revenues. Thus, the public
sector budget constraint may be written as follows (where DDNt
is new debt issued and corresponds to the term (DPt + etDFt)
in Eqn. (2):
DDNt pbt it P t iFC
t et F t  S t ;

where pbt is the primary balance, dened as nonseigniorage

revenues minus primary expenditures, it is the interest rate
on domestic-currency-denominated debt, and iFC
t is the interest
rate on foreign-currency-denominated debt.
A key variable in any sustainability analysis is the evolution
of the debt-to-GDP ratio over time. To the extent that both
variables (i.e., debt and GDP) are measured in the same currency, tracing the evolution of the ratio is easy. Things get a
bit more complicated when the two variables in the ratio are
denominated in dierent currencies or, more specically, when
part of the debt is denominated in foreign currency.
Let d represent the debt-to-GDP ratio:
dt

Dt
;
Yt

where both variables are measured in current local currency

and Dt comes from Eqn. (1).
It follows from (4), (1), and (2) and after some math that:

Ddt


DP t et DF t F t Det

 gt pt dt ;
Dt
Dt
Dt

where gt is real GDP growth and pt is the domestic rate of

ination. If there is no foreign currency-denominated debt,

then Ft = 0 and (5) collapses to the familiar expression for

public sector debt sustainability. 13
If, however, some public sector debt is denominated in foreign currency, the traditional analysis will be misleading. In
this case, using Eqn. (3) on the public sector budget constraint
and some algebra, we obtain the expression for the primary
balance consistent with steady-state scal sustainability (i.e.,
Ddt = 0): substituting (3) in (5):
Ddt

pbt it P t iFC
F t Det
t et F t  S t

 gt pt dt :
Yt
Yt

Note that in steady state, Ddt 0.

FC

Using the Fisher equation it rt pt ; iFC
t rt pt , and
the real exchange rate denition:
b t
R ER

et
Det
b t  p pit
pt  pt )
R ER
t
et
et

the following expression is obtained for the primary budget

balance:

pbt
St P t
et F t  FC
Ddt  rt  gt
r  gt
Yt
Yt Yt
Yt t
et F t b
R ERt :
6

Yt
In deriving this expression, we have also used the fact that the
nominal GDP growth rate is equal to real growth (gt) and
ination (pt). rt is the real interest rate on domestic-curthe real interest rate on forrency-denominated debt and rFC
t
eign-currency-denominated debt. The last term on the righthand side is the correction factor that arises from the existence of foreign-currency-denominated debt, RERt is the real
exchange rate and the hat denotes percentage change.
Notice that ignoring this correction factor will result in a
miscalculation of the primary decit consistent with a stable
debt-to-GDP ratio. In particular, if the country experiences
a real devaluationthat is, if (RERt/RERt) > 0, 14 ignoring
the correction factor will result in an underestimated primary balance consistent with sustainability.
The case where there is a BalassaSamuelson eect in which
the currency exhibits a real appreciation over time can be
thought of as leading to a larger sustainable primary decit
(see Eqn. (6)), since this implies that the value of the foreign
debt as a percentage of GDP is declining over time due to this
eect. This, however, does not invalidate our concern regarding the dramatic valuation eects that can take place if there is
a devaluation of considerable magnitude.
Assuming that d* is the sustainable public-debt-to-GDP ratio in the sense that it is consistent with the demand for that
debt, then if dt < d*, the country will be able to sustain smaller
primary balances than those of a steady state for a while. In
Eqn. (6) Ddt is positive (hence the primary balance smaller),
that is, the debt-to-GDP ratio can increase and still be consistent with debt sustainability. The inverse occurs if dt > d*.
We assume that the uncovered real interest rate parity holds:
b t qt rFC ;
rt rt ER ER
t

7
rt

where qt is the country risk premium and

is the (exogenously given) risk-free world interest rate. Additionally, we asb t 0.
sume that ER ER
On the other hand, as in Edwards (1986), Min (1998) and
Akitoby and Stratmann (2006), 15 we assume that the country
risk premium is a function of the level of public debt and a set
of economic variables (X t ). The larger the public debt, the
higher the risk premium. 16 Hence:

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

qt qdt ; X t with @qt =@dt > 0:

We also assume that after a devaluation occurs, the debt

holders (actual and potential) do not necessarily require the
country to return to its initial position immediately. They
understand that there are valuation eects. Hence, they give
some time t to the country to adjust and return to d*. This time
t will depend on the track record of the country and on the
credibility of its adjustment program.
Let dt denote the accepted level of public debt (as a ratio of
GDP) in the years after devaluation. The behavior of 
dt is represented by:

dt d d0  d 1  ct for t 1; 2; . . . ; t;
9
where c 1t .
Now, d0 represents the actual level of public debt just after
the devaluation. In other words,
d0 d

F t DEt
:
Yt

10

Expression (9) tells that as t approaches t, the level of accepted public debt approaches the steady-state level d*. On
one extreme, if the market gives the country just one period

to adjust to d*, then d
t d . This means that the primary
balance will have to increase as much as necessary so that
the ratio of public debt to GDP remains unchanged at level
d*. On the other extreme, if the market gives the country
innite time to adjust (t 1), then 
dt d0 . This means that
the new equilibrium debt-to-GDP ratio becomes d0 rather
than d*.
Finally, we assume that when public debt is above its sustainable level, GDP growth is below its potential, the reasons
being the higher interest rate discussed above and the increased uncertainty derived from the fact that taxes or seigniorage might be increased to nance the larger primary
decit. 17 The larger the gap between sustainable and actual
debt, the greater the impact on growth. Additionally, to explore potential nonlinearities of the relation between public
debt and GPD growth, we follow earlier studies by Reinhart
and Rogo (2010) and Kumar and Woo (2010) and work with
a threshold of initial public debt beyond which the debt begins
to have an adverse eect on growth. This relationship can be
expressed as:
 
if dt 6 d
gt
;
11
g
gt  bdt  d if dt > d
where g* is the potential GDP growth rate and d* is the debt
threshold beyond which debt will have adverse eect on
growth.
4. ESTIMATIONS
In this section, we are interested in estimating the eect of
public debt on the interest rate of that debt (Eqns. (7) and
(8)), the dynamics of the debt when there is a shock to the real
exchange rate (Eqns. (9) and (10)), and the impact of debt on
growth (Eqn. (11)).
Our estimations considered ve Latin American countries:
Brazil, Chile, Colombia, Mexico, and Peru. 18 We used quarterly data for the period 1999:I2007:IV. The variables included country risk as measured by the quarterly average of
the EMBI Plus index qt . 19 The data on external public debt
as a percentage of GDP were obtained from the World Bank.
For the real exchange rate, we used the traditional measure:

1765



Et P t
RERt
;
Pt
where P t is the dollar-denominated CPI of the main trading
partners, E is the nominal exchange rate (local currency units
per dollar), and P is the domestic CPI. All these data plus the
GDP growth data were obtained from the central banks of the
countries considered in the study. 20
(a) Public debt and interest rates
We now proceed to estimate the eect of a change in public
debt on the interest rate on that debt. Because there is no total
public debt quarterly data available for all the countries considered, we do our estimations using only foreign public
debt. 21 Our dependent variable is the risk premium (see Eqns.
(7) and (8)), that is, we want to estimate the eect of a change
in the public debt on the risk premium paid on that debt.
We estimate the following panel:

qit a1 a2 qit  1 a3
li nit ;

