      Orderly Sovereign Debt Restructuring:
                Missing in Action!
            (And Likely To Remain So)


                       Otaviano Canuto, Brian Pinto, and Mona Prasad*


An orderly sovereign debt restructuring should place the debtor nation’s public debt on a
sustainable trajectory while minimizing procrastination and contagion. However, the expe-
riences with the debt crisis of the 1980s, Russia 1998, Argentina 2001, and Greece
2010 indicate that orderly debt restructurings remain elusive, even with high-powered
ofﬁcial intervention. When solvency problems are present, the chances of success increase
if ofﬁcial money is lent at the risk-free rate, reﬂecting its low risk, and if private creditors
receive an upfront haircut. The paper examines the obstacles, which include moral hazard,
difﬁculty in distinguishing between solvency and liquidity crises, and the “political
economy” resistance to upfront haircuts. Orderly sovereign debt restructurings are likely
to remain elusive notwithstanding recent evidence that the ofﬁcial mindset may be chang-
ing. Sovereign Debt, Debt Restructuring, Solvency, Liquidity, Seniority, Sector Board,
Economic Policy (EPOL). JEL codes: E61, E65, F34



Introduction
This much we know from the experience of emerging market countries from the
1980s onwards: except for a few small cases, sovereign debt restructurings have
tended to be costly and chaotic, with orderly sovereign debt restructuring seemingly
impossible to achieve. This holds even when high-proﬁle ofﬁcial intervention occurs.
We explore why this happens and identify the impediments to an orderly debt
restructuring (ODR).
   The ﬁnding from emerging markets (EMs) about ODRs being elusive carries over
to Greece 2010. In March 2010, it became clear that Greece’s ﬁscal fundamentals

The World Bank Research Observer
# The Author 2014. Published by Oxford University Press on behalf of the International Bank for Reconstruction and
Development / THE WORLD BANK. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
doi:10.1093/wbro/lkt020          Advance Access publication January 13, 2014                                 29:109–135
were weak and that it would need ofﬁcial assistance to avoid a default. Discussions
on a bailout began. A counterintuitive feature was that market sentiment worsened
as negotiations proceeded between Greece and the European Union-European
Central Bank-IMF troika. The spread on Greece’s bonds rose signiﬁcantly, though
the size of the bailout package was upped substantially. Remarkably, the two-year
bond spread shot up, though Greece was effectively being “taken out of the market”,
that is, the announced bailout funds were more than enough to pay off maturing
short-term private creditors in full. This situation suggested deep-seated market
skepticism about solvency and the feasibility of the ﬁscal program accompanying
the bailout.1
   Notwithstanding these negative market signals about solvency, the authorities
made it abundantly clear that any haircut for Greece’s private creditors was out of
the question. The troika’s gamble was that structural and ﬁscal reforms would
restore Greece to a sustainable debt path that would lower interest rates to non-
default levels without a debt write down, which was believed would have costly con-
tagion effects.
   Barely a year later, in July 2011, the ofﬁcial position reversed dramatically. By
that time, contagion from Greece, Ireland, and Portugal (the latter two countries
had also received ofﬁcial bailouts by then) had begun to spread to the core of the
European Union (EU). A July 21, 2011 Eurozone summit announced support for a
haircut for Greece’s private creditors while also agreeing to a major softening of
loan terms to bring ofﬁcial EU lending rates closer to the risk-free rate while length-
ening maturities signiﬁcantly. A subsequent summit in October 2011 announced
that private Greek bondholders would receive a 50 percent write down on principal
and that the European Financial Stability Facility (EFSF) would be leveraged to E1
trillion to support Italy and Spain. Stock markets reacted euphorically, but Greece
announced and then withdrew a referendum on the bailout toward the end of
2011, while Italian 10-year bond yields approached the 7 percent threshold at
which other countries had been bailed out. The ECB saved the day by injecting
liquidity via two tranches of a Long-Term Reﬁnancing Operation (LTRO). This lent
commercial banks huge sums at 1 percent for 3 years, which they used to buy
Spanish and Italian sovereign bonds, lowering their yields substantially and
boosting conﬁdence.2
   In February 2012, the EU approved a second bailout for Greece amounting to
E130 billion but required a PSI ( private sector involvement) debt exchange, which
inﬂicted losses of some 70 percent in net present value (NPV) terms on E197 billion
in privately-held debt, equivalent to approximately 97 percent of the projected 2012
Greek GDP   . Even so, the government debt-to-GDP ratio under the program assump-
tions at the time was expected to fall to only 120 percent by 2020. The subsequent
bond price movements indicated that the debt deal had done little to alter market per-
ceptions about Greece’s credit standing. Figure 1 plots the 10-year Greek bond price

110                                    The World Bank Research Observer, vol. 29, no. 1 (February 2014)
Figure 1. Greece’s 10-Year Bond Prices and Spreads, January 2010 – May 2012




  Source: Bloomberg, IMF Staff Reports



and its spread from January 1, 2010 to the end of May 2012, noting key events.
Subsequent developments in the Eurozone are summarized in Section 3.
   As the Greek crisis indicates, sovereign debt restructuring is complicated, ofﬁcial
intervention notwithstanding. Does this mean that ofﬁcial intervention does more
harm than good? It is difﬁcult to answer this question conclusively because of the
difﬁculty in developing a counterfactual. The major lesson from the debt overhang
of the 1980s (Krugman 1988; Sachs 1986) is that such intervention is needed to
solve coordination and free-rider problems among creditors.
   Apart from the inherent complexity in sovereign debt restructuring, two reasons
may explain why ofﬁcial intervention (OI) tends not to work well in prominent sov-
ereign debt crises, such as the debt crisis of the 1980s in Latin America, Russia in
1998, Argentina in 2001, or Greece in 2010. The ﬁrst is a seeming inability to dis-
tinguish between liquidity and solvency crises, that is, situations where a sovereign
may be unable to rollover maturing debt versus one where the debt has simply
become too large to be serviced. Although the catalytic effect of ofﬁcial ﬁnance may

Canuto et al.                                                                     111
work well in persuading short-term creditors to roll over their loans in countries
with acceptable fundamentals (as in Morris and Shin 2006), these elegant results
tend to break down once one acknowledges that ofﬁcial loans may be senior to
private loans and that the country is facing a solvency instead of a liquidity problem
(Kharas, Pinto, and Ulatov 2001; Chamley and Pinto 2011). The second reason is
legal impediments to a smooth bankruptcy process for sovereigns.3 Our focus will
be on the basic economics of and political obstacles to an ODR.
   Section 2 presents the motivation for ODRs based on a survey of EM sovereign
debt restructuring and the part played by ofﬁcial intervention. This is followed by a
discussion of procrastination in sovereign debt restructuring in Section 3. Section 4
builds on Sections 2 and 3 to tease out the desirable attributes of an ODR. However,
the track record inevitably raises a question about the feasibility of an ODR. Hence,
Section 5 discusses the obstacles, which include political economy and the difﬁculty
in distinguishing between liquidity and solvency problems for countries. Section 6
concludes.



Context and Motivation for ODRs
The sovereign debt literature concerns itself with fundamental questions, such as
why sovereign debt exists in the ﬁrst place, considering difﬁculties in enforcing con-
tracts; why default by a sovereign does not mean permanent exclusion from future
borrowing; and why countercyclical ﬁscal policy (saving during good times, deplet-
ing accumulated saving during bad times) and self-insurance against shocks
cannot substitute for borrowing. An excellent survey is presented in chapter 2 of
Sturzenegger and Zettelmeyer (2006), which also contains a concise account of the
seminal papers.
   Our goal is different. We want to review the empirical experience with sovereign
debt restructuring since the landmark EM debt crisis of 1980s and use this as a
forward-looking platform for discussing the desirable attributes of an ODR. The ﬁrst
point to note is that the vast bulk of EM sovereign debt restructurings since the
1980s have involved private creditors (Table 1). Some US$325 billion in principal
owed to private creditors has been restructured, compared to just US$29 billion
with ofﬁcial creditors via the Paris Club.4
   In contrast, ofﬁcial (including bilateral and multilateral, such as IMF, World
Bank, International Development Association (IDA), African Development Fund)
creditors have accounted for the lion’s share of sovereign debt restructurings for
low-income countries, which typically have limited access to the international
capital markets. As of February 15, 2012, the Paris Club had treated debt amount-
ing to US$556 billion for 88 developing countries under 423 agreements.5
Multilateral creditors have provided debt relief through the Heavily Indebted Poor

112                                   The World Bank Research Observer, vol. 29, no. 1 (February 2014)
Table 1. Sovereign External Debt Restructurings with Private Creditors – 1980s and After
Plan/Country                                                                  Amount restructured (in US$ billion)

Brady Plan (1989)                                                                           200
Russia London Club (2000)                                                                    32
Argentina (2005 & 2010)                                                                      76
Ukraine (2000)                                                                                2.3
Uruguay (2003)                                                                                5.1
Others                                                                                        7.2
TOTAL                                                                                       322.6
   Sources: World Bank (1998), Chuhan and Sturzenegger (2005), Kharas et al. (2001), Paris Club (www.clubdeparis.
org).
   Notes: For the 1980s, the 1985 Baker Plan is not included because the restructured debt amounts are subsumed
under the Brady Plan. The Russian and Argentine pre-crisis swaps are not included but are discussed below. US$6
billion of defaulted debt owed to Argentina’s private creditors is still unresolved.



