The Level and Composition of Public Sector Debt in Emerging Market Crises

WP/06/186
The Level and Composition of

Public Sector Debt in
Emerging Market Crises
Monica de Bolle, Björn Rother, and Ivetta Hakobyan
© 2006 International Monetary Fund WP/06/186
IMF Working Paper
Policy Development and Review Department
The Level and Composition of Public Sector Debt in Emerging Market Crises
Prepared by Monica de Bolle, Björn Rother, and Ivetta Hakobyan1
Authorized for distribution by Mauro Mecagni
August 2006
Abstract
This Working Paper should not be reported as representing the views of the IMF.
The views expressed in this Working Paper are those of the author(s) and do not necessarily represent
those of the IMF or IMF policy. Working Papers describe research in progress by the author(s) and are
published to elicit comments and to further debate.
The paper examines the evolution of public sector debt levels and structures in 12 emerging
market countries around the time of financial crises. In particular, it focuses on whether the
debt situation of sovereign borrowers became more vulnerable in the aftermath of crises. The
principal findings are that (i) debt levels tend to increase significantly post-crisis, and
(ii) countries often experience more rigid debt structures following such events, with an
increase in the share of external public debt to multilateral creditors and a greater exposure of
the domestic banking system to sovereign debt.
JEL Classification Numbers: F30 and F34
Keywords: Financial crises, debt distress, sovereign debt
Author(s) E-Mail Address: [email protected]
1
The authors would like to thank Mark Allen, Russell Kincaid, Matthew Fisher,
Mauro Mecagni, and Krishna Srinivasan for their extensive comments and suggestions. The
authors are also grateful for the comments received from seminar participants in PDR and from
other IMF colleagues. Monica de Bolle was an economist in the IMF’s Crisis Resolution Issues
Division when the work on this paper was first undertaken. The authors assume responsibility
for any remaining errors.
-2-
Contents Page
I. Introduction ............................................................................................................................3
II. Evolution of Sovereign Debt Levels .....................................................................................4

A. Evolution of Debt Levels in the Immediate Post-Crisis Period .......................................4
B. Evidence on Debt Reversals.............................................................................................7
C. External Debt versus Domestic Debt ...............................................................................8
III. Changes in the Structure of Sovereign Debt......................................................................10