EPD
GDP


a4 RERit a5 git kt
it

12

where subscript i denotes the country (i = 1, 2, . . . , 5) and subscript t denotes the quarter:
EPD
GDP
it is the ratio of public foreign debt to GDP. As we
mentioned above, it is expected that the greater the public
debt, the higher the risk premium.
RERit is the real exchange rate index. Like in Edwards
(1986) and Min (1998), we analyze whether a less competitive real exchange rate (appreciation) can adversely aect
the risk premium. According to Cline (1983), real appreciations in LDCs played a major role in the overborrowing
process.
git is the rate of growth of GDP. We expect that a higher
GDP growth rate will reduce country risk.
Finally, kt is a temporary xed eect, li is a country-specic xed eect, and nit is an iid 0; r2n distributed-error
term.
We rst estimated Eqn. (12) using a static GLS xed-eect
method. The Hausmann test rejects the null hypothesis of no
correlations between the xed eects and the explanatory variables, thus validating the estimation through xed eects.
However, considering both the dynamic structure of the model
and the potential endogeneity of the explanatory variables
([EPD/GDP]iRERit and git ) we proceeded to estimate using
System GMM proposed by Arellano and Bover (1995) and
Blundell and Bond (1998).
The results are presented in Table 1. The rst regression
shows the results of the xed-eect estimation. The eect of
the debt on the risk premium (and hence on the interest rate)
is positive and statistically signicant. The coecient shows
that a one percentage-point increase in the debt-to-GDP ratio
(say, from 30% to 31%) produces an increase in the interest
rate of 21 basis points (say, from 5% to 5.21%). Regressions
24 show the results of the dynamic estimation using System
GMM methodology. The results are consistent with those
found in regression 1 in the sense that the coecient of the
debt variable is positive and statistically signicant, and the
magnitude of the eect is 0.11%. However, we are able, with
the dynamic specication, to obtain long-term eects. Note
that the coecient of the lagged dependent variable is positive
and statistically signicant, indicating persistence in the eect
on the risk premium. 22

1766

WORLD DEVELOPMENT
Table 1. Estimates using xed eects model and System GMM

Dependent variable: interest rate spread (qt)

Eqn. (1)

qt1
(External public debt/GDPt)
RERt
GDP growth ratet
Time eect
2

R
Observations
Number of id
Serial correlation order 1
Serial correlation order 2
Sargan p value

0.213***
(0.0265)
0.0504***
(0.00860)
0.0812
(0.0555)
Yes
0.877
155
5

Eqn. (2)

Eqn. (3)

Eqn. (4)

0.691***
(0.0729)
0.0969**
(0.0269)

0.537**
(0.139)
0.113***
(0.0259)
0.0473
(0.0276)

Yes

Yes

0.659***
(0.156)
0.113*
(0.0462)
0.0232
(0.0224)
0.288
(0.184)
Yes

153
5
0.0696
0.219
0.107

153
5
0.0878
0.202
0.118

153
5
0.0384
0.220
0.108

Standard errors in parentheses.

***
p < 0.01.
**
p < 0.05.
*
p < 0.1.

The Sargan test statistic for overidentifying restrictions conrms the exogeneity of the instruments validating our GMM
estimations. The second-order residual serial correlation tests
conrm the fact that the errors are serially uncorrelated.
Given that including Brazil in the estimations could be troublesome because of the persistently high interest rates, we perform a sensitivity analysis with Brazil excluded from our
sample. When excluding Brazil from the sample, the coecient
of public debt to GDP declines to 0.05, meaning that a onepercentage-point increase in the debt-to-GDP ratio produces
an increase in the interest rate of 5 basis points (see Appendix
Table 6). This coecient is about half of what we obtained
when Brazil was included in the sample.
(b) The dynamics of public debt
The second step in our analysis is to study the dynamics of
public debt when the exchange rate varies (Eqns. (9) and (10)).
To address this issue, we use vector autoregressions (VAR) to
analyze the dynamic impact on public debt in each country of
a random disturbance in the real exchange rate.
Let Xt be the vector of endogenous variables. It would be
written in reduced form as follows for the VAR system:
X t ALX t1 U t ;
h
i
EPD
; RERt ; qt ; gt and Ut is a reduced residuals
where X t GDPt
vector dened as:
h EPD
i
; uRER
; uqt ; ugt :
U t uGDP
t
t
The equations include a lag, established using Akaikes criteria.
The reduced residuals (Ut) and, more specically, the resid
EPD

,
uals of the ratio between foreign public debt and GDP uGDP
t
can also be determined as: (1) a linear combination of the response of the foreign-public-debt-to-GDP ratio when there are
unexpected shocks in the other variables, (2) the discretionary
response of the policy-maker to changes in the variables 23 and
(3) the random shocks of foreign debt. 24
As mentioned earlier, our analysis concentrates on the
response of the foreign-public-debt-to-GDP ratio to an
unexpected shock in the real exchange rate. We are specically
looking to estimate the time (t) that a certain country takes to

adapt to its sustainable public debt level. We therefore estimated the eect of an unexpected shock (one-time only) on
the RER, assuming that the RER will not react contemporaneously to changes in the other model variables and that the
other variables (EPD/GDP, q, g) do not react contemporaneously to shocks in the other variables, except for changes in the
RER. Moreover, the path of EPD/GDP is estimated separately from the shocks to the rest of the variables (i.e., the only
shock that is received throughout the period of analysis hits
the RER).
The results are presented in Figure 1, panels AE. This gure shows the response of the foreign-public-debt-to-GDP ratio to a one standard deviation shock in the RER using the
Cholesky decomposition with the small sample degrees of freedom correction to orthogonalize the impulses. Asymptotic
standard errors are used to construct the condence bands.
The Cholesky scheme implies that RER shocks have no contemporaneous eect on the interest rate and GDP growth rate.
The RER shock aects only the debt within the same quarter.
The results show the responses over a horizon of 15 years. The
gure shows that the foreign public debt-to-GDP ratio increases and that this eect lasts for approximately two to three
years in the cases of Brazil, Chile, Colombia, and Mexico. In
the case of Peru we found no signicant eect. 25
Hence, for the simulations in our model, we will assume that
the time in which the country has to return to an acceptable
level of public debt (in the sense discussed in the previous section) is three years, but we also use a range going from 1 to
10 years.
(c) Public debt and growth
Finally, we turn to the question about the eect of public
debt on growth (Eqn. (11)). To nd the answer, we estimate
the following dynamic regression:



External Public Debt
git b1 b2 git1 b3
GDP
it


External Public Debt
 Dum 30 b4
GDP
it
 Dum 30 60 b5 RERit b6 rit kt ui vit :

13

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

1767

Figure 1. Impulse response functions response of external debt-to-GDP ratio to Cholesky one S.D. RER innovation.

As in Eqn. (12), we address the possible endogeneity problem of explanatory variables (external public debt, RER,
and spread) using System GMM. Table 2 presents the results.
The coecient of external public debt when there is a low initial debt is negative and statistically insignicant in all of the
equations. In contrast, the coecient of external public debt
when debt is above 30% is negative and statistically signicant,
indicating that growth declines as the debt increases. The coefcient indicates that a one percentage point increase in the
public-debt-to-GDP ratio reduces the growth rate by 0.09%
points in the short term.
The Sargan test and the second-order residual serial correlation tests conrm both the fact that the errors are serially
uncorrelated and the validity of the instrumental variables.