Countries (HIPC) Initiative and the Multilateral Debt Relief Initiative (MDRI).6
However, this initiative is available only to low-income countries, and eligibility cri-
teria are restrictive. Given the eligibility requirements for HIPC and MDRI, none of
the EMs has beneﬁtted from multilateral debt restructurings.
   The rest of the review considers the origins of debt crises and goes on to the role
of ofﬁcial intervention. The 1980s experience showed that it is difﬁcult to design
efﬁcient ofﬁcial intervention. This was conﬁrmed by the subsequent experience
with Russia in 1998 and Argentina in 2001, which illuminated another important
issue: why ofﬁcial intervention may not be catalytic in terms of persuading private
creditors to roll over their loans.


Origins of Debt Crises
One set of constants has marked all serious EM debt crises since the 1980s: ﬁxed
exchange rates, open capital accounts, weak growth prospects, and concerns about
ﬁscal solvency.7 Fiscal fundamentals play a crucial role, either at the outset or even-
tually, as a result of bailing out the domestic private sector. In addition, though the
crisis itself typically involves an abrupt economic disruption, its seeds tend to be
sown over long periods, reﬂecting policy and political economy.
   Heavy external borrowing preceded the 1980s debt crisis. Such borrowing may
have been motivated by the need to ﬁnance development, sometimes via ill-advised
public investments; by social spending needs; and even by the desire to enrich well-
connected groups.8 Money-center banks were happy to roll over maturing principal
and even interest payments because the key creditworthiness indicator at that time
was the external debt-to-exports ratio, and nominal export prices in dollars contin-
ued to rise faster than the nominal interest rate, keeping this ratio under control.

Canuto et al.                                                                                                113
Sachs notes (1990, p 8), “During the heady days of the 1970s. . . . .countries and
their banks had the illusion of an unending Ponzi game . . .”. Eventually, with their
terms-of-trade declining sharply in the early 1980s along with the record rise in
interest rates in the US—a combination we shall refer to as the “twin shocks”—
the bubble burst, and countries now had to service their debt the old-fashioned
way: by generating current account surpluses to pay down their debt.9 This meant
politically unpalatable ﬁscal austerity and cuts in real wages.
   Three complications frequently arose. First, with ﬁxed pegs to the dollar the
norm, the private sector began speculating against their home currencies once they
realized that the exchange rate was becoming overvalued. This led the government
and central bank to borrow overseas in support of the peg. The acceleration of
private capital ﬂight exacerbated the eventual public debt burden while exerting
ruinous effects on domestic banks and the ﬁnancial system.
   Second, some central banks imposed restrictions on convertibility in an effort to
prevent foreign exchange reserve depletion, leading to a high black market
premium on foreign exchange. In this milieu, foreign banks were reluctant to keep
rolling over loans, forcing governments to switch to monetary ﬁnancing of the ﬁscal
deﬁcit. Furthermore, the rate of inﬂation to generate a given amount of seigniorage
for ﬁnancing the ﬁscal deﬁcit went up as the population’s ability to shift into dollars
raised the inﬂation elasticity of domestic money demand.10
   Third, inﬂation may have become entrenched as a result of the indexation of
wages and asset prices, as in Brazil during the 1970s and 1980s, making extrica-
tion from high inﬂation all the more difﬁcult. Not surprisingly, the major Latin
American countries entered a rut of repeated failures in stabilization, ever higher
public debt, and severe costs for growth and economic welfare, especially for vulner-
able groups.11
   The link between stabilization programs and debt crises provides a natural bridge
from the 1980s debt crises to those of Russia in 1998 and Argentina in 2001.
Russia achieved single-digit inﬂation in early 1998 but suffered a devastating triple
exchange rate-public debt-banking sector crisis less than six months later. This
1998 crisis had echoes in the 2001 crisis in Argentina. Both involved ﬁxed
(managed in the case of Russia, constitutionally mandated in the case of Argentina)
pegs to the dollar, which had been chosen to squeeze inﬂation out; both eventually
developed unsustainable debt dynamics (masked by real appreciation of the ex-
change rate in conjunction with a signiﬁcant share of public debt denominated in
dollars); and in both cases, banks became vulnerable to sovereign risk. In addition,
Argentina’s banks became vulnerable to currency mismatches, having borrowed in
US dollars but on-lent to companies with local currency revenues. The net result
was a downgrading of growth prospects and a rise in interest rates, which eventual-
ly fueled a meltdown. We shall not go into the details of these crisis episodes, which



114                                    The World Bank Research Observer, vol. 29, no. 1 (February 2014)
have been well-documented elsewhere, but we use these as a springboard for a dis-
cussion of the implications for sovereign debt restructuring later in the paper.12


Ofﬁcial Intervention: Insights from the 1980s
Following Mexico’s announcement in August 1982 that it could no longer service
its external commercial bank loans, 27 countries owing US$239 billion had either
rescheduled their bank loans or were engaged in doing so by October 1983. Sixteen
were from Latin America; of these, the four largest debtors, Mexico, Brazil, Venezuela,
and Argentina, owed US$176 billion, or 74 percent of the total outstanding EM
debt.13 Although it was evident by 1985 that the debtor countries were not recover-
ing, debt reduction remained politically unacceptable. Instead, the Baker Plan,
named after US Treasury James Baker, was launched in October 1985. It emphasized
new lending from commercial banks in exchange for market-based reforms. The 10
Baker Plan agreements rescheduled US$165 billion of debt. The World Bank was
expected to play a large role with “structural adjustment loans” in helping implement
the market-based reforms. However, the plan did not work, and with the decade
being inexorably lost, the US government ﬁnally threw its weight behind debt reduc-
tion. The Brady Plan was announced by US Treasury Secretary Nicholas Brady in
March 1989, with Mexico becoming the ﬁrst major test. In total, US$60 billion of
debt was forgiven, and US$200 billion of bank claims were converted into US$154
billion of Brady bonds.14
   From the perspective of achieving an ODR, three questions stand out: (i) Is ofﬁcial
intervention needed? (ii) When is ofﬁcial intervention most likely? and (iii) does ofﬁ-
cial intervention help? We consider each question in turn.
   Is ofﬁcial intervention needed? Krugman (1994, 710) noted with regard to the
Brady Plan that the idea of a voluntary approach was soon dropped, and a “combi-
nation of legal maneuvering and pressure on banks” left no option but to partici-
pate in a debt reduction program. The notion that a voluntary approach would not
work is intuitively plausible; no creditor would willingly write down their claims,
and this could scuttle what is a collectively superior outcome. This understanding is
reinforced by the events during the Baker Plan intervention. Though commercial
banks were supposed to come up with new money, Krugman (1994) and Dooley
(1994) note that the main outcome was that they managed to substantially reduce
their exposure to the debtor countries over the course of the 1980s and that loans
from ofﬁcial creditors rose sharply. Therefore, ofﬁcial intervention is needed for an
ODR because otherwise an impasse would result.
   When is ofﬁcial intervention most likely? The economic self-interest of the more
advanced and inﬂuential countries plays a powerful role. The immediate response
after Mexico’s default announcement in 1982 was to provide government-to-
government bridge loans so that debtor countries could remain current on their

Canuto et al.                                                                      115
interest payments and thereby avoid imperiling the US banking system. The four
largest debtors, Mexico, Brazil, Venezuela, and Argentina, owed $37 billion to the
eight largest US banks, which amounted to 147 percent of these banks’ capital
and reserves. At the time, the US was in a deepening recession, and there were
widespread fears that a debt default could trigger another Great Depression. The
bridge loans eventually led to the Baker Plan. The need for concerted lending
stemmed from the “free-rider” problem; each bank on its own would have pre-
ferred to reduce its exposure, but every bank doing so might have forced a default
and, eventually, an international ﬁnancial meltdown. Solving this collective action
problem provided the rationale for intervention by the US.
   Once the need for debt reduction was recognized, a beneﬁcial solution for both
debtors and creditors would have been for the commercial banks to voluntarily
accept a haircut in the hope of lowering the probability of default and raising the
market value of the remaining debt as per the debt overhang argument (Sachs
1986; Krugman 1988). In other words, indebtedness had reached a crippling level
which, in the event of a default, would lead to such a low recovery value for credi-
tors that it would be better for them to forgive part of the debt and recover a larger
amount as the country resumed growing and restored solvency—an outcome that
would be facilitated by the partial debt forgiveness. However, such forgiveness was
unlikely to happen spontaneously because individual creditors would be tempted to
hold out under a voluntary scheme to gain on their entire holding of the country’s
debt. Solving this free-rider problem provided the rationale for the Brady Plan.
Thus, the economic arguments for ofﬁcial intervention—solving the collective
action problem and lowering transactions costs—are clear. However, the question
remains whether such intervention will materialize without the interests of inﬂuen-
tial countries being at stake.
   Does ofﬁcial intervention help? Dooley (1994) conjectures, “It is difﬁcult to rule
out the possibility that all the direct beneﬁts of the Brady deals to date went to the
banks. Moreover, it is generally agreed that the direct beneﬁts of Brady restructur-
ings have been too small to account for much of the increase in the secondary
market prices since 1990”. The speciﬁc case of Mexico is insightful. While acknowl-
edging the ofﬁcial arm-twisting needed for debt reduction, Krugman nevertheless
notes (1994, 702), “Mexico achieved a reduction in the present value of its debt of
approximately $14 billion or 14 percent. This was clearly insufﬁcient . . .”.15
   Krugman (1994) describes the outcome of Baker and Brady as exhibiting two
features: ﬁnancial stability was maintained, but the debtor countries did badly. He
cites Cline (1990) as arguing that the debtor countries were going to do badly on
growth anyway, so saving the ﬁnancial system was a signal success. However,
Eichengreen and Portes (1989) showed that countries willing to default early and
massively during the 1930s crisis did better than those not willing to do so. This
eventuality was pre-empted by the Baker and Brady Plans during the 1980s.