A. External Public Debt .....................................................................................................10
B. Domestic Public Debt....................................................................................................17
IV. Conclusions........................................................................................................................22
Boxes
1. Did the Relative Importance of Multilateral Lending Increase Over Time? .......................14
2. Evolution of Foreign Direct Investment (FDI) Inflows in Crisis Episodes .........................16
Tables
1. Decomposition of Changes in Debt-to-GDP Ratios, from t–1 to t+1....................................6
2. Domestic Banking Sector Claims on Public Sector.............................................................18
A1. Public Sector Debt by Country: Evolution from t–2 to t+2 ..............................................28
Figures
1. Public Sector Debt by Country as a Share of GDP................................................................5
2. Public Sector Debt by Country: Pre-Crisis Levels and Evolution Through t+3....................8
3. Public Sector Debt: Average Evolution of External and Domestic Debt Components.........9
4. Debt to Multilateral Institutions...........................................................................................11
5. External Sovereign Debt: Creditor Composition, Cross-Country Averages .......................13
6. Domestic Public Sector Debt by Country: Maturity and Currency Composition ...............21
A1. External Sovereign Debt by Country: Creditor Composition...........................................26
Annexes
Data Issues and Methodology..................................................................................................25
References................................................................................................................................32
-3-
I. INTRODUCTION
The rapid international integration of capital markets provides emerging market countries
(EMCs) with considerable opportunities for investment-led growth. At the same time, it exposes
them to sudden shifts in market sentiment and the risk of financial crises. The resolution of these
crises has typically involved the provision of sizable financial assistance by the official
community in support of economic adjustment programs aimed at easing liquidity pressures and
restoring confidence. It is conceivable that under some circumstances, however, countries may
emerge from a financial crisis with increased debt-related vulnerabilities that may leave them
exposed to the risk of recurrent distress.
Conceptually, debt-related vulnerabilities can be expected to increase during and following the
resolution of a crisis through a number of channels. In particular, the level of debt relative to a
country’s repayment capacity may increase because of a sharp depreciation of the exchange rate,
an increase in real interest rates, and a contraction in output. In addition, the debt burden may
increase as a result of new borrowing from external and domestic sources in the context of
efforts aimed at resolving the crisis. Post-crisis vulnerabilities can also stem from crisis-induced
changes in the composition of sovereign debt, including in the maturity and currency
composition of the debt stock and changes in the creditor base which increase the debt’s rigidity,
an issue that has received relatively little attention thus far in the literature.
Against this background, the paper examines the stylized facts regarding changes in the level and
structure of sovereign debt for 12 emerging market economies that experienced a major financial
crisis in recent years.2 The paper focuses on the following questions:
• How did public debt levels evolve over the crisis cycle, and what was the contribution of the
domestic and external components to overall debt dynamics?3
• Did efforts aimed at crisis resolution result in a more rigid external debt structure, defined as a
higher share of senior multilateral claims in total public debt that are less amenable to (or
need to be excluded from) a restructuring?4 In which situations could such a shift in debt
structure contribute significantly to post-crisis vulnerability?
2
The sample includes Argentina, Brazil, Ecuador, Indonesia, Korea, Mexico, Philippines, Russia, Thailand, Turkey,
Ukraine, and Uruguay. For a discussion of data-related issues, see Annex.
3
External and domestic debt refers to the claims of nonresidents and residents, respectively, on the public sector.
This definition follows the World Bank’s Global Development Finance, which is used as one of the main data
sources for this study.
4
The de facto seniority of multilateral claims—or the preferred creditor status of multilateral financial institutions
relative to other creditors—helps insulate these institutions from the risk of nonpayment and debt restructuring. This
enables them to take on risks in the general public’s interest in crisis situations when other creditors are unwilling to
provide financing.
-4-
• Is there evidence to suggest that domestically issued public debt also became more rigid in the
aftermath of crises, owing in particular to an increase in the exposure of the financial sector to
such debt?
• How did the maturity and currency composition of domestic public debt change throughout
the crisis cycle?
The paper is organized as follows. Section II examines the evolution of public debt levels for the
12 countries over the crisis cycle. Section III discusses changes in the structure of sovereign debt
during a crisis and its aftermath, focusing both on its external and domestic components.
Section IV concludes.
II. EVOLUTION OF SOVEREIGN DEBT LEVELS
The experience with financial crises indicates that these episodes were associated with
significant changes in public debt levels. Typically, the combined effect of new financing
requirements, the depreciation of the exchange rate, changes in real interest rates, and output
contractions led initially to a significant increase in the level of debt as a share of GDP. While
some countries were successful in reducing debt as a share of GDP to pre-crisis levels within
three years following the crisis event,5 debt levels stabilized at relatively high levels for others.
A. Evolution of Debt Levels in the Immediate Post-Crisis Period
Public sector debt as a share of GDP typically increased sharply following the onset of a crisis.
In particular, for the countries in the sample, gross public debt as a share of GDP increased on
average by 36 percentage points from the year prior to the crisis (t–1) to the year following the
crisis (t+1) (Figure 1, left panel). A broadly similar trend can be discerned for net public debt
(gross debt adjusted for international reserve holdings of the central bank), with an average
increase of about 32 percentage points over the same two-year horizon (Figure 1, right panel).
Comparing the evolution of gross and net debt reveals that for some countries, particularly in
Asia, the growth in gross sovereign debt was associated with an increase in central banks’
international reserves.6
5
Due to data constraints, the analysis is confined to a three-year window following the crisis, and thus does not
capture longer-term effects of the post-crisis recovery on debt dynamics. Moreover, since countries differ regarding
the reporting of contingent liabilities in public sector debt figures, cross-country comparisons of post-crisis debt
levels should be interpreted with some caution.
6
Between t–1 and t+1, the stock of international reserves increased by 11 percentage points of GDP in Indonesia,
9 percentage points in Korea, 8 percentage points in the Philippines, and 6 percentage points in Thailand,
respectively.
-5-
Figure 1. Public Sector Debt by Country as a Share of GDP
Gross debt Net debt (gross debt minus international reserves)
12% -4%
KOR THA
39% Pre-crisis average: 44% 21% Pre-crisis average: 36%
43% Post-crisis average: 80% 6% Post-crisis average: 68%
MEX KOR
42% 24%
16% 41%
THA MEX
47% 36%
30% 25%
UKR UKR
51% 47%
44% 38%
BRA BRA
75% 68%
52% 59%
RUS ECU
82% 76%
66% 49%
ECU RUS
82% 77%
25% 18%
IDN IDN
95% 77%
83% 74%
PHL PHL
100% 83%
60% 49%
TUR TUR
101% 86%
58% 42%
URY URY
115% 96%
45% 36%
ARG ARG
135% 125%
0% 20% 40% 60% 80% 100% 120% 140% 160% -20% 0% 20% 40% 60% 80% 100% 120% 140%
Pre-crisis (t–1) Post-crisis (t+1) Pre-crisis (t–1) Post-crisis (t+1)
Sources: World Bank, Global Development Finance; IMF, World Economic Outlook; national authorities; and IMF staff
estimates.
Notwithstanding this general trend, the rise in debt ratios was uneven across countries. For
example, in the three cases that experienced the highest increase in their debt burden (Argentina,
Indonesia, and Uruguay), the net debt-to-GDP ratio rose by more than half the size of GDP over
the two-year period. These exceptional dynamics contrast with the experience of the other
countries in the sample, where the increase in the debt ratio was more moderate. In one case,
Mexico, the ratio declined from 41 percent of GDP in t–1 to 36 percent in t+1.
A decomposition of the change in the debt ratios reveals that the depreciation of the exchange
rate typically contributed most to the worsening of the debt situation (Table 1). On average, the
nominal exchange rate depreciation alone caused the net debt-to-GDP ratio to increase by
36 percentage points. In addition, debt dynamics were affected by crisis-induced new borrowing,
including to fund primary deficits and, in some country cases, financial sector restructuring
(reflected in the residual category of Table 1), changes in real interest rates, and the evolution of
output:
• The impact of exchange rate depreciations was very large in the cases of Argentina, Ecuador,
Russia, and Uruguay, where it led to a rise in the net debt-to-GDP ratio exceeding
50 percentage points in the two-year period beginning at the end of t–1. The impact of an
-6-
exchange rate depreciation on debt levels was, however, partially offset by other factors in the
cases of Ecuador and Russia.
Table 1. Decomposition of Changes in Debt-to-GDP Ratio, from t–1 to t+1 1/