As in our estimates of interest rates we performed a sensitivity analysis excluding Brazil from our sample. Results are
shown in Appendix Table 7. The results show that a change
of one percentage point in the debt ratio would reduce growth
by 0.06% at levels above 30%, The impact of a change in debt
on growth rate is about two thirds of what we nd in the full
sample regressions. Reinhart (2010) nd that growth deteriorates markedly at debt levels over 60%. 26
5. SIMULATIONS
In this section, we simulate the sustainable primary balance
and debt dynamics of a country facing an initial real deprecia-

1768

WORLD DEVELOPMENT
Table 2. Estimates using System GMM

Dependent variable: GDP growth ratet

Eqn. (1)

Eqn. (2)

Eqn. (3)

GDP growth ratet1

0.535***
(0.115)
0.127*
(0.0533)
0.219
(0.143)

0.535***
(0.106)
0.0914*
(0.0358)
0.0925
(0.0541)
0.0412
(0.116)

Yes

Yes

0.536***
(0.107)
0.0912**
(0.0348)
0.0930
(0.0515)
0.0448
(0.151)
0.000711
(0.0117)
Yes

155
5
0.0979
0.552
0.406

154
5
0.0398
0.187
0.239

154
5
0.0388
0.189
0.224

(External public debt/GDPt)over30%

(External public debt/GDPt)below30%
rt
RERt
Time eect
Observations
Number of id
Serial correlation order 1
Serial correlation order 2
Sargan p
Robust standard errors in parentheses.
***
p < 0.01.
**
p < 0.05.
*
p < 0.1.

that the parameter that relates growth to debt is 0.09, and

the parameter that relates interest rate to debt is 0.11.
We also assume that the country starts at a point where
dt = d*. This is where the desired stock of this countrys public
debt is equal to its actual stock. For d*, we assume a value of
0.35 based on recent literature that considers that the sustainable public debt-to-GDP ratio in emerging economies is in the
range of 3040%. 27 Note that for countries that have access to
concessional debt, the relevant number is the PV of that debt
rather than its face value. In our rst simulation, we assume
that 55% of that debt is either denominated in foreign currency

tion. The primary balance path is represented by Eqn. (6), while

the public debt path is represented by Eqn. (9).
Table 3 contains the parameters used in our simulation.
From our empirical analysis in Section 3, we assume a range
of 110 years for t, although in most of the cases it is about
three years. This means that in three years time the country
has to adjust to the desired long-term public debt ratio. In
the lowest part of the range the country has to adjust within
the rst year to the desired long-term public debt ratio while
in the upper part of the range, it will have 10 years to return
to that level. From our growth panel regression, we assume

Table 3. Parameter values used in the scal sustainability analysis

Parameter

Symbol

Desired (and initial) external public

debt to GDP ratio

Initial foreign currency denominated debt as

percentage of total debt

(e0F0/D0)

Real interest rate when d = d*

r0

5.0%

Parameter that relates interest rate in t with

interest rate in t  1 (persistence)
Parameter that relates changes in interest rates with debt changes

a2

0.659

a3

0.113

Rate of real GDP growth when d = d*

g0

4%

Parameter that relates GDP growth rate

in t with GDP growth rate in t  1
Parameter that relates changes in growth with debt changes

b2

0.536

b3

0.09

Time necessary to come back to the desired

public debt ratio after the devaluation
Seigniorage

t
S/Y

Assumed value

Comments and sources

35%

Common gure used for

sustainability analyses based
on demand for public sector debt
(see Edwards & Vergara, 2002)
Fraction of public debt that is in
foreign currency. From the actual data
of LA countries

5570%

Between 1 and 10
0.4%

It is assumed to be equal to
r* + q = rFC
SGMM estimator in
autoregressive panel data model
SGMM estimator in
autoregressive panel data model
Simulation can be made using
dierent rates of growth
SGMM estimator in
autoregressive panel data model
SGMM estimator in
autoregressive panel data model
According to impulse response
functions
Consistent with a monetary base
of 6% of GDP and a nominal
GDP growth rate of 7%

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

1769

Table 4. Real devaluation of 25% initial foreign currency-denominated debt as a percentage of total debt: 55%
Panel A
Primary balance

Panel B
Public debt (% of GDP)

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

0.05
1.60
0.08
0.03
0.05
0.05
0.05
0.05
0.05
0.05
0.05

0.05
0.97
0.95
0.66
0.02
0.04
0.05
0.05
0.05
0.05
0.05

0.05
0.85
0.95
0.78
0.58
0.38
0.01
0.04
0.05
0.05
0.05

0.05
0.79
0.95
0.84
0.69
0.54
0.40
0.26
0.02
0.05
0.05

0.05
0.75
0.95
0.88
0.78
0.67
0.57
0.46
0.36
0.26
0.16

0
1
2
3
4
5
6
7
8
9
10

39.8
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

39.8
38.2
36.6
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

39.8
38.9
37.9
36.9
36.0
35.0
35.0
35.0
35.0
35.0
35.0

39.8
39.1
38.4
37.8
37.1
36.4
35.7
35.0
35.0
35.0
35.0

39.8
39.3
38.9
38.4
37.9
37.4
36.9
36.4
36.0
35.5
35.0

Panel C
Real interest rate

Panel D
Rate of real GDP growth

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

5.00
5.65
5.12
5.02
5.00
5.00
5.00
5.00
5.00
5.00
5.00

5.00
6.08
5.85
5.38
5.07
5.01
5.00
5.00
5.00
5.00
5.00

5.00
6.16
6.13
5.86
5.55
5.23
5.04
5.01
5.00
5.00
5.00

5.00
6.20
6.24
6.07
5.85
5.62
5.39
5.17
5.03
5.01
5.00

5.00
6.23
6.33
6.22
6.07
5.91
5.75
5.60
5.44
5.28
5.12

0
1
2
3
4
5
6
7
8
9
10

4.0
3.6
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.3
3.5
3.8
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.2
3.3
3.5
3.7
3.9
4.0
4.0
4.0
4.0
4.0

4.0
3.2
3.3
3.4
3.5
3.7
3.8
3.9
4.0
4.0
4.0

4.0
3.2
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
3.9

or indexed to the exchange rate. In the countries considered in

this paper, the average is around that gure, with Chile having
the lowest (10%) and Peru the highest (80%). However, in the
poorest countries, this fraction approaches 100%, which is
why we performed a second set of simulations where 70% of
the public debt is denominated in foreign currency.
For both scenarios, the real GDP is assumed to grow at 4%
per year when dt = d*. As dt increases, GDP growth declines
according to the parameter mentioned above. As shown in Edwards and Vergara (2001), Edwards and Vergara (2002), the
GDP growth rate is critical for the sustainable path of the primary balance. As the main focus of this paper is debt valuation issues, we do not simulate for dierent base rates of
growth (i.e., the rate of growth when d = d*), although it is
clear that as g* increases, the sustainable primary balance declines. Finally, the real interest rate on domestic public debt
when d = d* is assumed to be 5%. This is consistent with a
2.5% international real interest rate plus a risk premium of
250 basis points. 28
Seigniorage is assumed to equal 0.4% of GDP, which is consistent with a monetary base of 6% of GDP, an ination rate
of 3% per year (hence a nominal GDP growth of 7%), and a
unitary elasticity of the demand for money with respect to
GDP.
(a) Results
From our assumptions and using (6), we nd that with no
devaluation, the steady-state primary balance consistent with
scal sustainability for this theoretical country is a decit of
0.05% of GDP.