116                                   The World Bank Research Observer, vol. 29, no. 1 (February 2014)
Ofﬁcial Intervention: Insights from 1997– 2001
Turning to the last round of EM crises during 1997 –2001, the experiences of
Russia in 1998 and Argentina in 2001 clearly show that the intervention of the
IFIs in situations of low foreign exchange reserves and unsustainable debt dynamics
carries a serious risk of prolonging crisis, eventually requiring the country to
address a much larger debt problem. This is what Rogoff (2003), former chief econ-
omist of the IMF, had to say about the Russian rescue package of 1998:
   As a result, the ofﬁcial lending community, typically led by the IMF , is often unwilling
to force the issue and sometimes ﬁnds itself trying to keep a country aﬂoat far beyond the
point of no return. In Russia in 1998, for example, the ofﬁcial community threw money
behind a ﬁxed exchange-rate regime that was patently doomed. Eventually, the Fund cut
the cord and allowed a default, proving wrong those many private investors who thought
Russia was “too nuclear to fail.” But if the Fund had allowed the default to take place at an
earlier stage, Russia might well have come out of its subsequent downturn at least as
quickly and with less ofﬁcial debt.
   Interestingly, Rogoff (2003) attributes procrastination in Russia 1998 to the fact
that “. . . .current international law makes bankruptcies by sovereign states extraor-
dinarily messy and chaotic”. However, the analysis in Kharas, Pinto, and Ulatov
(2001) points to mistakes in diagnosis that resulted in a rescue package that
emphasized liquidity over solvency and eventually led to a much larger problem.16
   In its post-mortem of Argentina 2001, the IMF’s Independent Evaluation Ofﬁce
noted that ofﬁcial rescue packages are unlikely to be catalytic in insolvency situa-
tions, that ﬁnancial engineering in the form of voluntary debt swaps is ineffective,
and that procrastination is costly.17 These were very much the lessons from Russia
1998. The ﬁrst two lessons, on why rescue packages may not be catalytic and the
inefﬁcacy of sovereign debt swaps, are discussed below. The third, on procrastina-
tion, is discussed in Section 3.
   Ofﬁcial intervention that is catalytic could mean one or all of three things: (i)
private holders of government bonds are persuaded not merely to rollover maturing
loans but also to increase their exposure to the debtor government; (ii) the govern-
ment implements ﬁscal and structural reform as part of the ofﬁcial rescue package
that places government debt on a sustainable trajectory and improves growth pros-
pects; and (iii) interest rates come down because risk spreads relative to the bench-
mark country (for example, the US or Germany) decline.
   It should be obvious that all three would be much easier to achieve if the gov-
ernment faced a liquidity but not a solvency problem. In fact, one could argue
that in the case of a pure liquidity problem, only (i) would be needed as part of the
catalytic effect (as in Morris and Shin 2006). However, in a solvency problem, an
ofﬁcial rescue loan package could worsen the situation due to the seniority of ofﬁ-
cial loans in conjunction with certain design aspects of the package, which we


Canuto et al.                                                                            117
discuss later. This brings us to why sovereign debt swaps tend not to work when
ﬁscal fundamentals are weak.
   Neither the 1998 Russian swap out of GKOs (ruble T-bills) into Eurobonds nor
Argentina’s 2001 mega-swap, which were designed to lower borrowing costs and
lengthen maturities, worked. Even worse, they backﬁred. 18 The reason is that in a
market-based voluntary debt swap (the case both for Russia 1998 and Argentina
2001), investors work to protect the value of their assets. For debt swaps to work,
they have to reduce the debt burden of countries. Creditors are unlikely to let this
happen in a voluntary fashion—a result that the reader will recognize as a variant
of the Modigliani-Miller Theorem from corporate ﬁnance.19 Indeed, creditors could
demand additional compensation that would worsen the ﬁscal situation. For
example, Argentina’s swap was concluded at a spread of 1,100 basis points,
whereas according to Mussa (2002), calculations showed that at spreads of over
1,000 basis points, Argentina’s debt dynamics were “virtually hopeless”. After the
swap, meltdown proceeded as tax collections continued to ﬂag, bond spreads
rose further, and bank runs intensiﬁed because of concerns about the viability of
the hard peg. Six months later, Argentina defaulted on its debt, including the bonds
restructured as part of the mega-swap.



Procrastination in Sovereign Debt Restructurings
Procrastination is a major impediment to ODRs. We discuss two points: ﬁrst, why
procrastination is costly; second, we attempt to understand why procrastination
occurs.


Why Procrastination Is Costly
With adverse debt dynamics and diminished chances of a positive catalytic effect
of ofﬁcial bailout funds due to ofﬁcial seniority and debtor country solvency
concerns—illustrated most vividly by Russia 1998 and the common lessons from
this crisis and Argentina 2001 that were discussed above—procrastination
becomes costly. This is because the ratio of debt-to-GDP continues rising until a
default or debt restructuring becomes unavoidable. The ofﬁcial bailout funds only
enable short-term creditors to exit at 100 cents on the dollar. The costs to the
debtor nation and creditors can then rise substantially, except for the short-term
creditors. The long-term prospects for the debtor country and for the remaining
private creditors are likely to worsen. To the extent that private creditors hold both
short- and long-term claims, they are liable to lose on the latter whatever they gain
on the former. Greece is the latest illustration of this point and is discussed further
below.

118                                    The World Bank Research Observer, vol. 29, no. 1 (February 2014)
   Returning to the 1980s, an FDIC study (FDIC 1997) noted that the four largest
debtors, Mexico, Brazil, Venezuela, and Argentina, owed $37 billion to the eight
largest US banks, which amounted to 147 percent of the banks’ capital and re-
serves. Suppose the banks had to take a 25 percent haircut on these loans. This
would have amounted to $9.25 billion and wiped out some 37 percent of these
banks’ capital. However, even their total exposure, $37 billion, was less than 1
percent of the US GDP in 1985. Different and less costly strategies were conceivable
that would have beneﬁted millions of poor people in Latin America, who eventually
bore the brunt of its lost decade even as ofﬁcial intervention kept the banks going.
Borrowing access by the debtor countries would most likely have been disrupted.
However, such a hardening of their budget constraint was needed to address the un-
derlying ﬁscal and governance problems, as the country studies by Sachs (1990)
indicate.
   Similarly, in Russia’s case, it became obvious by mid-May 1998 that government
debt was on an unsustainable course and that the ruble was hugely overvalued.
Suppose the ofﬁcial community had persuaded Russia at that point to ﬂoat the
ruble, backed it in seeking a haircut for private creditors, and given it an ofﬁcial
rescue package at the risk-free rate (reﬂecting its seniority and risk status). The sit-
uation would still have been difﬁcult, and the U.S. most likely would still have had to
bail out the systemic hedge fund operated by Long Term Capital Management
(LTCM), which fell victim to contagion from the Russian default (Dungey et al.
2006). However, the problem would have been smaller, as noted by Rogoff. Dollar-
denominated public debt had increased by $16 billion, or 8 percent of the post-
crisis GDP , over the 10 weeks preceding the date of the crisis.20
   In the case of Greece, the October 2011 EU Summit announcements on the need
for haircuts for private creditors were in line with the conclusion in Chamley and
Pinto (2011). They were eventually implemented in March 2012, two years after
negotiations on a bailout began. In the meanwhile, a debt problem amounting to 3-
4 percent of the euro area GDP in Greece had mushroomed by the summer of 2011
into a situation in which “nearly half of the E6.5 trillion stock of government debt
issued by euro area governments . . .” was at risk, with the crisis having spread to
Italy and Spain (IMF 2011, p. 16). As a result, the euro area has become locked
into an interacting vulnerability linking sovereign debt and exposed banks.
   This situation raises a tantalizing question: had the haircut for Greece’s private
creditors been imposed upfront in March 2010 when the solvency problem was ﬁrst
detected and the vulnerable foreign banks ring-fenced, would the wider contagion
have been avoided? A speculative attack on the debt of other vulnerable sovereigns
might well have followed, but this could have also spurred a more decisive response
by the ofﬁcial community, including the major central banks. The back-to-the-wall
effects of crises in concentrating the mind and a two-year head start in implement-
ing ﬁscal, banking, and structural reforms in the vulnerable Eurozone countries

Canuto et al.                                                                       119
should not be dismissed. As it turned out, contagion spread to the core of the
Eurozone. Notwithstanding the palliative effects of ECB’s LTRO interventions (see
Introduction), Italian 10-year bond yields were once again approaching the 7
percent “bailout” threshold by June 2012.
   Two subsequent events have kept a lid on the spreading sovereign debt crisis in
the Eurozone. The ﬁrst was a now-famous announcement at the end of July 2012
by Mario Draghi, President of the ECB, that the ECB would do “whatever it takes” to
support the euro and keep the Eurozone together. The second was an announce-
ment in September 2012 by the ECB about its government bond buying program,
Outright Monetary Transactions (OMT), provided the country concerned agreed to
a corrective program. At the same time, attention turned to creating a fuller ﬁscal
and banking union to save the euro. As of November 2013, the ﬁscal and banking
unions as well as the OMT program remain works in progress. In its April 2013
Fiscal Monitor, the IMF observed that of the ten advanced countries with a debt-to-
GDP ratio over 90 percent and with adverse dynamics, seven are in the Eurozone.
The sovereign debt situation has noticeably worsened in the euro periphery.