(in percent of GDP from the end of t–1 to the end of t+1)
Individual Contributions to Variations
Change in public Real interest Real GDP Exchange rate

debt-to-GDP ratio Primary balance rate growth depreciation Residual
Argentina (2001) 88.5 –1.1 3.2 6.2 63.4 16.7

Brazil (1998) 30.5 –3.2 19.0 –0.5 16.8 –1.6
Ecuador (1999) 16.9 –11.1 –60.5 1.2 90.7 –3.3
Indonesia (1998) 59.5 –3.3 –5.8 –0.1 22.1 46.7
Korea (1997) 18.7 4.7 0.5 0.7 2.2 10.7
Mexico (1995) –5.2 –14.7 5.4 0.2 38.3 –34.3
Philippines (1998) 8.5 –2.3 –1.3 –2.1 10.6 3.5
Russia (1998) 28.5 0.7 –27.5 –0.1 79.5 –24
Thailand (1997) 25.7 4.2 –0.4 0.6 0.7 20.5
Turkey (2001) 36.9 –9.6 –7.3 –1.7 42.9 12.6
Ukraine (1998) 22.4 0.2 –6.6 0.5 15.7 12.6
Uruguay (2002) 54.3 –3.4 –9.3 2.4 54.2 10.3
Average 32.1 –3.2 –7.6 0.6 36.4 5.9
Sources: World Bank, Global Development Finance; IMF, World Economic Outlook; national authorities; and IMF staff
estimates.
1/ Crisis years in parenthesis.
• The impact of exchange rate movements on debt levels was less pronounced in the cases of
Brazil, Mexico, Turkey, and Ukraine; and was modest in the case of the Southeast Asian
countries, largely because of their relatively small pre-crisis holdings of foreign
currency-denominated debt.
• A significant tightening of the primary fiscal balance helped contain debt dynamics in some
countries, including Ecuador, Mexico, and Turkey. In most other cases, the impact of fiscal
consolidation on the debt dynamics was relatively modest.7
• With the exception of Argentina and Uruguay—where deep recessions contributed to a

significant rise in debt ratios—the impact of output growth on debt dynamics remained
limited.
7
In the cases of Korea and Thailand, expansionary fiscal policies, which were adopted with a view to buttress
domestic demand and thus limit the real effects of their financial crises, actually led to a significant rise in the
debt-to-GDP ratio.
-7-
B. Evidence on Debt Reversals
The ability to roll back or contain the crisis-induced increase in debt levels differed across
countries (Figure 2). This suggests that adjustment programs aimed at recovery from crises
through a combination of fiscal consolidation, growth-enhancing structural reforms, and prudent
monetary policies do not always translate into a significant reduction in the debt burden in the
three years following the onset of a crisis. Furthermore, evidence suggests that the post-crisis
evolution of debt ratios appears to be generally unrelated to the countries’ pre-crisis debt levels.8
Looking at the evolution of (net) sovereign debt ratios over time, countries can be grouped as
follows:
• In some cases, the initial increase in the debt-to-GDP ratio was reversed. In Ecuador, Mexico,
and Russia, debt ratios were at or below their pre-crisis levels within the third year following
the crisis. This can be attributed to a combination of economic adjustment, including fiscal
consolidation, favorable trend in the terms of trade, and, in the cases of Mexico and Russia,
relatively strong economic recovery. Debt restructurings further contributed to improving the
debt dynamics in the cases of Ecuador and Russia, where the debt-to-GDP ratio declined in
2000, the year of a major restructuring in both cases, by 13 percentage points and
32 percentage points, respectively.9 Korea, the Philippines, and Ukraine also succeeded in
broadly neutralizing the impact of the crisis on their sovereign debt levels by the end of the
third post-crisis year, although debt did not fully revert to pre-crisis levels.
• For other countries, the crisis-induced increase in the debt-to-GDP ratio was more persistent.
In the cases of Brazil, Thailand, Turkey, and Uruguay, debt-to-GDP ratios at the end of t+3
(t+2 for the cases of Turkey and Uruguay) remained at about 30 percentage points above their
pre-crisis levels. While the debt-to-GDP ratio exhibited a declining trend for Turkey and
Uruguay, it continued to grow in years t+2 and t+3 in the cases of Brazil and Thailand.10
• In a few cases, net sovereign debt as a share of GDP at t+3 continued to exceed pre-crisis
levels by more than 50 percentage points. Notwithstanding fiscal tightening, negative real
interest rates, and, in the case of Argentina, a strong pick-up in output growth, both Argentina
and Indonesia were not able to reverse significantly the sharp rise in debt ratios in the first
three years following the crisis. This said, in the case of Argentina, net sovereign debt as a
share of GDP declined sharply following the debt exchange with external private sector
creditors in 2005 and the early repayment of IMF credit in January 2006. In Indonesia, the
8
This may be surprising, given that debt ratios reflect, inter alia, past macroeconomic performance and underlying
characteristics of the country such as the quality of institutions that could generally be thought to be relevant for the
success and speed of debt reversals.
9
This evidence tends to confirm the findings of IMF (2003) and Reinhart and others (2003) that substantial reversals
of debt are often linked to debt defaults and/or restructurings.
10
In the case of Brazil, the persistence of high debt levels reflected largely high real interest rates owing to
continued concerns about policy credibility, a further depreciation of the exchange rate, as well as an increase in
IMF exposure in 2001. Thailand’s debt burden remained high since (i) the economy was slow to recover from the
1997 crisis; (ii) the primary balance remained in deficit through t+3; and (iii) high real interest rates persisted.
-8-
debt ratio was reduced by 22 percentage points of GDP between t+3 and t+5, but still
remained high at 49 percent of GDP at end-2003.
Figure 2. Public Sector Debt by Country: Pre-Crisis Levels and Evolution Through t+3 1/ 2/
100
Argentina
80
Change of Debt-to-GDP Ratio from t–1 through t+3
60 Indonesia
Brazil
(in percentage points)
40
Uruguay
Thailand
20 Turkey
Korea Philippines
Ukraine
0
Ecuador
Mexico
-20 Russia
-40
-60
-20% 0% 20% 40% 60% 80% 100%
Pre-Crisis Debt-to-GDP Ratio (t–1)
Sources: World Bank, Global Development Finance; IMF, World Economic Outlook; national authorities; and
IMF staff estimates.
1/ Net debt (gross debt minus international reserves).