Now we assume a real devaluation of 25% at the end of period zero. As mentioned earlier, it is assumed that there are no
further expected changes in the real exchange rate. 29 Table 4
shows the results of our simulations for dierent values of t
and for 10 years after the devaluation occurs, when 55% of
public debt is denominated in foreign currency. Panel A shows
the primary balance sustainable path. If the country has three
years to adjust to the equilibrium debt-to-GDP ratio, it has to
have primary surpluses of 0.97%, 0.95%, and 0.66% respectively in the rst three years. This means a primary balance
adjustment of between 1.02% and 0.71% of GDP per year
for three years as compared to the nondevaluation situation.
This is clearly not a minor adjustment for a country that
was supposed to be scally sustainable. Panel B shows the
path of public debt as a percentage of GDP. For the same
t 3, it goes up to 39.8% of GDP right after the devaluation.
At the end of the rst year, it is down to 38.2% of GDP, and it
is back to its equilibrium level (35%) at the end of period 3.
Interest rates (panel C) go up to 6.08% at the end of year 1
and back to 5.0% in year 5. 30 Growth declines to 3.3% in
the rst year and returns to 4% after the fourth year.
If the country has to adjust in just one year, debt goes back
to its initial level at the end of year 1. But the required adjustment in the primary balance in that year is 1.6% of GDP.
In Table 5 we simulate the case where 70% of public debt is
denominated in foreign currency. If the country has three
years to return to the equilibrium debt-to-GDP ratio, it has
to adjust its primary balance between 1.27% and 0.86% of
GDP per year for three years as compared to the nondevaluation situation. Public debt goes up to 41.1% of GDP immediately after the devaluation. At the end of the rst year, it is

1770

WORLD DEVELOPMENT
Table 5. Real devaluation of 25% initial foreign currency-denominated debt as a percentage of total debt: 70%
Panel A
Primary balance

Panel B
Public debt (% of GDP)

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

0.05
2.06
0.11
0.02
0.05
0.05
0.05
0.05
0.05
0.05
0.05

0.05
1.27
1.24
0.86
0.04
0.04
0.05
0.05
0.05
0.05
0.05

0.05
1.11
1.24
1.02
0.75
0.50
0.00
0.04
0.05
0.05
0.05

0.05
1.05
1.25
1.10
0.90
0.71
0.52
0.34
0.01
0.04
0.05

0.05
1.00
1.25
1.16
1.03
0.88
0.74
0.61
0.48
0.35
0.22

0
1
2
3
4
5
6
7
8
9
10

41.1
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

41.1
39.1
37.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

41.1
39.9
38.7
37.5
36.2
35.0
35.0
35.0
35.0
35.0
35.0

41.1
40.3
39.4
38.5
37.6
36.8
35.9
35.0
35.0
35.0
35.0

41.1
40.5
39.9
39.3
38.7
38.1
37.5
36.8
36.2
35.6
35.0

Panel C
Real interest rate

Panel D
Rate of real GDP growth

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

5.00
5.82
5.16
5.03
5.01
5.00
5.00
5.00
5.00
5.00
5.00

5.00
6.37
6.08
5.48
5.09
5.02
5.00
5.00
5.00
5.00
5.00

5.00
6.48
6.43
6.09
5.70
5.30
5.06
5.01
5.00
5.00
5.00

5.00
6.53
6.58
6.36
6.08
5.79
5.50
5.21
5.04
5.01
5.00

5.00
6.56
6.70
6.55
6.36
6.16
5.96
5.76
5.55
5.35
5.15

0
1
2
3
4
5
6
7
8
9
10

4.0
3.5
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.1
3.4
3.8
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.0
3.2
3.4
3.6
3.9
4.0
4.0
4.0
4.0
4.0

4.0
3.0
3.1
3.2
3.4
3.6
3.7
3.9
4.0
4.0
4.0

4.0
3.0
3.0
3.1
3.2
3.3
3.5
3.6
3.7
3.8
3.9

down to 39.1% of GDP and it is back to its equilibrium level

(35%) at the end of period 3. Interest rates (panel C) go up to
6.37% at the end of year 1 and back to 5% in year 6. Growth
declines to 3.1% in the rst year and returns to 4% after the
fourth year.
In a more extreme scenario, where the fraction of public
debt denominated in foreign currency is 70% and the real
devaluation is 50%, the average primary balance adjustment,
when the adjustment period is three years, is 2.4% of GDP
per year as compared to the nondevaluation situation. In this
case, interest rates would jump to 7.8% and GDP growth
would decline to 2.2%.
If we use the parameters estimated in the regressions excluding Brazil, and assume that 55% of public debt is denominated
in foreign currency and that the country has three years to adjust to the equilibrium debt-to-GDP ratio after a real devaluation of 25%, our simulations show that the country has to
have primary surpluses of 0.8% in the rst two years after
devaluation and a surplus of 0.6% in the third year. And in
the scenario where the country has one year to adjust, the primary surplus needed to reach the equilibrium is 1.4% (see Table 9).
We also performed alternative simulations applying parameters from Baldacci and Kumar (2010) for the impact of debt
on the interest rate, and Kumar and Woo (2010) for the impact of debt on the GDP growth rate. The rst work nds that
that an increase in the debt ratio of 1% of GDP leads to an
increase in bond yields of around 5 basis points. On the other
hand, Kumar and Woo nd a 1% increase in initial debt-toGDP ratio is associated with a growth slowdown of around

0.04%. Results of the simulations using the latter parameters

are shown in Appendix Table 7. As the results show, after a
real devaluation of 25% in the scenario where 55% of public
debt is denominated in foreign currency, the country would
need a primary surplus of 0.5% in the rst two years after
devaluation and of 0.4% in the third year. This is about half
of the primary surplus obtained in our base case (with the full
sample).
The discrepancy arises because of dierences in the samples
used (advanced and emerging market economies versus Latin
American countries). There are also dierences in the methodologies of estimation. In particular, Baldacci and Kumar
(2010) do not account for persistence in interest rates and
although Kumar and Woo (2010) introduce dynamics in their
estimations they do not report the coecient of the lagged
dependent variable for the case of the impact of debt on the
GDP growth rate. When we take out Brazil from our sample
(Appendix Table 9) the simulations results are still higher but
closer to the results when using the parameters of the authors
mentioned earlier.
6. CONCLUSIONS
Ignoring valuation issues can result in misleading conclusions regarding scal sustainability. We have shown that,
starting from a point where the country is in a sustainable scal position, a devaluation can dramatically change the path of
primary balances consistent with current scal sustainability,
aecting not only the value of the foreign-currency-denomi-

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

nated public debt, but also considering the adverse eects that
external debt could have on the overall economic performance, more specically on the countrys interest rate and
GDP growth.
Assuming that the country has three years to return to the
equilibrium public debt-to-GDP ratio, the required primary
balance adjustment after the devaluation of 25% is about 1%
of GDP for each of those three years. Moreover, not taking
into account the dynamics of the interest rate and GDP
growth rate, would have led us to the mistaken conclusion that
the required scal adjustment was about half of the actual required adjustment.
The longer the adjustment period, the larger the underestimation of the adjustment in the primary balance required to
return to the equilibrium level of public debt, since as the
adjustment period increases, both interest rates and growth
are aected for a longer period. By ignoring the impact on
interest rate and growth assuming them to be exogenous, previous studies underestimate the scal costs of a real exchange
rate devaluation.