Why Is There Procrastination?
Let us return to the 1980s. It took seven years to accept that debt reduction was
needed. It took a few years more to negotiate and implement the Brady debt deals
based on the menu of options available. However, even if the Brady deals had been
negotiated and implemented instantaneously, seven years would still have been lost.
Thus, the ﬁrst impediment to an ODR seems to be an inbuilt bias towards procrasti-
nation. Where does this procrastination come from? One might think that politi-
cians in the debtor country have an incentive to procrastinate rather than to admit
that mismanagement on their watch led to a default. However, Mexico was quick to
admit in 1982 that it could no longer service its external debt. Similarly, Russia
pulled the plug on its international rescue package by devaluing and defaulting on
August 17, 1998, less than a month after it had been approved by the IMF.
   Do private creditors have an incentive to procrastinate? If the prospects for the
country are bad and the debt dynamics adverse, then individual creditors have an
incentive to sell and exit before the others do to minimize their own losses in line
with the prisoners’ dilemma. This would bring matters to a head, but might not
happen for two reasons. First, if all creditors exit simultaneously and panic results,
everyone loses much more, as in a ﬁre sale. This might keep creditors from exiting.
A second pivotal reason is the anticipation of an ofﬁcial bailout package. However,
creditor reaction would also depend on the maturity of the debt held. If it is short
term, there is a clear incentive to hang on if the probability of a large rescue
package is high and exit at 100 cents on the dollar. However, this would be at the
expense of long-term bondholders.

120                                   The World Bank Research Observer, vol. 29, no. 1 (February 2014)
    What about economists? Implicit in the preceding Rogoff quote on Russia 1998
is that economists should know when a currency is overvalued and when there is a
solvency rather than a liquidity problem. In fact, Rogoff was arguing that econo-
mists knew it all in the case of Russia 1998 but were driven to continue with the
(unsustainable) status quo because there was no easy bankruptcy process for sover-
eigns.21 We sympathize with the view that economists should be able to assess
whether a currency is overvalued and whether the public ﬁnances are salvageable
without a debt write down.
    What about the ofﬁcial community, including the IFIs? One would have to admit
that the record is mixed, with Russia 1998 and Argentina 2001 as examples of pro-
crastination and the ﬂawed design of rescue packages.22 The latter, in particular,
means one of two things: either (a) that the economists involved were not sufﬁcient-
ly astute in assessing the sustainability of the ﬁxed exchange rates or of the public
ﬁnances or (b) that debt reduction is anathema and that the ofﬁcial community will
do whatever it takes to bail out private creditors and avoid setting a moral hazard-
inducing precedent for debtor countries. Although one might have believed that (b)
was true, the experience from Russia 1998, Argentina 2001, and Greece 2010 sug-
gests that it must be revisited.



ODRs – What Should They Look Like?
Successful ODRs have been few and far between, typically involving tiny amounts
of debt. One was Ukraine’s debt exchange offer of February 2000 involving
$2.6 billion, which achieved an NPV reduction of 22 to 35 percent (Table 5.4,
Sturzenegger and Zettelmeyer 2006) and elicited a high participation rate.23
Ukraine was then under the IMF’s three-year US$2.2 billion Extended Fund Facility
(signed in September 1998), and the IMF made it clear that Ukraine could not use
its low reserves to service maturing debt and that the IMF program depended upon
a satisfactory debt restructuring.24 This unambiguous signal of “no bailout” per-
suaded private creditors to agree rapidly to a deal. In addition, Pakistan’s debt
restructuring of 1999 involving $610 million (Table 6.3, Sturzenegger and
Zettelmeyer 2006) was in large part due to a comparability requirement imposed by
the Paris Club, which had rescheduled Pakistan’s loans in January of that year.25
The swap offer attracted a participation rate of close to 99 percent, partly because of
default concerns with the original bond, and it achieved a reduction of 30 percent
in NPV terms.
   However, lingering dissatisfaction with the process and outcome of debt restruc-
turings in more complicated cases has prompted a few corrective proposals. Sachs
(1995) proposed an international bankruptcy mechanism to achieve ODRs
that would entail a payment moratorium by the debtor country during debt

Canuto et al.                                                                      121
renegotiations. The Sovereign Debt Restructuring Mechanism (SDRM) was pro-
posed by the IMF in 2001 to facilitate creditor coordination in the event of debt
restructurings for bond debt, the holdings of which are much more dispersed than
the concentrated syndicated bank loans that featured in the debt crisis of
the 1980s.26 In addition, a voluntary code of conduct was proposed by Jean-
Claude Trichet in 2001 that spelled out nine principles governing creditor-debtor
relations during debt restructurings. 27 However, none of these proposals has
gained traction.
   The only mechanism that has been widely accepted by the market has been
Collective Action Clauses (CACs).28 These are part of the terms and conditions gov-
erning a bond issue and can be invoked by the debtor government. The most fre-
quently used CAC is one that entails a modiﬁcation of payment terms requiring a
favorable vote by a majority of the outstanding bond holders (75 percent, typically;
85 percent in some cases, but it could be lower). Empirical evidence on the impact
of CACs on bond pricing has been inconclusive, and their usefulness in achieving
an ODR is questionable.29
   How should the way forward look? Based on Sections 2 and 3, we posit three con-
ditions that an ODR should fulﬁll at a minimum, with which we believe most econo-
mists would be comfortable:
  †   Restore the debtor country’s government debt to a sustainable trajectory;
  †   Minimize procrastination and costs for both the debtor country and its credi-
      tors; and
  †   Minimize any harmful contagion effects in our interconnected world.
   However, the cumulative EM experience augmented with Greece 2010 demon-
strates that the preceding conditions are seriously incomplete when there is ofﬁcial
intervention in an insolvency setting, that is, when the present value of primary sur-
pluses is less than the present value of outstanding debt obligations. In this case,
either primary surpluses will need to be raised (“ﬁscal reform”) or debt will have to
be written down (“haircuts”) to restore solvency.30 Suppose the market does not
believe that primary surpluses can be raised to restore solvency. In this case, a
numerical example based on Chamley and Pinto (2011) in Annex 1 shows that two
additional conditions are needed: ﬁrst, there should be an upfront haircut for
private creditors to help restore debt sustainability; second, ofﬁcial funds should be
lent at the risk-free rate, reﬂecting their seniority. These two conditions will lead to
a less onerous and therefore more credible ﬁscal program to restore solvency
because there is less debt to address and a smaller ofﬁcial loan will be required.
Moreover, an upfront haircut imposed on all private creditors is more equitable
than a situation in which short-term creditors gain at the expense of long bond-
holders. This might induce long bond holders to hang on instead of selling off. It
could also have political economy beneﬁts: with private creditors receiving an

122                                    The World Bank Research Observer, vol. 29, no. 1 (February 2014)
upfront haircut, the less severe ﬁscal austerity program becomes easier to sell to the
public.



Obstacles to an ODR
The most controversial aspect of an ODR discussed above is likely to be the idea of
an upfront haircut for private creditors in the event of a solvency problem for the
debtor country. Three objections could be raised: moral hazard, the difﬁculty in dis-
tinguishing solvency from liquidity problems, and political economy considerations.

Moral Hazard
For governments, moral hazard implies that countries deliberately and irresponsibly
run up debt to precipitate a solvency problem in which private debt will be written
down. Although it is conceivable that countries have behaved in this manner in the
past and could do so again in the future, such behavior is unlikely to be the norm.
Three points are worth noting in the speciﬁc context where the IFIs (international
ﬁnancial institutions, such as the IMF and World Bank) are brought in to orches-
trate a rescue aiming to restore the government to solvency.
    First, consider who is really being bailed out. It cannot be the country because
any ofﬁcial funds received have to be paid back in full, and such debt is difﬁcult to
renegotiate.31 Therefore, engineering a situation in which ofﬁcial loans are obtained
to pay off maturing private debts at 100 cents on the dollar does not “subsidize” the
country’s “bad” behavior, although one cannot rule out unfair redistributions within
the country itself as well-connected people beneﬁt from the external loans that are
then serviced by the taxpayers, as frequently happened in Latin America during the
1970s and 1980s.
    Second, moral hazard implies a proclivity by countries to default strategically,
that is, to default based on unwillingness to pay rather than an inability to do so.
Once again, there is little evidence to support such a position.32 Countries typically
default only as a last resort.33
    Perhaps the most compelling argument against moral hazard by debtor countries
is the unambiguous trend toward self-insurance by EMs, documented in Aizenman
and Pinto (2011, 2013) and Pinto (2014, chapter 9). By deﬁnition, a country
prone to moral hazard will not self-insure because this would be contradictory to
the idea that someone else is insuring the country’s behavior.34 However, EMs
moved aggressively to self-insure at three levels after their 1997 – 2001 crises,
taking steps to (a) restore sustainable public debt dynamics by raising primary sur-
pluses and strengthening ﬁscal institutions; (b) insure against shifts in market senti-
ment and sudden stops by building up foreign exchange reserves and restricting