2/ For Uruguay, data are only available through t+2.
C. External Debt versus Domestic Debt
External and domestic debt evolved over the crisis period with a clear pattern.11 While domestic
debt increased, on average, by 16 percentage points of GDP through a seven-year time span
beginning two years prior to the crisis event, external debt (net of financing used to build-up
international reserves) contributed only 10 percentage points to the overall growth in the
debt-to-GDP ratio over the same horizon. The relative contribution of the domestic and external
components to overall debt growth varied significantly across the different stages of the crisis
cycle (Figure 3):
• Domestic debt, on average, grew faster than external debt in the run-up to the crisis. This
pattern was strongest in Brazil, Mexico, and Uruguay, where domestic debt in the year prior
to the crisis increased by more than 8 percentage points of GDP, while external public debt
remained broadly constant. A significant factor driving the increase in domestic debt was the
absorption of sovereign debt by the domestic financial system, including because of moral
11
The distinction between the two components is less useful in cases where international and domestic debt markets
had become essentially integrated (as was the case in Argentina, Mexico, Russia, and Uruguay), with nonresidents
holding a large share of domestically issued debt (see Borensztein and others, 2004).
-9-
suasion, in the context of a general decline in investor confidence and in the sovereign’s
access to international capital markets.
• In the midst of a crisis, however, the external debt component tended to dominate the overall
debt dynamics. Typically, external public debt as a share of GDP rose sharply, while the
increases in domestic debt were relatively modest. The sharp increase in the relative share of
external debt can be explained mainly by an exchange rate depreciation effect, given the
significant weight of foreign currency-denominated debt in total external debt, as well as the
countries’ recourse to emergency financial support from multilateral institutions.
Figure 3. Public Sector Debt: Average Evolution of External and Domestic Debt Components 1/
(change in percentage points of GDP)
30%
25%
Domestic External
20%
16% 17%
15%
10%
10% 7%
6%
5% 3%
1%
0%
-5%
-10% -8%
Total window Run-up Crisis Aftermath
t–2 through t+4 t–2 and t–1 t t+1 through t+4
Sources: World Bank, Global Development Finance; IMF, World Economic Outlook; national authorities; and IMF
staff estimates.
1/Net debt (gross debt minus international reserves).
• In the aftermath of crises, the share of domestic debt typically continued to rise while that of
external debt declined. This pattern was pronounced in Southeast Asia, where the
unprecedented issuance of domestic debt was linked to financial sector restructuring efforts.12
More generally, countries had to rely more on domestic sources for new borrowing in the
aftermath of crises after having temporarily lost access to international capital markets (see
IMF, 2005). The decline of external debt-to-GDP ratios was also facilitated by the recovery of
12
Domestic debt ratios for the Asian sample countries increased by 19 percentage points of GDP, on average,
between the end of year t+1 and t+4, but significant cross-country differences are observable: domestic debt grew by
23 percentage points of GDP in Indonesia (after a prior increase of 14 percentage points of GDP in year t), and by
22 percentage points and 20 percentage points in Korea and Thailand. The rise was limited to 10 percentage points
for the Philippines. By contrast, in the non-Asian crisis cases, the average post-crisis domestic debt ratio increased
by 1 percentage point of GDP.
- 10 -
the exchange rate following the initial overshooting and, in some cases, the early repayment
of crisis-related external borrowing as a result of improvements in the balance of payments.
III. CHANGES IN THE STRUCTURE OF SOVEREIGN DEBT
Financial crises are associated not only with changes in the level of public debt, but also in its
composition. In this section, we first elaborate on this issue by illustrating the effect of crises on
the creditor composition of external sovereign debt for the countries included in the sample
(Section III.A). In Section III.B, the focus is then shifted to a discussion of how crisis episodes
were associated with changes in the creditor, currency, and maturity composition of domestic
public debt.
A. External Public Debt13
The resolution of financial crises has generally involved the country seeking recourse to official
financing, particularly from the IMF. The resulting increase in the share of debt to official
creditors, notably multilaterals, led to a significant change in the composition of the debt stock.
In particular, the structure of sovereign debt tended to become more rigid, as claims of
international financial institutions are typically accorded senior creditor status.
• All countries in the sample received multilateral financing at the time of their financial crises.
While exposure to multilateral institutions increased on average by 4.3 percentage points of
pre-crisis GDP in the two years following the onset of the crisis (Figure 4, top panel), there
were significant cross-country differences in terms of the size of emergency financing.14
• The increase in recourse to multilateral financing was most significant in the cases of Turkey
and Uruguay, where changes in the stock of multilateral debt amounted to 17 percentage
points and 9 percentage points of pre-crisis GDP, respectively.
• Recourse to such financing by other countries in the sample was more modest, ranging from
about 0.9 percentage points of pre-crisis GDP in the case of Russia to about 5 percentage
points of pre-crisis GDP in the cases of Korea and Indonesia.
• The IMF was the main source of balance of payments assistance in the context of crisis
resolution. As a result, the IMF claims as a share of the claims of all multilateral creditors
increased rapidly for most countries in the sample (Figure 4, lower panel). While there was a
shift in the composition of multilateral lending towards the IMF, the U.S. dollar exposure of
other multilateral creditors also increased. In some cases, their engagement was substantial:
for example, the World Bank increased its exposure by more than 50 percent between t–1 and
13
The analyses presented in Section III.A are based on gross debt rather than net debt and mainly draw on data from
the World Bank’s Global Development Finance database. Moreover, as data on short-term external sovereign debt
are not available, the analysis is confined to debt having a contractual maturity of one year or more.
14
Interestingly, the cross-country variation with regards to additional multilateral financing does not appear to be
closely related to the pre-crisis debt exposure of the sovereign. This suggests that creditors did not systematically
differentiate the size of their emergency lending packages based on debt sustainability considerations.
- 11 -
Figure 4. Debt to Multilateral Institutions
Evolution in Percent of Pre-Crisis GDP by Country Change in U.S. dollar terms
1%
Brazil 3% Brazil 175%
1% Thailand 184%
Thailand 4%
0% Korea 1,106%
Korea 5%
5% Russia 20%
Russia 5%
5%
Mexico 7% Mexico 47%
8%
Ukraine 10% Ukraine 32%
8% Argentina 41%
Argentina 11%
10%
Philippines 12% Philippines 16%
8%
Indonesia 13% Indonesia 60%
6%
Turkey 15% Turkey 169%
17%
Ecuador 18% Ecuador 5%
12%
Uruguay 29% Uruguay 139%
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
t–1 t+1
Composition of Multilateral Debt, Cross-Country Averages