1771

The longer the adjustment period, the smaller the change

needed to return to the equilibrium debt. The larger the share
of foreign-currency-denominated debt, the greater the required scal adjustment, while the greater the initial devaluation, the greater the scal adjustment.
These conclusions are important for developing countries
that have a large share of their public debt in foreign currency
and frequently face terms of trade shocks or other disturbances that result in considerable depreciations of the domestic currency. With scal positions that are usually very tight,
requiring an adjustment of the magnitudes found in this paper
can be very stressful for these countries.
Finally, it is important to bear in mind that in this paper we
have assumed that the real exchange rate shock has a direct effect on the value of the stock of public debt and indirect eects
on other economic variables. However, some developing countries have state-owned companies that produce commodities
that are exported. 31 If this is the case, the devaluation also
produces a positive revenue eect which works in the opposite
direction.

NOTES
1. Naturally, the debt ratio may be calculated relative to alternative
benchmarks. On sustainability analyses see, for example, Milesi-Ferretti
and Razin (1996, 2000) and Edwards (2002).
2. The primary balance is dened as the nominal balance, excluding
interest payments.
3. See, for example, World Bank and IMF (2002, 2004) and Lachler
(2001).
4. See Cuddington (1997) and Arnone, Bandiera, and Presbitero (2008).
5. See IMF and World Bank (2010).
6. It could be argued that a devaluation from a position of an
unsustainably appreciated currency can make room for reductions rather
than increases in interest rates. Although this is a plausible case, it is usual
to see that devaluations initially bring turmoil to the market hence
increasing interest rates. In our model and simulations we assume that this
is the case (via an increase in public debt) as it has been so in most cases of
signicant depreciations in Latin American countries. In addition, our
empirical estimates indicate that interest rates rise after a depreciation (the
channel is an increase in public debt). On the other hand, it could also be
argued that devaluations are expansionary rather than contractionary,
hence GDP growth would increase rather than decrease. It is generally
observed, though, that this is not the case in the short term. Initially
devaluations are traumatic and reduce economic growth. In our simulations we assume that growth returns to its long-term value after the
country returns to a sustainable level of public debt. In addition our
empirical estimates indicate that growth declines after a depreciation.
7. See also Eichengreen, Hausmann and Panizza (2003).
8. We consider this sample is representative of Latin American economies. Although for more general conclusions for the whole set of emerging
economies it would be necessary a broader sample.
9. See Edwards and Vergara (2002) for a detailed discussion of some of
these parameters.
10. See for example Wessel (2007) and Dizard (2008).

11. As noted by Panizza (2008) in Latin American countries over the

19942005 period domestic public debt increased from 14% to 23% of
GDP. Meanwhile, average debt levels were decreasing from 72% to 62% of
GDP. The latter implies that the share of domestic debt over total public
debt went from 24% to 40%. Hence 60% of public debt is foreign currency
denominated; for the purposes of our simulations we used gures around
this level (55% and 70%).
12. In a sensitivity analysis they show an alternative calibration where a
10% point increase in the debt burden raises the risk premium by 150 basis
points. The main results do not change.
13. In a steady state, where Ddt = 0, then DPt/Pt = DDt/Dt = gt + pt,
or DDt/Yt = Dt/Yt(gt + pt). This implies that, to maintain the ratio of
public debt to GDP constant, the budget decit has to be equal to the
debt-to-GDP ratio times the nominal GDP growth rate.
14. Note that we are using the usual Latin American denition of RER,
this is international prices in domestic currency divided by the domestic
CPI. This diers from the usual IMF denition which is just the inverse.
Hence in our denition an increase in the RER is a real depreciation of the
domestic currency.
15. For further discussion of the eects of public debt on sovereign bond
yields, see Baldacci and Kumar (2010).
16. As Bordo and Meissner (2006) discuss,
matches matter for a nancial crisis , a strong
exports relative to hard currency liabilities
likelihood of a debt crisis, a currency crisis or a

although currency misreserve position or high

could help reduce the
banking crisis.

17. Barro (1979) and Dotsey (1994) argue that high debt can aect
growth via higher future distortionary taxation.
18. The sample is limited by the availability of country risk data for
Latin American countries. In a sensitivity analysis we excluded Brazil from
our sample. Estimates without Brazil are shown in Appendix tables.
19. The EMBI Plus measures the interest rate dierential between the
dollar bonds issued by governments and the US Treasury bonds. The
source is www.valorfuturo.cl.

1772

WORLD DEVELOPMENT

20. The unit root tests of our variables reject the null hypothesis that
variables are order-1 integrated.
21. For the ve countries considered in this paper, foreign public debt
represents about 50% of total public debt. For less developed countries,
the share of foreign debt is even higher.

26. Reinhart (2010) nd that average GDP growth falls 3% for external
debt levels above 60%. As they point out, In light of this, it is more
understandable that over one half of all defaults on external debt in
emerging markets since 1970 occurred at level of debt that would have met
the Maastricht criteria of 60% or less.
27. See Edwards and Vergara (2002) and Edwards (2002).

22. We also tested specications with nonlinear terms, but we did not
obtain dierent results.
23. Like in Blanchard and Perotti (2002), the discretionary responses are
assumed to take more than one quarter to appear and, therefore, they do
not capture the quarterly data from the series used.
24. See Perotti (2005) for an in-depth discussion of this subject.
25. In addition, we included, in Appendix gures, a set of rolling IRFs of
External Debt to Cholesky One S.D. RER innovation, in order to examine
more closely the sensitivity of external debt responses to sample variation.
In the recursive regressions, the initial sample is xed (rst 12 quarters)
and then observations are added one at a time. Appendix Figures A1A5
show snapshots of the rolling regression for dierent time periods (one,
three, ve, seven and 10 years). As the gure shows, results for most of the
samples point to an increase in public debt-to-GDP ratio that lasts for two
to three years.

b 0.
28. Remember that we assume that ER ER
29. This is clearly a simplifying assumption since the scal adjustment
can be expected to produce further changes in the exchange rate.
Nonetheless, although the scal adjustment obtained in this exercise is
large as compared to scal expenditure, it is relatively small as compared
to the overall aggregate demand of the economy. Hence, it is reasonable to
assume that the further eects over the real exchange rate will be rather
small.
30. Notice that neither growth nor the interest rate return to their
previous levels immediately after t years due to the presence of persistence
in these variables.
31. Even if the companies are not state-owned, there is an eect through
tax revenues obtained from private companies producing and exporting
commodities.