Canuto et al.                                                                      123
currency mismatches on government and private balance sheets; and (c) lower con-
tingent liabilities from the private sector by shifting to ﬂexible exchange rates, moni-
toring private external borrowing, and strengthening ﬁnancial institutions.
   What about moral hazard for private creditors? First, these creditors price risk
and are diversiﬁed. Nevertheless, as documented in Kharas, Pinto, and Ulatov
(2001, Box 2), private creditors are often in the forefront of the drumbeat for big
bailout packages. What would be better than to price government debt at default
levels and exit at 100 cents on the dollar? In other words, they are not innocent
bystanders.35 Second, the economic beneﬁts of external ﬁnancial integration for
developing countries are seriously questioned (Aizenman and Pinto 2011, 2013),
and one should take threats that haircuts for private creditors will have disastrous
effects for EMs (by shutting off market access) with a grain of salt.36 If anything,
experience shows that disruptions in market access force countries to ﬁnally address
the fundamental ﬁscal problem at the root of sovereign debt crises. The self-insur-
ance by EMs after 1997 – 2001 discussed above is an extreme manifestation of
precisely such behavior. Third, where ofﬁcial funds are used to bailout private credi-
tors, the primary ﬁscal surplus targets needed to assuage default fears on the
remaining private debt might simply be out of reach, as argued earlier.
   Ultimately, the prospect of an upfront haircut for private creditors is a matter of
pragmatism because it increases the chances of a successful and credible ﬁscal
program and implies equal treatment for short- and long-term creditors. Otherwise,
it makes little sense to inject senior ofﬁcial funds into an insolvency situation. The
knowledge that they would be subject to an upfront haircut in the event of insolven-
cy would also make private creditors exercise greater caution ex ante in lending to
sovereigns, reducing an important source of moral hazard, which is in line with the
caveat emptor principle.


Solvency versus Liquidity
The seriousness of insolvency can be gauged from two vantage points. First, what
are the market signals? If bond spreads indicate a high probability of default and
keep rising even as ofﬁcial bailout discussions continue (the case with Russia 1998,
Argentina 2001, and Greece 2010), then this should be considered a red ﬂag.
Second, what are the country’s ﬁscal and growth prospects as conveyed by an
assessment of its fundamentals by economists at the IFIs? How likely is the country
to generate the needed primary surpluses to stabilize and even lower debt? If the
chances are slim, accompanying ofﬁcial loans at the risk-free rate with an equal
upfront haircut for all investors will enable more credible ﬁscal targets and lead to
lower long-bond spreads, minimizing reputation costs.
   However, we admit that it is not always easy to judge whether a country is
dealing with multiple equilibria (liquidity and conﬁdence) or a fundamental

124                                    The World Bank Research Observer, vol. 29, no. 1 (February 2014)
(insolvency) problem. A case in point is Brazil in the summer of 2002, just before
Luiz Inacio Lula da Silva, the candidate of the Brazilian Workers’ Party, was elected
president. Bond spreads reached 2000 basis points that July as presidential election
polls “indicated that Lula would win the presidential election . . ..can [investors] be
certain that a Brazil run by a president with a past record of sympathizing with
default will not take the easy way out?” In the article from which this quote was
taken, Williamson (2002) argued that fundamentals were sound; primary ﬁscal
surpluses had been raised substantially, and budget constraints hardened for the
state governments. Additionally, the real had been ﬂoated in 1999.37
    In his classic on multiple equilibria, Obstfeld (1994) recognizes, “Ultimately,. . ..crises
based on limited foreign reserves [liquidity] must also be based on overall ﬁscal
weakness: [otherwise] . . . it would be . . ...feasible to borrow sufﬁcient reserves to
. . ...fend off any attack [on the ﬁxed exchange rate]” (material in square brackets
added). Brazil was different in that it had taken clear steps to strengthen its ﬁscal
fundamentals. Not only was it running signiﬁcant primary surpluses, it had raised
these substantially compared to the period prior to the ﬂoat of the real in January
1999. Adopting a ﬂoat while moving to address currency mismatches (the govern-
ment was simultaneously shifting toward local currency debt) would substantially
alleviate the international liquidity problem because the central bank would no
longer be in the position of having to defend a ﬁxed peg. Therefore, at the time that
the bond vigilantes went after it in the summer of 2002, Brazil’s problem was one of
political risk and conﬁdence, which was boosted, albeit feebly, by the announce-
ment of a $30 billion loan from the IMF. In the case of Greece, bond spreads contin-
ued rising with each successive bailout augmentation.
    Hence, if a country is taking steps to self-insure along the comprehensive lines
discussed above (including the adoption of ﬂexible exchange rates and hardening
budgets), then one might be a bit more cautious about confusing a liquidity with a
solvency problem. However, this was not the case with Russia 1998, Argentina
2001, or Greece 2010, which were much more clear-cut ex ante on both market
signals and fundamentals.


Political Economy
We interpret “political economy” as anything that would lead to procrastination.
The greatest resistance is likely to come from the creditors themselves. For example,
the large commercial banks holding Greek debt were in the forefront of warning
against any Greek debt restructuring because of the contagion risks. It was in their
interest to allow ofﬁcial creditors to replace exiting private creditors at 100 cents on
the dollar while ﬁscal and structural reforms were implemented.
   However, three caveats are in order. First, creditors are not a homogeneous
bunch. They are distinguished by the maturity of the debt they hold, with

Canuto et al.                                                                              125
short-term creditors beneﬁting most from the strategy of taking the sovereign out of
the market for a few years with ofﬁcial creditors replacing exiting private creditors;
by the size of their exposure; and by whether they are covered by insurance, for
example, in the form of credit default swaps, especially if such insurance was pur-
chased before the solvency problem was detected. Second, replacing exiting private
lenders with ofﬁcial lenders does not lower the country’s debt burden in a present
value sense and results in more ambitious ﬁscal targets to restore solvency, which
are by deﬁnition unattainable; otherwise, we would be dealing with a liquidity and
not a solvency problem. Therefore, there is a risk that the country could abandon
the program for political reasons before short-term creditors have exited. This was
deﬁnitely the case in Russia 1998 and, judging by the extraordinarily high two-
year bond spreads (see Introduction), appeared to be the fear in the case of Greece
2010 as well. Third, creditors may hold both short and long maturity debt. In this
case, provided the ofﬁcial loans are priced in accordance with their risk (and the
interest rate should equal the risk-free rate in the case where ofﬁcial loans are ﬁrst
in the queue and small enough to be paid in full), creditors could lose on their hold-
ings of long-term debt what they gain on their short-term debt holdings. This is
because in an insolvency case, a default and debt write down become inevitable,
with all the burden of the restructuring falling on the remaining private debt.
   Therefore, an upfront haircut is simply a way of distributing restructuring costs
more fairly across creditors in the event of insolvency. It is the analog of the “con-
certed lending” approach applied to the money center banks during the 1980s
crisis to pre-empt the free-rider problem—that is, some banks reducing their expo-
sure as others roll over their loans—once a debt overhang develops.



Conclusions
The record on ofﬁcial intervention in sovereign debt crises is not ﬂattering, whether
it be the 1980s, Russia 1998, Argentina 2001, or Greece 2010. An important
reason is the tendency to procrastinate, treating solvency problems as liquidity
problems even when the distinction between the two is clear. If the goal of ofﬁcial
intervention in such circumstances is to teach debtor countries a lesson, nothing
needs to change. However, if the goal is to increase the likelihood of an orderly debt
restructuring, then pricing ofﬁcial loans at the risk-free rate (in line with their
more-or-less zero risk) and insisting on an upfront haircut—when the bargaining
power of the ofﬁcial sector is the greatest—for private creditors will help.38 It will
also share the burden more equitably between short- and long-term creditors.
Although this may appear to be a recipe for moral hazard, the aggressive self-insur-
ing behavior of emerging market countries after their crises of 1997-2001 suggests
behavior diametrically opposed to what one might expect from countries conﬁdent

126                                   The World Bank Research Observer, vol. 29, no. 1 (February 2014)
of being bailed out should they run up debt irresponsibly. Additionally, private credi-
tors are hardly innocent bystanders; they are sophisticated investors who price risk
and are diversiﬁed. Therefore, an upfront haircut in the event of a solvency problem
should not come as a total surprise to them and could make their ex ante lending
behavior more diligent.
   Ultimately, there is a stark choice between two strategies: gambling for redemp-
tion as in Conesa and Kehoe (2011), in which an immediate haircut is avoided, and
insisting on an upfront haircut for private creditors while keeping in mind that the
cost of default is large and that the chances of a default increase with procrastina-
tion, as in Chamley and Pinto (2013). The approach to sovereign debt restructuring
favored by ofﬁcial agencies has been to gamble for redemption, reﬂected in the use
of ofﬁcial funds to take the country out of the market while implementing ﬁscal and
structural reform to raise primary ﬁscal surpluses and spur growth. It has tended
not to work—Latin America in the 1980s, Russia 1998, Argentina 2001, and
Greece 2010 are examples.
   However, the ofﬁcial mindset may be changing in favor of inﬂicting upfront hair-
cuts on private creditors when a country is obviously insolvent. The ofﬁcial
approach to the Cypriot banking crisis, which came to a head in March 2013, was
to insist on losses for bank depositors because the size of the banking system (800
percent of GDP) precluded a government bailout. Indeed, in a May 2013 draft law
on bank resolution, the EU embraced the idea that in future banking crises, burden
sharing might be needed with shareholders, bondholders, and uninsured deposi-
tors. On June 5, 2013, the IMF published its post-mortem of the Greek bailout,
noting that (IMF 2013, 28; words in square brackets added) “not tackling the
public debt problem decisively at the outset or early in the program created uncer-
tainty about the euro area’s capacity to resolve the crisis and likely aggravated the
contraction in output. An upfront debt restructuring would have been better for
Greece . . .. A delayed debt restructuring also provided a window for private credi-
tors to reduce exposures and shift debt into ofﬁcial hands . . ..[which] occurred on a
signiﬁcant scale and limited the bail-in of creditors when PSI eventually took place,
leaving taxpayers and the ofﬁcial sector on the hook.”39
   In conclusion, we have laid out the desirable attributes of an orderly sovereign
debt restructuring (ODR) when a solvency problem is involved. In addition to ensur-
ing that government debt attains a sustainable trajectory and that procrastination
and contagion costs are minimized, pricing ofﬁcial funds in line with their risk and
using ofﬁcial bargaining power to insist on an upfront haircut for private creditors
would be desirable. However, there is no perfect formula for distinguishing between
liquidity and solvency problems and upfront haircuts are going to encounter stiff
political resistance.40 Therefore, ODRs are likely to remain elusive, recent signs of a
change in the ofﬁcial mindset notwithstanding.