100%
90%
28% 27% 29% 28% 29%
35% 34%
80%
70%
60%
30% 29%
33% 35%
50% 39%
47% 46%
40%
30%
20% 42% 43%

38% 38%
31%
10% 18% 20%
0%
t–2 t–1 t t+1 t+2 t+3 t+4
IMF IBRD/IDA RDBs
Sources: World Bank, Global Development Finance; IMF, World Economic Outlook; national authorities;
and IMF staff estimates.
- 12 -
t+1 in the cases of Korea and Ukraine; and the Regional Development Banks more than doubled
their lending relative to the pre-crisis period in the cases of Brazil, Korea, and Thailand (see
Table A1, Annex).
The rapid increase in multilateral financing combined with a more modest growth in the
exposure of other creditor groups implied an increase in the share of senior debt in total debt.
Multilateral debt as a share of total (medium- and long-term) external public debt rose on
average by 9 percentage points through the crisis year, from about 24 percent to 33 percent, and
remained at about that level through the following years (Figure 5, top panel).15 This said,
available evidence suggests that bilateral and private sector creditors typically maintained their
exposure to the sovereign at the time of the financial crisis in most of the country cases (see also
Box 1 discussing the relative importance of multilateral lending over time).
• In U.S. dollar terms, official bilateral exposure increased, on average, by 11 percent over the
two years following the onset of a crisis, underscoring the role of official creditors in the
resolution of crises (Figure 5, lower panel). This source of emergency financing was
particularly important in some of the earlier crisis cases included in the sample. In particular,
with the exception of Argentina’s crisis in 2001, all of the other cases in which bilateral
creditors significantly increased their exposure to the country in the context of a crisis
occurred between 1995 and 1998 (Indonesia, Korea, Philippines, Thailand, and Ukraine).16
• Typically, private sector exposure to the sovereign measured in U.S. dollar terms either
remained unchanged at pre-crisis levels or, in some cases, increased at the time of the crisis.
In all cases, except Argentina and Ecuador, private debt exposure in U.S. dollar terms grew
between the end of t–1 and t+1. In some cases, including Korea, Indonesia, and Thailand, this
was facilitated by the recourse to rollover agreements with international banks to maintain
exposure.17
15
The significant increase of multilateral claims can largely be explained by the debt dynamics in a group of five
countries (Argentina, Brazil, Korea, Turkey, and Uruguay). For these countries, the average increase in the share of
multilateral debt in total external sovereign debt amounted to about 18 percentage points between t–1 and t+1, while
it was limited to 3 percentage points for the remaining countries. Within the above group of five countries, initial
conditions differed significantly—in Brazil and Korea the build-up of multilateral claims began from a very low
base, but came on top of already substantial exposure to multilateral institutions in the cases of Argentina, Turkey,
and Uruguay.
16
Interestingly, the strongest average increase in U.S. dollar exposure can be observed for t+2 and thus with
considerable delay relative to the onset of crisis. That said, the average growth rate for t+2 is heavily biased by a
strong increase in exposure to Korea (plus 252 percent) and Argentina (plus 33 percent).
17
For example, in Korea, the authorities rolled over short-term interbank debts into one to three-year government
guaranteed bonds. A similar, though less successful scheme was put into place in Indonesia (see Roubini & Setser,
2004). The fact that such strategies appeared to have been less successful in some of the more recent crisis episodes
(e.g., Brazil, Turkey) could reflect a progressive change in the investor base from international banks to bond
holders and thereby a more diverse investor base that (i) would make it more difficult to overcome collective action
problems, and (ii) does not have the same level of long-term commitment to the country, as is usually the case with
banks. In addition, the banks’ appetite for risk may have declined since, related to the deterioration of the global
environment (September 11, heightened credit risk in the OECD world) and a worsening of bank balance sheets.
- 13 -
Figure 5. External Sovereign Debt: Creditor Composition, Cross-Country Averages 1/
Exposure as a Share of Total Debt
100%
90%
80% 43% 44% 43% 43%