REFERENCES
Akitoby, B., & Stratmann, T. (2006). Fiscal policy and nancial markets.
International Monetary Fund working paper 06/16. <http://www.imf.org/external/pubs/ft/wp/2006/wp0616.pdf> Retrieved 16.12.11.
Arellano, M., & Bover, O. (1995). Another look at the instrumental
variable estimation of error-components models. Journal of Econometrics, 68, 2951.
Arnone, M., Bandiera, L., & Presbitero, A. (2008). Debt sustainability
framework in Hipcs: A critical assessment and suggested improvements.
<http://ssrn.com/abstract=871171> Retrieved 16.12.11.
Baldacci, E., & Kumar, M. (2010). Fiscal decits, public debt and sovereign
bond yields. International Monetary Fund working paper 10/184.
<http://www.imf.org/external/pubs/ft/wp/2010/wp10184.pdf>
Retrieved 16.12.11.
Barro, R. (1979). On the determinants of the public debt. Journal of
Political Economy, 85(5), 940971.
Blanchard, O., & Perotti, R. (2002). An empirical characterization of the
dynamic eects of changes in government spending and taxes on
output. The Quarterly Journal of Economics, 117(4), 13291368.
Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions
in dynamic panel data models. Journal of Econometrics, 87, 115143.
Bohn, H. (1998). The behavior of U.S. public debt and decits. Quarterly
Journal of Economics, 113(3), 949963.
Bohn, H. (2005). The sustainability of scal policy in the United States.
<http://www.econ.ucsb.edu/~bohn/papers/DebtUS.pdf>
Retrieved
16.12.11.
Bordo, M., & Meissner, C. (2006). The role of foreign currency debt in
nancial crises: 18801913 vs. 19721997. Journal of Banking Finance,
30(12), 32993329.
Calvo, G., & Reinhart, C. (2002). Fear of oating. Quarterly Journal of
Economics, 117(2), 379408.
Calvo, G., Izquierdo, A., & Talvi, E. (2003). Sudden stops, the real
exchange rate and scal sustainability: Argentinas lessons. NBER
working paper 9828. <http://www.nber.org/papers/w9828> Retrieved
16.12.11.
Cuddington, J. (1997). Analyzing the sustainability of scal decits in
developing countries. World Bank policy research working paper 1784.
<http://papers.ssrn.com/sol3/papers.cfm?abstract_id=597231>
Retrieved 16.12.11.
Cline, W. (1983). Interest and debt: Systematic risk and policy response.
Washington, DC: Institute for International Economics, MIT Press.
Dizard, J. (2008, October 21). Now the suering spreads to emerging
market banks. Financial Times. <http://www.ft.com/cms/s/0/

9da86a32-9f08-11dd-98bd-000077b07658.html#axzz1ghlaTg5q> Retrieved 16.12.11.

Dotsey, M. (1994). Some unpleasant supply side arithmetic. Journal of
Monetary Economics, 33(3), 507524.
Edwards, S. (1986). The pricing of bonds and bank loans in international
markets: An empirical analysis of developing countries foreign borrowing. NBER working paper 1689. <http://www.nber.org/papers/
w1689> Retrieved 16.12.11.
Edwards, S. (2002). Debt relief and scal sustainability. NBER working
paper 8939. <http://www.nber.org/papers/w8939> Retrieved 16.12.11.
Edwards, S., & Vergara, R. (2001). La Poltica Macroeconomica en
Nicaragua: Evaluacion y Opciones. Unpublished manuscript.
Edwards, S., & Vergara, R. (2002). Fiscal sustainability, debt dynamics and
debt relief: The cases of Nicaragua and Honduras. Unpublished paper.
Eichengreen, B., & Hausmann, R. (1999). Exchange rates and nancial
fragility. NBER working paper 7418. <http://www.nber.org/papers/
w7418> Retrieved 16.12.11.
Eichengreen, B., Hausmann, R., & Panizza, U. (2003). Currency
mismatches, debt intolerance and original sin: Why they are not the
same and why it matters. NBER working paper 10036. <http://
www.nber.org/papers/w10036> Retrieved 16.12.11.
Garcia, M., & Rigobon, R. (2004). A risk management approach to
emerging markets sovereign debt sustainability with an application to
Brazilian data. NBER working papers 10336. <http://www.nber.org/
papers/w10336.pdf > Retrieved 16.12.11.
Hausmann, R. (2004). Good credit ratios, bad credit ratings: The role of
debt structure. In G. Kopits (Ed.), Rules-based scal policy in emerging
markets: Background, analysis and prospects (pp. 3052). New York:
Palgrave Macmillan.
Hausmann, R., & Panizza, U. (2003). The determinants of original sin: An
empirical investigation. Journal of International Money and Finance,
22(7), 957990.
Hausmann, R., & Panizza, U. (2010). Redemption or abstinence? Original
sin, currency mismatches and counter-cyclical policies in the new
millennium. Center for International Development working paper
194. <http://www.hks.harvard.edu/centers/cid/publications/facultyworking-papers/cid-working-paper-no.-194> Retrieved 16.12.11.
Hausmann, R., Panizza, U., & Stein, E. (2001). Why do countries oat the
way they oat?. Journal of Developing Economies, 66, 387414.
Hostland, D., & Karam, P. (2006). Assessing debt sustainability in
emerging market economies using stochastic simulation methods. World
Bank policy research working paper series 3821. <http://www-

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

wds.worldbank.org/servlet/WDSContentServer/WDSP/IB/2006/01/
13/000016406_20060113143802/Rendered/PDF/wps3821.pdf>
Retrieved 16.12.11.
International Monetary Fund and World Bank (2002, December).
Guidelines for public debt management. <http://www.imf.org/external/
np/mfd/pdebt/2003/eng/am/120903.pdf> Retrieved 16.12.11.
International Monetary Fund and World Bank (2004, February). Debt
sustainability in low-income countries Proposal for an operational
framework and policy implications. <http://www.imf.org/external/np/
pdr/sustain/2004/020304.htm> Retrieved 16.12.11.
International Monetary Fund and World Bank (2010, January). Sta
guidance note on the application of the joint bank-fund debt sustainability
framework for low-income countries. <http://www.imf.org/external/np/
pp/eng/2010/012210.pdf> Retrieved 16.12.11.
Kumar, M., & Woo, J. (2010). Public debt and growth. International
Monetary Fund working paper 10/174. <http://www.imf.org/external/
pubs/ft/wp/2010/wp10174.pdf> Retrieved 16.12.11.
Lachler, U. (2001, April). Nicaragua: Fiscal sustainability after HIPC.
World Bank, Unpublished manuscript.
Mendoza, E., & Ostry, J. (2008). International evidence on scal solvency:
Is scal policy Responsible?. Journal of Monetary Economics, 55(6),
10811093.
Milesi-Ferretti, G., & Razin, A. (1996, October). Current-account sustainability. Princeton Studies in International Finance 81. <http://
www.princeton.edu/~ies/IES_Studies/S81.pdf> Retrieved 16.12.11.

1773

Milesi-Ferretti, G., & Razin, A. (2000). Current account reversals and

currency crises: Empirical regularities. In P. Krugman (Ed.), Currency
crises (pp. 285326). Chicago: University of Chicago Press.
Min, H. (1998). Determinants of emerging market bond spread: Do
economic fundamentals matter? World Bank policy research working
paper 1899. <http://www-wds.worldbank.org/servlet/WDSContentServer/WDSP/IB/1998/03/01/000009265_3980429111506/Rendered/
PDF/multi0page.pdf> Retrieved 16.12.11.
Panizza, U. (2008). Domestic and external public debt in developing
countries. UNCTAD discussion papers 188. <http://www.unctad.org/
en/docs/osgdp20083_en.pdf> Retrieved 16.12.11.
Penalver, A., & Thwaites, G. (2006). Fiscal rules for debt sustainability in
emerging markets: The impact of volatility and default risk. Bank of
England working paper 307. <http://www.bankofengland.co.uk/publications/workingpapers/wp307.pdf> Retrieved 16.12.11.
Perotti, R. (2005). Estimating the eects of scal policy in OECD countries.
CEPR discussion paper 4842. <http://www.cepr.org/pubs/new-dps/
dplist.asp?dpno=4842> Retrieved 16.12.11.
Reinhart, C., & Rogo, K. (2010). Growth in a time of debt. NBER
working paper 15639. <http://www.nber.org/papers/w15639> Retrieved 16.12.11.
Wessel, D. (2007, May 3). How an emerging-market trend may ease
volatility. The Wall Street Journal. <http://online.wsj.com/article/
SB117815149999390238.html?mod=todays_us_page_one> Retrieved
16.12.11.