Canuto et al.                                                                      127
Notes
* The authors are Senior Adviser, Development Economics Department of the World Bank; Chief
Economist, Emerging Markets, GLG Partners LP; and Senior Economist, Poverty Reduction and
Economic Management Department in the Europe and Central Asia region of the World Bank, respec-
tively. They acknowledge valuable comments from three anonymous referees. The views herein are
entirely those of the authors. They do not necessarily represent the views of the International Bank for
Reconstruction and Development/World Bank and its afﬁliated organizations or those of the Executive
Directors of the World Bank or the governments they represent.
    1. See Chamley and Pinto (2011).
    2. The ECB lent 523 banks E498 billion in December, with another E530 billion to 800 banks at
the end of February 2012.
    3. The New York Federal District Court ruling in November 2012 on the Argentine default is a
case in point. The ruling required Argentina to honor the pari passu clause, which means equal treat-
ment of all bondholders. Hence, it required the Argentine government to make payments to holdout
creditors if the restructured bonds of 2005 and 2010 were honored. The U.S. Federal Appeals Court
upheld this decision in August 2013. However, as noted by Pinto (2014) and in contrast to the discus-
sion on self-insurance later in this paper, Argentina is an exception to the rule that EM government
behavior changed for the better following the crises of 1997–2001.
    4. A detailed account of speciﬁc country episodes involving sovereign debt restructuring is con-
tained in the work of Sturzenegger and Zettelmeyer (2006). Pinto and Tanaka (2005) describe the
various instruments and options availed of during these restructuring episodes, starting with the
1980s debt crisis.
    5. Visit http://www.clubdeparis.org/.
    6. Details       at     http://web.worldbank.org/WBSITE/EXTERNAL/TOPICS/EXTDEBTDEPT/0,,
contentMDK:20260411~menuPK:64166739~pagePK:64166689~piPK:64166646~theSitePK:
469043,00.html. Also see World Bank (2011).
    7. This statement applies to the period under review. This does not rule out macroeconomic crises
with ﬂexible exchange rates. For a theoretical example of the latter, see Kumhof, Li, and Yan (2007).
    8. Drawn from Sachs (1990), an overview of a volume of country studies on the 1980s debt crisis.
    9. An exacerbating factor was that although they were of long maturity, these external hard cur-
rency debts had ﬂoating rates, with the interest rate adjusted every six months based on a market
index. Therefore, once the U.S. started raising interest rates, the interest burden of the developing
countries quickly increased.
    10. See Pinto (1991). For an application to Bolivia, see Kharas and Pinto (1989) and Morales and
Sachs (1990).
    11. See, for example, the case study on Argentina by Dornbusch and de Pablo (1990). Brazil had
several stabilization programs during the 1980s and until 1994. In July 1994, after six failed price
stabilization plans over the previous ten years, Brazil ﬁnally initiated a successful stabilization effort
embedded in the Real plan. It lowered consumer inﬂation from 2287 percent in 1994 to 71.9 percent
in 1995, 18.2 percent in 1996, and ﬁnally to 7.7 percent in 1997. See Blanco et al. (2011).
    12. For Russia 1998, see Kharas, Pinto, and Ulatov (2001), Pinto, Gurvich, and Ulatov (2005)
and Pinto and Ulatov (2012). For Argentina 2000–1, see Serven and Perry (2005) and De la Torre,
Levy Yeyati, and Schmukler (2003).
    13. Pinto and Tanaka (2005).
    14. Chuhan and Sturzenegger (2005).
    15. Claessens, Oks and van Wijnbergen (1993) make a similar argument.
    16. The arguments that follow were made prior to the crisis in real time by the economics unit of
the World Bank ofﬁce in Moscow, which the second author of this paper then headed.
    17. IMF (2004, lessons 7–9).




128                                            The World Bank Research Observer, vol. 29, no. 1 (February 2014)
    18. For a description of the swaps and a formal argument for why they backﬁred, see Aizenman,
Kletzer, and Pinto (2005). In a similar vein, the Mexican Government began rolling over its short-term
peso-denominated debt (Cetes) into short-term dollar indexed debt (Tesobonos) after March 1994 to
avoid raising interest rates to deal with rising devaluation risk. This became a major source of
ﬁnancial vulnerability. See Sachs et al. (1996).
    19. In the sovereign case, the market value of the debt is determined by the present value of the
future primary ﬁscal surpluses. As long as this is ﬁxed (when discounted at the risk-free rate), shufﬂing
the mix of debt instruments through market-based swaps will not change the present value of the debt
burden.
    20. Details may be found in Kharas, Pinto, and Ulatov (2001) and Pinto and Ulatov (2012).
    21. Although Rogoff ’s is an excellent point, the absence of an easy bankruptcy procedure was not
the only reason the doomed Russian rescue package of July 1998 proceeded. There was a strong belief
in inﬂuential quarters that with Russia having achieved single-digit inﬂation in February 1998, these
hard-won stabilization gains had to be preserved, and that with Russian government debt much less
than the Maastricht criterion of 60 percent of GDP  , the market was overreacting.
    22. On Russia 1998, see Kharas, Pinto, and Ulatov (2001). For Argentina 2001, see Mussa
(2002) and IMF (2004).
    23. In contrast, the debt amount involved in Russia 1998 was $77 billion ($45 billion in ruble
debt, $32 billion owed to the London Club). Pinto, Gurvich, and Ulatov (2005, 431).
    24. See commentaries in “Ukraine: Heading to Default and Restructuring”, Commerzbank Global
Fixed Income, 7 February 2000, and “Ukraine: Further Upside for Eurobonds”, Emerging Markets
Economics Research, Credit Suisse First Boston, February 3, 2000.
    25. See Pinto and Tanaka (2005).
    26. See Krueger (2002).
    27. See Trichet (2001).
    28. See Bradley and Gulati (2011), Eichengreen and Mody (2000), and Weinschelbaum and
Wynne (2005).
    29. For example, it was reported in the news on March 9, 2012 that Greece was able to secure a
95.7 percent participation rate among private-sector creditors in its bond exchange by invoking CACs
to make the deal binding on holders of Greek-law bonds (until the bailout began, much of Greek debt
was under Greek law). However, by then, severe damage had been done to the Greek economy and the
wider euro area.
    30. For technical details, see Burnside (2005).
    31. See, for example, de Bolle, Rother, and Hakobyan (2006).
    32. Sturzenegger and Zettelmeyer (2006, 4, 38) argue that most sovereign defaults since the
1970s were driven by interactions between domestic policies and economic shocks (including exogenous
shocks), sometimes worsened by political shocks. In this sense, ability and willingness to pay are difﬁcult
to disentangle. However, Ecuador’s default of 2008 on its US$3.2 billion Eurobonds was a rare instance
of a country that did not repay its debt even though it had the resources to do so. The Eurobonds were
declared “illegitimate”, and the government bought back 91 percent of the defaulted debt in the second-
ary market at 35 percent of face value.
    33. Tomz and Wright (2007) ﬁnd a negative but weak relationship between economic output and
default on external loans from private creditors. Eden, Kraay, and Qian (2012) also come to a similar
conclusion that defaults are more likely to occur during growth slowdowns in countries with weak
policy performance and that have seen rapid debt accumulation.
    34. This is further corroborated by Aguiar and Amador (2011), who build on the debt overhang
argument and ﬁnd that countries that grow rapidly are those that accumulate net foreign assets
because growth in capital requires a reduction in the stock of debt.
    35. See Canuto, Pinto, and Prasad (2012, 22 –3) for some pointed examples.
    36. For example, Eichengreen and Ruhl (2000) argued, in the context of Ecuador, Pakistan,
Romania, and Ukraine following the East Asian and Russian crises, that IFIs acted to avoid “a costly,



Canuto et al.                                                                                          129
extended interruption to market access” and were therefore not credible when they sought to impose
haircuts on private creditors.
   37. Williamson (2002) was partly responding to an estimate by Morris Goldstein that there was a
70 percent chance that Brazil would be forced to restructure its debt by the end of 2003. See Goldstein
(2003) and the excellent overview in Giavazzi, Goldfajn, and Herrera (2005).
   38. With adverse debt dynamics and the growing risk of contagion, the ofﬁcial sector may ﬁnd its
bargaining power eroding as time passes.
   39. These conclusions were anticipated in Chamley and Pinto (2011) and Canuto, Pinto, and
Prasad (2012), the working paper version of this publication.
   40. Even the ofﬁcial sector may ﬁnd it difﬁcult to assume a fully objective stance. For example, the
interests of the US in the 1980s debt crisis and those of the ECB and EU in the ECB-EU-IMF troika in
the Eurozone crisis may not always coincide with what a dispassionate body like the IMF is likely to
recommend.