45%
49% 51%
70%
60%
50%
22% 23% 24% 24% 25%
40%
27% 25%
30%
20%
33% 34% 33% 33% 32%
23% 24%
10%
0%
t–2 t–1 t t+1 t+2 t+3 t+4
Multilaterals Bilaterals Private sector
Change in U.S. Dollar Exposure
120%
100%
80%
60%
40%
20%
0%
-20%
t–2 t–1 t t+1 t+2 t+3 t+4
Multilaterals Bilaterals Private sector
Sources: World Bank, Global Development Finance; IMF, World Economic Outlook; national
authorities; and IMF staff estimates.
1/ Medium- and long-term sovereign debt.

- 14 -
Box 1. Did the Relative Importance of Multilateral Lending Increase Over Time?
In recent years, the role of debt-related vulnerabilities in triggering confidence crises has received growing attention among
policymakers and scholars (see, for example, European Central Bank, 2005; Goldstein, 2003). Moreover, high debt
vulnerabilities in some recent crisis cases, including Argentina, Turkey, and Uruguay, may have led to an increased
reluctance of private Debt to Multilateral Institutions by Country: Evolution in Percent of Pre-crisis GDP 1/
sector creditors to
14%
increase or even to
Uruguay
maintain their exposure 12%
(see, for example,
Bordo and others, 2004; 10%
Eichengreen and others,
8%
2005). In this context, Turkey
did the more recent 6%
crisis cases involve a R2 = 0.4362
4%
larger role for official
financing compared to 2% Korea Indonesia
earlier crisis episodes?
In particular, did the 0%
Mexico Thailand Argentina
Brazil Ukraine
size of financing Philippines
–2% Russia Ecuador
packages provided by
the multilateral –4%
institutions actually 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
increase over time?
And did the share of Sources: World Bank, Global Development Finance; IMF, World Economic Outlook; national authorities; and IMF staff
multilateral exposure estimates.
increase relative to 1/ Between t–1 and t+1.