1774

WORLD DEVELOPMENT

APPENDIX A
See Figures A1A5 and Tables 613.

Figure A1. Brazil rolling IRFs of External Debt to Cholesky One S.D. RER innovation at one, three, ve, seven, and 10 years.

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

Figure A2. Chile rolling IRFs of External Debt to Cholesky One S.D. RER innovation at one, three, ve, seven, and 10 years.

1775

1776

WORLD DEVELOPMENT

Figure A3. Colombia rolling IRFs of External Debt to Cholesky One S.D. RER innovation at 1 quarter, one, three, ve, seven, and 10 years.

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

Figure A4. Mexico rolling IRFs of External Debt to Cholesky One S.D. RER innovation at one, three, ve, seven, and 10 years.

1777

1778

WORLD DEVELOPMENT

Figure A5. Peru rolling IRFs of External Debt to Cholesky One S.D. RER innovation at 1 quarter, one, three, ve, seven, and 10 years.

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

1779

Table 6. Estimates using xed eects model and System GMM Excluding Brazil
Dependent variable: interest rate spread (qt)

Eqn. (1)

qt1
(External public debt/GDPt)
RERt
GDP growth ratet
Time eect
Observations
R2
Number of id
Serial correlation order 1
Serial correlation order 2
Sargan p

0.219***
(0.0252)
0.00259
(0.0110)
0.0955**
(0.0430)
Yes
122
0.910
4

Eqn. (2)

Eqn. (3)

Eqn. (4)

0.782***
(0.0942)
0.0642**
(0.0191)

0.813***
(0.0753)
0.0311**
(0.00895)
0.0133
(0.00703)

Yes

Yes

0.693***
(0.0476)
0.0548***
(0.0116)
0.00144
(0.00781)
0.0892**
(0.0252)
Yes

120

120

120

4
0.134
0.348
0.958

4
0.0775
0.189
0.123

4
0.0818
0.217
0.132

Standard errors in parentheses.

***
p < 0.01.
**
p < 0.05.

Table 7. Estimates using Dierence GMM Excluding Brazil

Dependent variable: GDP growth ratet
GDP growth ratet1
(External public debt/GDPt)over30%
(External public debt/GDPt)below30%
RERt
rt
Time eect
Observations
Number of id
Serial correlation order 1
Serial correlation order 2
Sargan p value
Robust standard errors in parentheses.
***
p < 0.01.
**
p < 0.05.
*
p < 0.1.

Eqn. (1)

Eqn. (2)

***

***

Eqn. (3)

0.573
(0.0373)
0.0636*
(0.0245)
0.0516
(0.125)
0.00879
(0.0806)
0.479
(0.237)
Yes

0.710
(0.0760)
0.0672***
(0.00777)
0.202
(0.0996)
0.0396
(0.0337)

0.737***
(0.0669)
0.0628**
(0.0169)
0.180
(0.103)

Yes

Yes

121
4
0.0534
0.423
0.109

122
4
0.118
0.895
0.168

122
4
0.0997
0.878
0.147

1780

WORLD DEVELOPMENT
Table 8. Parameter values used in the scal sustainability analysis

Parameter

Symbol

Assumed value

Desired (and initial) external

public debt to GDP ratio

d*

Initial foreign currency denominated

debt as percentage of total debt

(e0F0/D0)

Real interest rate when d = d*

r0

5.0%

Parameter that relates interest rate

in t with interest rate in t  1 (persistence)
Parameter that relates changes in
interest rates with debt changes

a2

0.693

a3

0.0548

Rate of real GDP growth when d = d*

g0

4%

Parameter that relates GDP growth rate

in t with GDP growth rate in t  1
Parameter that relates changes in growth with debt changes

b2

0.573

b3

0.0636

Comments and sources

35%

Time necessary to come back to the

desired public debt ratio after the devaluation

t

Seigniorage

S/Y

Common gure used for sustainability

analyses based on demand for public sector
debt (see Edwards & Vergara, 2002)
Fraction of public debt that is in
foreign currency
From the actual data of LA countries
It is assumed to be equal to
r* + q = rFC
SGMM estimator in
autoregressive panel data model
SGMM estimator in
autoregressive panel data model

5570%

Simulation can be made using

dierent rates of growth
SGMM estimator in
autoregressive panel data model
SGMM estimator in
autoregressive panel data model

Between 1 and 10

According to impulse response

functions
Consistent with a monetary base of 6% of
GDP and a nominal GDP growth rate of 7%

0.4%

Table 9. Simulations using parameters estimated excluding Brazil Real devaluation of 25% initial foreign currency-denominated debt as a percentage of total
debt: 55%
Panel A
Primary balance

Panel B
Public debt (% of GDP)

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

0.05
1.44
0.05
0.03
0.04
0.05
0.05
0.05
0.05
0.05
0.05

0.05
0.75
0.77
0.59
0.02
0.04
0.05
0.05
0.05
0.05
0.05

0.05
0.61
0.72
0.62
0.48
0.34
0.01
0.04
0.05
0.05
0.05

0.05
0.56
0.70
0.64
0.54
0.43
0.33
0.23
0.02
0.04
0.05

0.05
0.51
0.68
0.65
0.58
0.51
0.43
0.36
0.29
0.21
0.15

0
1
2
3
4
5
6
7
8
9
10

39.8
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

39.8
38.2
36.6
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

39.8
38.9
37.9
36.9
36.0
35.0
35.0
35.0
35.0
35.0
35.0

39.8
39.1
38.4
37.8
37.1
36.4
35.7
35.0
35.0
35.0
35.0

39.8
39.3
38.9
38.4
37.9
37.4
36.9
36.4
36.0
35.5
35.0

Panel C
Real interest rate

Panel D
Rate of real GDP growth

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

5.00
5.41
5.12
5.03
5.01
5.00
5.00
5.00
5.00
5.00
5.00

5.00
5.64
5.59
5.31
5.08
5.02
5.00
5.00
5.00
5.00
5.00

5.00
5.69
5.75
5.60
5.41
5.20
5.05
5.01
5.00
5.00
5.00

5.00
5.71
5.82
5.73
5.59
5.45
5.29
5.14
5.04
5.01
5.00

5.00
5.72
5.87
5.83
5.73
5.63
5.53
5.42
5.31
5.21
5.10

0
1
2
3
4
5
6
7
8
9
10

4.0
3.7
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.5
3.6
3.9
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.4
3.5
3.6
3.8
3.9
4.0
4.0
4.0
4.0
4.0

4.0
3.4
3.4
3.5
3.6
3.7
3.8
3.9
4.0
4.0
4.0

4.0
3.4
3.4
3.5
3.5
3.6
3.7
3.7
3.8
3.9
4.0

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

1781

Table 10. Simulations using parameters estimated excluding Brazil real devaluation of 25% initial foreign currency-denominated debt as a percentage of total
debt: 70%
Panel A
Primary balance

Panel B
Public debt (% of GDP)