Annex 1: Numerical Example on Ofﬁcial Intervention
in Insolvency Situations
Consider the 2-period situation in Annex Table 1. The debt service due in each
period is shown in the second row of the table. The risk-free sovereign yield in a
benchmark country such as the US or Germany rate is assumed to be 5 percent. In
Scenario 1, the government faces a liquidity problem because the debt service pay-
ments falling due of 100 in period 0 exceed the primary surplus of 75. However, it is
solvent in the sense that the present value of primary ﬁscal surpluses equals that of
the debt to be repaid; both equal 250 when discounted at the risk-free rate of 5
percent. In this case, the government can borrow 25 at the risk-free rate of 5
percent either from the markets or the IMF to make up the difference. The total
amount it must repay in period 1 is 25X1.05 þ 157.50 ¼ 183.75, which can be
exactly met out of the primary surplus in period 1.
   Now suppose an adverse shock occurs and the period 1 primary surplus falls to
175, as in Scenario 2. The government now has a solvency problem in the sense
that the present value of the primary surpluses at the risk-free discount rate of 5
percent falls to 241.67. Equilibrium can be restored to the government’s intertem-
poral budget constraint if the price of the debt were to fall from 1 to 0.967 (the ratio
of the present value of primary surpluses to that of debt service due at the risk-free


Annex Table 1. Two Hypothetical Fiscal Scenarios
                                                                          Period 0                   Period 1

                                      Debt service due                     100                       157.50
Scenario 1 (liquidity)                Primary surplus                       75                       183.75
Scenario 2 (solvency)                 Primary surplus                       75                       175




130                                           The World Bank Research Observer, vol. 29, no. 1 (February 2014)
discount rate ¼ 241.67/250). Even with this haircut, the government would still
need to borrow an amount given by 96.7-75 ¼ 21.7 in period 0 to pay off the
maturing debt. In other words, it has both a liquidity and solvency problem.
    Suppose it were to go to the market to borrow this amount of 21.7. What would
it be charged? Anticipating the haircut, the market would charge an interest rate i
given by the arbitrage condition: 0.967(1 þ i ) ¼ 1.05. This can be solved to give
i ¼ 8.58 percent.1 The spread jumps from 0 to 358 basis points. The amount of
21.7 can also be sought from ofﬁcial sources, such as the IMF. In this case, the
amount due in period 2 is (21.7x1.05 þ 0.967X157.5) ¼ 175, where the IMF’s
seniority means it gets repaid in full; but in line with the Modigliani-Miller theorem,
this effect is offset by its charging the risk-free rate. However, the amount due to
period 1 private creditors is still subject to the same haircut imposed on period 0
creditors.
    This brings us to the situation reﬂective of Greece 2010, as well as both Russia
1998 and Argentina 2001. Anxious to avoid a ﬁrst period default, the government
goes to ofﬁcial creditors and borrows 25 at the risk-free rate to pay off period 1 credi-
tors in full. The big difference is that the size of the ofﬁcial loan goes up from 21.7 to
25. In this case, the price of second-period debt falls from 0.967 to (175-25X1.05)/
157.5 ¼ 0.944. This is equivalent to an interest rate of 11.2 percent, or a spread of
620 basis points. This is exactly what we have witnessed in practice, with long
bond spreads rising substantially and persisting at elevated levels following the
announcement and implementation of the bailout for Greece.
    A crucial difference in the two responses to the insolvency situation is that in one
case, an upfront haircut is imposed on all creditors, leading to a smaller ofﬁcial
loan. In the second case, the one witnessed in practice, short-term creditors get paid
in full, leaving less for long-term creditors. The point of bringing in ofﬁcial creditors
is to engender positive catalytic effects, namely, persuading private creditors to roll
over their loans instead of exiting; and putting pressure on the debtor country for
ﬁscal reform and austerity to increase primary surpluses. In practice, short-term
creditors have been exiting and bond spreads have continued to rise; in Greece’s
case, the share of ofﬁcial loans (ofﬁcial creditors plus the European Central Bank,
ECB, and Eurosystem) had risen to 58 percent by the end of April 2012. In effect,
the ofﬁcial bailouts have taken the countries out of the debt market in the hope that
in the meanwhile primary surpluses will increase to levels consistent with solvency.
In the case of Greece, at least, the market does not seem to have ever believed this
would happen, as conveyed by the evolution of the long bond spread.
    Returning to our numerical example, suppose the country is better off the closer
the price of period 2 debt is to 1, which would imply a lower spread and smaller rep-
utation costs. Raising the price all the way back to 1 would require restoring the
primary surplus in period 2 to 183.75; then we would be back to a liquidity
problem, with the solvency problem solved. This raises the question of how serious

Canuto et al.                                                                         131
the solvency problem was in the ﬁrst place, a point we return to later. In the mean-
while, suppose the interest rate on the ofﬁcial bailout package were higher than the
risk-free rate in spite of the seniority of the ofﬁcial loans.2 Then, by continuity, the
second period primary surplus would have to be even higher than 183.75 in order
for the price of period 2 debt to return to 1, which is likely to severely strain creduli-
ty and derail any catalytic effects of ofﬁcial ﬁnance.3 This points to the importance
of pricing ofﬁcial funds at the risk-free rate in view of their seniority as otherwise
the credibility of the accompanying ﬁscal package would be lowered: it would
require primary surplus targets that would be too onerous to be believable.
   The numerical example shows that when the market believes that there is a sol-
vency problem, an upfront haircut for all creditors combined with ofﬁcial funds
priced at the risk-free rate will lead to more believable ﬁscal targets to restore the
country’s reputation.4 The upfront haircut will require that the ofﬁcial loan be just
21.7 instead of 25; and the period 1 primary surplus target will need to be 180.3
instead of the pre-shock 183.75 to return bond prices to 1.5



Notes
    1. It can be cross-checked that 0.967[21.7X1.0858 þ 157.5] ¼ 175, where the expression in
square brackets equals the new amount payable in period 1.
    2. For example, the IMF loan interest rate for Greece, while well below what the market may
charge, involves signiﬁcant spreads above the IMF’s own borrowing cost (a spread of 200 basis points
for amounts in excess of 300 percent of quota, which goes up to 300 basis points after 3 years if the
credit is still above 300 percent. Greece’s loan was 3,200 percent of quota).
    3. The result here on ofﬁcial ﬁnance is diametrically opposed to that in Morris and Shin (2006),
who treat ofﬁcial and private loans as strategic complements. Here, they become imperfect substitutes
because of insolvency and ofﬁcial seniority.
    4. Notice that if the ofﬁcial loans were also subject to default, they would be priced above the risk-
free rate in anticipation of the haircut; the eventual expected payout would be the same. Therefore, so
long as a haircut on ofﬁcial loans is ruled out, these loans should be priced at the risk-free rate.
    5. 21.7X1.05 þ 157.5 ¼ 180.3.




References
Aguiar, M., and M. Amador. 2011. “Growth in the Shadow of Expropriation.” The Quarterly Journal of
  Economics 126 (2): 651 –97.
Aizenman, J., K. M. Kletzer, and B. Pinto. 2005. “Sargent-Wallace Meets Krugman-Flood-Garber,
   or: Why Sovereign Debt Swaps Don’t Avert Macroeconomic Crises.” Economic Journal 115 (3):
   343 –67.
Aizenman, J., and B. Pinto. 2011. “Managing Financial Integration and Capital Mobility.” Vox:
   http://www.voxeu.org/index.php?q=node/7058.
      . 2013. “Managing Financial Integration and Capital Mobility—Policy Lessons from the Past
   Two Decades.” Review of International Economics 21(4): 636 –53.