other creditors?
The first chart plots for each country the deviation (in percentage points of pre-crisis GDP) from the average increase in
multilateral exposure between t–1 and t+1. It shows that the rise in multilateral exposure in the two recent crisis cases of
Turkey (2001) and Uruguay (2002) exceeded the average cross-country increase in this debt category (4.3 percent of
pre-crisis GDP) by a wide margin. At the same time, two other recent crisis cases, Ecuador (1999) and Argentina (2001),
did not involve “unusual” amounts of multilateral assistance. Thus, the evidence with regards to the first
question appears to be mixed.
The second chart Debt to Multilateral Institutions by Country: Changes as a Share of Total Debt 1/
shows for each 35%
country the
30% Korea
percentage point
change in the ratio of 25%
multilateral exposure Uruguay
20% Turkey
to total medium- and
Brazil R2 = 0.1241
long-term external 15%
debt between t–1 and Argentina
10%
t+1. It can be Ecuador
observed that the 5% Thailand Indonesia
exposure to Mexico
Russia
0%
multilateral creditors Philippines Ukraine
relative to other –5%
sources of external 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
financing rose more
Sources: World Bank, Global Development Finance; IMF, World Economic Outlook; national authorities; and IMF
strongly in some of staff estimates.
the recent cases
1/ Between t–1 and t+1.
including Turkey and
Uruguay than in most
of the earlier crises. This said, the most significant increase was experienced by Korea, which underwent a
crisis in 1997.
- 15 -
• When crises abated, private sector positions were temporarily unwound in a majority of cases,
including in countries where mechanisms to encourage the maintenance of exposure in the
crisis context were successfully implemented. The decline in exposure was often a
consequence of a temporary loss of access to international capital markets, with spells of
market exclusion ranging from several months to five years and more (see IMF, 2005).18 Over
the longer term, however, countries emerging from crises have typically regained access to
private sector flows, owing largely to credible adjustment programs supported by official
financing (see Box 2 for evidence on the behavior of FDI inflows at the time of crisis).19
• A large increase in the weight of multilateral debt may arguably heighten a country’s
post-crisis vulnerability to the extent that it is associated with very high and rising debt
levels.20 Most adjustment programs supported by multilateral financing have been successful
in restoring confidence and helping crisis countries progress towards debt sustainability. In
rare cases, however, the evolution of the crisis may cause a sovereign’s debt situation to
deteriorate to such an extent that no feasible set of macroeconomic policies exists that would
allow the sovereign to regain medium-term debt sustainability without a restructuring of its
debt. In such circumstances, a high share of senior multilateral debt in total debt may
complicate reaching agreement with creditors on debt relief sufficient to ensure a durable exit
from the crisis. In particular, all else equal, the higher the share of multilateral debt in total
debt and the deeper the reduction of the present value of private claims sought by the
borrower, the more reluctant will private creditors be with regard to their participation in the
debt operation.21
18
Countries that restructured their bonded debt to private creditors—and particularly countries that underwent a
restructuring in a post-default setting—experienced longer spells of market exclusion than others. For example, in
the cases of Ecuador and Ukraine, the sovereign reaccessed international markets only in the fifth and sixth
post-crisis year, respectively (see IMF, 2005).
19
External private sector claims on the sovereign at the end of the fourth post-crisis year exceeded the exposure in
the last pre-crisis year by at least 30 percent in the cases of Mexico, Philippines, and Turkey. By contrast, private
sector exposure to Argentina and Ecuador fell by more than 25 percent (for Argentina, the calculation is based on
t+3). For the remaining countries in the sample, private creditor exposure at the end of t+4 stood at broadly
pre-crisis levels.
20
Reinhart and others (2003) suggest that for most emerging market countries, the external debt-to-GDP ratio
should not exceed 35 percent to be considered “safe,” and could need to be substantially lower in countries with a
history of recurrent crises and/or debt defaults. A study by the IMF (IMF 2003) reaches a similar conclusion. In
general, a sovereign’s liability position is deemed sustainable if it satisfies the solvency condition without a major
policy adjustment and given the costs of financing it faces in the market. The solvency condition stipulates that the
present discounted value (PDV) of a sovereign’s current and future primary expenditure is no greater than the PDV
of its current and future path of income, net of any initial indebtedness. By contrast, the liquidity condition is
satisfied if a sovereign’s liquid assets and available financing are sufficient to meet or roll over its maturing
liabilities. Finally, vulnerability is the risk that the liquidity or solvency conditions are violated and the borrower
enters a crisis (see IMF, 2002, for a more extended discussion of these concepts).
21
Such a scenario is most likely to arise in cases where the sovereign’s solvency is the main source of concern. By
contrast, if a country is facing liquidity pressures but its long-term solvency is not at risk, multilateral financial
assistance should typically be associated with a decrease in vulnerability, as the longer maturities and more
(continued…)
- 16 -
Box 2. Evolution of Foreign Direct Investment Flows in Crisis Episodes
A comprehensive analysis of private sector exposure to a crisis country would need to assess how flows
other than the financing of the sovereign have evolved around the time of the event. Of particular interest
may be the behavior of foreign direct investment (FDI) inflows, as FDI is considered one of the most
stable components of capital flows to developing countries and can also be a vehicle for technological
progress (see, for example, Mody 2004).
Net FDI Inflows: Average Evolution over Time
Using data for the 12 emerging market
countries included in the sample, the
evolution of net FDI inflows scaled to 3.5% 70
Average net FDI inflows as a share of GDP, left scale
GDP suggests that inflows increased, on 3.0% Total net FDI inflows in billions of U.S. dollars, right scale 60
average, from about 1.3 percent of GDP 2.5% 50
in t–1 to 1.5 percent of GDP in t, and
subsequently to 2.5 percent of GDP in 2.0% 40
t+1. This pattern is confirmed by data on 1.5% 30
aggregate FDI flows into the 12 crisis 1.0% 20

countries measured in U.S. dollar terms, 0.5% 10
showing that net inflows continued to
grow strongly after the crisis occurred. In 0.0%
T–2 T–1 T T+1 T+2 T+3
0
fact, Indonesia and Russia were the only -0.5% -10
countries in the sample that experienced -1.0% -20