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

0.05
1.85
0.08
0.02
0.04
0.05
0.05
0.05
0.05
0.05
0.05

0.05
0.98
1.01
0.76
0.03
0.03
0.05
0.05
0.05
0.05
0.05

0.05
0.81
0.94
0.81
0.63
0.45
0.00
0.04
0.05
0.05
0.05

0.05
0.74
0.92
0.84
0.71
0.57
0.44
0.31
0.01
0.04
0.05

0.05
0.68
0.90
0.86
0.77
0.67
0.57
0.47
0.38
0.29
0.20

0
1
2
3
4
5
6
7
8
9
10

41.1
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

41.1
39.1
37.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

41.1
39.9
38.7
37.5
36.2
35.0
35.0
35.0
35.0
35.0
35.0

41.1
40.3
39.4
38.5
37.6
36.8
35.9
35.0
35.0
35.0
35.0

41.1
40.5
39.9
39.3
38.7
38.1
37.5
36.8
36.2
35.6
35.0

Panel C
Real interest rate

Panel D
Rate of real GDP growth

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

5.00
5.53
5.16
5.04
5.01
5.00
5.00
5.00
5.00
5.00
5.00

5.00
5.82
5.74
5.39
5.11
5.03
5.01
5.00
5.00
5.00
5.00

5.00
5.87
5.95
5.77
5.52
5.25
5.07
5.02
5.00
5.00
5.00

5.00
5.90
6.04
5.93
5.76
5.57
5.37
5.18
5.05
5.01
5.00

5.00
5.92
6.10
6.05
5.94
5.80
5.67
5.53
5.40
5.26
5.13

0
1
2
3
4
5
6
7
8
9
10

4.0
3.6
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.3
3.5
3.8
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.3
3.4
3.5
3.7
3.9
4.0
4.0
4.0
4.0
4.0

4.0
3.2
3.3
3.4
3.5
3.7
3.8
3.9
4.0
4.0
4.0

4.0
3.2
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
3.9

Table 11. Parameter values used in the scal sustainability analysis

Parameter

Symbol

Desired (and initial) external

public debt to GDP ratio

Initial foreign currency denominated

debt as percentage of total debt

(e0F0/D0)

Real interest rate when d = d*

Parameter that relates changes in
interest rates with debt changes

r0
a3

5.0%
0.053

Rate of real GDP growth when d = d*

g0

4%

Parameter that relates changes in

growth with debt changes
Time necessary to come back to the desired
public debt ratio after the devaluation
Seigniorage

b3

0.04

t
S/Y

Assumed value

Comments and sources

35%

Common gure used for sustainability

analyses based on demand for public sector
debt (see Edwards & Vergara, 2002)
Fraction of public debt that is in foreign currency
From the actual data of LA countries

5570%

Between 1 and 10
0.4%

It is assumed to be equal to r* + q = rFC

Baldacci and Kumar (2010)
Simulation can be made using dierent
rates of growth
Kumar and Woo (2010)
According to impulse response functions
Consistent with a monetary base of 6% of
GDP and a nominal GDP growth rate of 7%

1782

WORLD DEVELOPMENT

Table 12. Simulations using parameters from BaldacciKumar and KumarWoo Real devaluation of 25% initial foreign currency-denominated debt as a
percentage of total debt: 55%
Panel A
Primary balance

Panel B
Public debt (% of GDP)

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

0.05
1.28
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05

0.05
0.54
0.46
0.39
0.05
0.05
0.05
0.05
0.05
0.05
0.05

0.05
0.39
0.35
0.30
0.26
0.21
0.05
0.05
0.05
0.05
0.05

0.05
0.33
0.30
0.26
0.23
0.20
0.17
0.14
0.05
0.05
0.05

0.05
0.29
0.26
0.24
0.21
0.19
0.17
0.15
0.12
0.10
0.08

0
1
2
3
4
5
6
7
8
9
10

39.8
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

39.8
38.2
36.6
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

39.8
38.9
37.9
36.9
36.0
35.0
35.0
35.0
35.0
35.0
35.0

39.8
39.1
38.4
37.8
37.1
36.4
35.7
35.0
35.0
35.0
35.0

39.8
39.3
38.9
38.4
37.9
37.4
36.9
36.4
36.0
35.5
35.0

Panel C
Real interest rate

Panel D
Rate of real GDP growth

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

5.00
5.06
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00

5.00
5.19
5.11
5.02
5.00
5.00
5.00
5.00
5.00
5.00
5.00

5.00
5.22
5.17
5.11
5.06
5.01
5.00
5.00
5.00
5.00
5.00

5.00
5.23
5.19
5.15
5.12
5.08
5.05
5.01
5.00
5.00
5.00

5.00
5.24
5.21
5.18
5.16
5.13
5.11
5.08
5.06
5.03
5.01

0
1
2
3
4
5
6
7
8
9
10

4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.9
3.9
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.8
3.9
3.9
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.8
3.9
3.9
3.9
3.9
4.0
4.0
4.0
4.0
4.0

4.0
3.8
3.8
3.9
3.9
3.9
3.9
3.9
4.0
4.0
4.0

FISCAL SUSTAINABILITY: THE IMPACT OF REAL EXCHANGE RATE SHOCKS

1783

Table 13. Simulations using parameters from BaldacciKumar and KumarWoo Real devaluation of 25% initial foreign currency-denominated debt as a
percentage of total debt: 70%
Panel A
Primary balance

Panel B
Public debt (% of GDP)

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

0.05
1.64
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05

0.05
0.71
0.61
0.51
0.05
0.05
0.05
0.05
0.05
0.05
0.05

0.05
0.52
0.46
0.40
0.34
0.29
0.05
0.05
0.05
0.05
0.05

0.05
0.44
0.40
0.35
0.31
0.27
0.23
0.19
0.05
0.05
0.05

0.05
0.38
0.35
0.32
0.29
0.26
0.23
0.20
0.17
0.14
0.12

0
1
2
3
4
5
6
7
8
9
10

41.1
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

41.1
39.1
37.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0
35.0

41.1
39.9
38.7
37.5
36.2
35.0
35.0
35.0
35.0
35.0
35.0

41.1
40.3
39.4
38.5
37.6
36.8
35.9
35.0
35.0
35.0
35.0

41.1
40.5
39.9
39.3
38.7
38.1
37.5
36.8
36.2
35.6
35.0

Panel C
Real interest rate

Panel D
Rate of real GDP growth

Period

t
3

10

Period

t
3

10

0
1
2
3
4
5
6
7
8
9
10

5.00
5.08
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00

5.00
5.24
5.13
5.03
5.00
5.00
5.00
5.00
5.00
5.00
5.00

5.00
5.27
5.21
5.15
5.08
5.02
5.00
5.00
5.00
5.00
5.00

5.00
5.29
5.24
5.20
5.15
5.10
5.06
5.01
5.00
5.00
5.00

5.00
5.30
5.27
5.23
5.20
5.17
5.14
5.11
5.07
5.04
5.01

0
1
2
3
4
5
6
7
8
9
10

4.0
3.9
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.8
3.9
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.8
3.8
3.9
3.9
4.0
4.0
4.0
4.0
4.0
4.0

4.0
3.8
3.8
3.9
3.9
3.9
4.0
4.0
4.0
4.0
4.0

4.0
3.8
3.8
3.8
3.8
3.9
3.9
3.9
3.9
4.0
4.0