132                                            The World Bank Research Observer, vol. 29, no. 1 (February 2014)
Blanco, F., F. de Holanda, B. Filho, and S. Pessoa. 2011. “Brazil: Resilience in the Face of the Global
   Crisis.” In The Great Recession and Developing Countries, ed. M. K. Nabli, 91 –160. Washington, DC:
   World Bank.
Bradley, M., and M. Gulati. 2011. “Collective Action Clauses for the Eurozone: An Empirical Analysis.”
   Duke Law Faculty Scholarship, Paper 2455.
Burnside, C. 2005. Fiscal Sustainability in Theory and Practice: A Handbook. Washington, DC: World
  Bank.
Canuto, O., B. Pinto, and M. Prasad. 2012. “Orderly Sovereign debt Restructuring: Missing in Action!”
  World Bank Policy Research Working Paper WPS 6054.
Chamley, C. P., and B. Pinto. 2011. “Why Ofﬁcial Bailouts Tend Not To Work: An Example Motivated
  by Greece 2010,” The Economists’ Voice 8 (1): 1 –5.
Chamley, C., and B. Pinto. 2013. “Sovereign Bailouts and Senior Loans.” In NBER International
  Seminar on Macroeconomics 2012, ed. Giavazzi, F., and K. D. West, 269–91. Chicago: University of
  Chicago Press.
Chuhan, P ., and F. Sturzenegger. 2005. “Default Episodes in the 1980s and 1990s: What have we
  learned?” In Managing Economic Volatility and Crises: A Practitioner’s Guide, ed. J. Aizenman, and B.
  Pinto, 471–520. Cambridge University Press.
Claessens, S., D. Oks, and S. van Wijnbergen. 1993. “Interest Rates, Growth and External Debt – The
   Macroeconomic Impact of Mexico’s Brady Deal.” World Bank Policy Research Working Paper No.
   1147.
Cline, W. 1990. “From Baker to Brady: Managing International Debt.” In Finance and the International
   Economy 3: The AMEX Bank Review Prize Essays ed. R. O’Brien, and I Iversen. 84-101. Oxford:
   Oxford University Press.
Conesa, J. C., and T. J. Kehoe. 2011. “Gambling for Redemption and Self-Fulﬁlling Debt Crises.”
  Mimeograph.
de Bolle, M., B. Rother, and I. Hakobyan. 2006. “The Level and Composition of Public Sector Debt in
   Emerging Market Crises.” IMF Working Paper 186.
de la Torre, A., E. L. Yeyati, and S. Schmukler. 2003. “Living and Dying with Hard Pegs: The Rise and
   Fall of Argentina’s Currency Board.” Working Paper, Centro de Investigacion en Finanzas,
   Universidad Torcuata di Tella.
Development Committee. 2010. “How Resilient Have Developing Countries Been During the Global
   Crisis?” DC2010-0015, September 30 2010. http://siteresources.worldbank.org/DEVCOMMINT/
   Documentation/22723862/DC2010-0015(E)IDAResilience.pdf
Dooley, P. M. 1994. “A Retrospective on the Debt Crisis.” NBER Working Paper, No. 4963.
Dornbusch, R., and J. C. de Pablo. 1990. “Debt and Macroeconomic Instability in Argentina.”
  In Developing Country Debt and Economic Performance, Volume 2: The Country Studies – Argentina,
  Bolivia, Brazil, Mexico, ed. J. Sachs, 39 –156. Chicago: The University of Chicago Press.
Dungey, M., R. Fry, B. Gonzalez-Hermosillo, and V. Martin. 2006. “Contagion in international bond
  markets during the Russian and the LTCM crises”, Journal of Financial Stability 2 (1): 1– 27.
Eden, M., A. Kraay, and R. Qian. 2012. “Sovereign Defaults and Expropriations: Empirical
   Regularities.” World Bank Policy Research Working Paper No. 6218.
Eichengreen, B., and A. Mody. 2000. “Would Collective Action Clauses Raise Borrowing Costs? An
   Update and Additional Results.” World Bank Policy Research Working Paper No. 2363.
Eichengreen, B., and R. Portes. 1989. “Dealing with debt: The 1930s and the 1980s.” In Dealing with
   the debt crisis, ed. I. Husain, and I. Diwan, 69 –85. Washington, DC: World Bank.



Canuto et al.                                                                                      133
Eichengreen, B., and C. Ruhl. 2000. “The Bail-In Problem: Systematic Goals, Ad Hoc Means” NBER
   Working Paper 7653.
Federal Deposit Insurance Corporation. 1997. “The LDC Debt Crisis.” In History of the Eighties—
   Lessons for the Future. http://www.fdic.gov/bank/historical/history/191_210.pdf.
Giavazzi, F., I. Goldfajn, and S. Herrera. 2005. Inﬂation targeting, Debt, and the Brazilian Experience,
   1999 to 2003. Cambridge: MIT Press.
Goldstein, M. 2003. “Debt Sustainability, Brazil, and the IMF.” Institute for International Economics
   Working Paper 03-1.
Heavily Indebted Poor Countries (HIPC) Website. http://web.worldbank.org/WBSITE/EXTERNAL/
  TOPICS/EXTDEBTDEPT/0,,contentMDK:20260411~menuPK:64166739~pagePK
  :64166689~piPK:64166646~theSitePK:469043,00.html
IMF. 2004. Report on the Evaluation of the Role of the IMF in Argentina, 1991–2001. Independent
   Evaluation Ofﬁce. Washington, DC: International Monetary Fund.
       . 2011. Global Financial Stability Report. Grappling with Crisis Legacies. September. Washington,
   DC. http://www.imf.org/external/pubs/ft/gfsr/2011/02/index.htm
      . 2013. Greece: Ex Post Evaluation of Exceptional Access under the 2010 Stand-By Arrangement.
   IMF Country Report No. 13/156.
Kharas, H., and B. Pinto.1989. “Exchange Rate Rules, Black Market Premia and Fiscal Deﬁcits: The
  Bolivian Hyperinﬂation.” Review of Economic Studies 56 (3): 435 –47.
Kharas, H., B. Pinto, and S. Ulatov. 2001. “An Analysis of Russia’s 1998 Meltdown: Fundamentals
  and Market Signals.” Brookings Papers on Economic Activity 1:1 –50.
Krueger, A. 2002. “New Approaches to Sovereign Debt Restructuring: An Update on Our Thinking.”
  Paper prepared for the conference “Sovereign Debt Workouts: Hopes and Hazards,” Institute for
  International Economics, Washington, April 1.
Krugman, P. 1988. “Financing vs. Forgiving a Debt Overhang.” Journal of Development Economics 29
  (3): 253– 68.
      . 1994. “LDC Debt Policy.” In American Economic Policy in the 1980s, ed. P. M. Feldstein,
   691 –740. Chicago: University of Chicago Press.
Kumhof, M., S. Li, and I. Yan. 2007. “Balance of Payments Crises under Inﬂation Targeting.” IMF WP/
  07/84. http://www.imf.org/external/pubs/ft/wp/2007/wp0784.pdf
Morales, J. A., and J. D. Sachs. 1990. “Bolivia’s Economic Crisis.” In Developing Country Debt and
  Economic Performance, Volume 2: The Country Studies – Argentina, Bolivia, Brazil, Mexico, ed. J. Sachs,
  157 –268. Chicago: University of Chicago Press.
Morris, S., and H. S. Shin. 2006. “Catalytic Finance: When does it work?” Journal of International
  Economics 70 (1): 161–77.
Mussa, M. 2002. Argentina and the Fund: From Triumph to Tragedy. Washington, DC: Institute for
  International Economics.
Obstfeld, M. 1994. “The Logic of Currency Crises.” Banque de France – Cahiers economiques et monetaires
  43:189 –213.
Paris Club. www.clubdeparis.org
Pinto, B.. 1991. “Black Markets for Foreign Exchange, Real Exchange Rates and Inﬂation.” Journal of
   International Economics 30 (1 –2): 121–35.
Pinto, B., and S. Tanaka. 2005. “Sovereign Debt Swaps with Private Creditors.” PRMED Note, PREM
   Anchor. World Bank, Washington, DC.



134                                            The World Bank Research Observer, vol. 29, no. 1 (February 2014)
Pinto, B., E. Gurvich, and S. Ulatov. 2005. “Lessons from the Russian Crisis of 1998 and Recovery.”
   In Managing Economic Volatility and Crises: A Practitioner’s Guide, ed. J. Aizenman, and B. Pinto,
   406–38. Cambridge: Cambridge University Press.
Pinto, B., and S. Ulatov. 2012. “Financial Globalization and the Russian Crisis of 1998.” In The
   Evidence and Impact of Financial Globalization, GLFI3, ed. G. Caprio, 689 –708. UK: Academic
   Press.
Pinto, B. 2014. How Does My Country Grow? Economic Advice Through Story-Telling. Oxford University
   Press, Forthcoming.
Rogoff, K. 2003. “The IMF Strikes Back.” Foreign Policy 134 (January/February). [http://www.imf.org/
   external/np/vc/2003/021003.htm]
Sachs, J. 1986. “Managing the LDC Debt Crisis.” Brookings Papers on Economic Activity, 1986:2.
       . 1990. “Introduction”. In Developing Country Debt and Economic Performance, Volume 2: The
   Country Studies – Argentina, Bolivia, Brazil, Mexico, ed. J. Sachs, 1–28. Chicago: University of
   Chicago Press.
       . 1995. “Do We Need an International Lender of Last Resort?” Frank D. Graham Lecture,
   Princeton University.
Sachs, J., A. Tornell, and A. Velasco. 1996. “The Collapse of the Mexican Peso: What Have We
   Learned.” Economic Policy 22 (April): 15 –63.
Serven, L., and G. Perry. 2005. “   Argentina’s Macroeconomic Collapse: Causes and Lessons.”
   In Managing Economic Volatility and Crises: A Practitioner’s Guide, ed. J. Aizenman, and B. Pinto,
   439–70. Cambridge: Cambridge University Press.
Sturzenegger, F., and J. Zettelmeyer. 2006. Debt Defaults and Lessons from a Decade of Crises.
Tomz, M., and M. L. J. Wright. 2007. “Do Countries Default in “Bad Times”?” Journal of the European
  Economic Association 5 (2-3): 352 –60.
Trichet, J.-C. 2001. “Preserving Financial Stability in an Increasingly Globalized World.” Keynote
   speech at the European Financial Markets Convention, Paris.
Weinschelbaum, F., and J. Wynne. 2005. “Renegotiation, Collective Action Clauses and Sovereign Debt
  Markets.” Journal of International Economics 67:47 –62.
Williamson, J. 2002. “Is Brazil Next?” International Economics Policy Brief 02-7. Institute of
   International Economics, Washington, DC.
World Bank. 1998. Global Development Finance: Analysis and Summary Tables. Washington, DC: World
  Bank.
World Bank and International Monetary Fund. 2011. Report on “Heavily Indebted Poor Countries
  (HIPC) Initiative and Multilateral Debt Relief Initiative (MDRI)–Status of Implementation and
  Proposals for the Future of the HIPC Initiative.” Washington, D.C.: World Bank.




Canuto et al.                                                                                    135