net FDI outflows in at least one year Sources: World Bank, Global Development Finance; IMF, World Economic Outlook; national authorities,
following the event. and IMF staff estimates.
Notwithstanding the conceptual link, it is difficult to establish a strong association between debt
rigidity and post-crisis vulnerability for any of the sample countries.
• In the majority of the country cases included in the sample, multilateral financing was
successful in supporting an adjustment program aimed at addressing temporary liquidity
pressures and restoring confidence among private investors. In many of these cases, sovereign
bond spreads—a measure of the risk premium requested by private investors compared with a
risk-free asset—declined rapidly to at least pre-crisis levels, allowing the sovereign to regain
reasonably robust access to international capital markets. This, together with the
reconstitution of central bank reserves in the context of successful policy adjustment, typically
helped the debtor to quickly unwind its exposure to multilateral creditors.22
• In some of the more recent cases, however, including Argentina, Ecuador, and Uruguay, the
fact that rigid multilateral claims represented more than one-fifth of total sovereign debt,
which, in turn, stood at elevated levels (exceeded 80 percent of GDP) at the end of t+1,
favorable terms of such financing can be expected to improve the maturity profile of the debt structure and lower
debt service costs.
22
For a discussion of crisis resolution strategies for some of the earlier crisis cases included in the sample, see
Ghosh et al. 2002. Country cases in which sovereign bond spreads (J.P. Morgan’s EMBI or EMBIG) decreased
substantially in the first-post crisis year include Brazil, Korea, the Philippines, and Thailand, while no data is
available for Indonesia, Mexico, and Ukraine.
- 17 -
contributed initially to concerns about post-crisis vulnerability. For example, some private
creditors openly challenged the seniority of multilateral debt in the cases of Argentina and
Ecuador;23 and creditors representing 24 percent of the eligible claims did not participate in
the Argentine debt offer (as of February 25, 2005, the closing date). Debt levels eventually
declined in all of the three country cases—with some delay and in the context of a favorable
external environment—rendering the issue of debt rigidity less critical.24
B. Domestic Public Debt
Evidence suggests that financial crises had an effect on the structure of domestic sovereign debt
as well. In particular, in some of the country cases, domestic debt became more rigid post-crisis
due to a significant increase in the claims of a fragile financial sector, and notably banks, on the
sovereign. Evidence from some of the sample countries also shows that countries typically do
not succeed in improving the maturity and currency composition of domestic public debt in the
immediate post-crisis period.25
Domestic Debt Rigidity
Crisis resolution efforts often resulted in substantially increased holdings of sovereign debt by
the domestic banking sector. While in the context of crises, there was a sharp increase in banks’
claims on the sovereign for about half of the countries in the sample, the post-crisis bank
exposure to the sovereign was particularly high in Argentina, Brazil, Indonesia, Mexico, and
Turkey, where it exceeded 15 percent of GDP in year t+3 (Table 2). Underlying the increased
exposure of the banking sector on the sovereign were typically the large fiscal costs related to the
23
In the context of the litigation arising from Argentina’s default on its external debt, creditors contemplated to rely
on a broad interpretation of the “pari passu” provision in the bonds so as to constrain Argentina from making
payments to any creditors including the IMF. The question of whether creditors could actually use the “pari
passu “clause to attack Argentina’s new payments was raised and deferred in the Southern District Court of New
York in 2004 (see, for example, Gelpern, 2005 and International Monetary and Financial Committee, 2005). In the
case of Ecuador, the seniority of such debt was challenged by several commercial banks. For example, the
Commerzbank, in its Emerging Markets This Week (No. 26/1999, October 15) publication, states that [the IMF and
the World Bank] “will be concerned with protecting their own balance sheets rather than with fair ‘burden
sharing’” [and that therefore the] “IMF and the World Bank are not suited either as arbitrators or as objective
regulators of sovereign insolvency procedures.” (See http://www.new-rules.org/docs/sdrm0902.pdf). By contrast,
the 2003 bond exchange in Uruguay was well received by investors, but it did hardly affect the sovereign’s
debt-to-GDP ratio.
24
While multilateral claims at the end of 2005 still amounted to more than 32 percent of GDP in the case of
Uruguay, Argentina fully retired its outstanding debt to the IMF of US$9.5 billion (5.3 percent of GDP) in
January 2006 and hence reduced its liabilities to multilateral institutions to about 10 percent of GDP. Ecuador’s
multilateral debt represented 12 percent of GDP at end-2005, most of which was owed to the World Bank and
Regional Development Banks. For a recent discussion of these countries’ debt sustainability outlook, see IMF, 2006.
25
However, given data constraints, the results presented in this section should be interpreted with caution: the
analysis of changes in the maturity structure is based on the experience of six countries; the discussion of the
currency decomposition is informed by data from five crisis cases.
- 18 -
Table 2. Domestic Banking Sector Claims on Public Sector 1/ 2/

(in percent of GDP)
t–1 t+1 t+3
Argentina 7 26 23
Brazil 11 12 17
Indonesia 1 25 29
Korea 1 3 4
Mexico 7 7 16
Philippines 13 11 15
Russia 9 10 7
Turkey 28 29 26
Thailand 0 3 6
Uruguay 5 9 6
Sources: IMF, International Financial Statistics and World Economic Outlook.
1/ Includes only countries with banking sector exposure to government exceeding

5 percent of GDP in at least one year between t–1 and t+3.
2/ Deposit bank claims on central government for all countries, except Russia and
Ukraine (general government, and Turkey (government).
resolution of financial sector crises.26 In the cases for which data are available (Ecuador,
Indonesia, Korea, Mexico, Thailand, and Turkey), estimates of net costs range between
19 percent and 52 percent of GDP, comprising, inter alia, outlays for liquidity assistance from
the central bank, government guarantees of deposits, and bank recapitalization (see Hoelscher
and others, 2003). These outlays often represented a substantial share of overall post-crisis public
debt, particularly in the Southeast Asian countries.
As in the case of senior multilateral debt, a high exposure of the financial sector to the sovereign
may add to the sovereign’s vulnerability by rendering its debt structure more rigid. This said, the
risks associated with the rigidity of these liabilities of the sovereign have to be weighed against
the potentially benign impact of such obligations on debt servicing costs and the maturity and
currency composition of public debt.
26
The sources of financial sector crises varied across the sample countries. In Korea, Indonesia, and Thailand,
investor concerns focused on imbalances in the private sector, including heavily leveraged corporate sectors, inflated
asset values, and large unhedged short foreign currency positions (see Lindgren and others, 1999). In other
countries, including Argentina, Ecuador, Russia, and Uruguay, banking crises were closely associated with
sovereign debt distress. That said, in most cases, the financial sector showed weaknesses already in the run-up to the
financial crisis. For a more detailed discussion of financial sector restructuring costs, see, for example, Hemming
and others (2003), Hoelscher and others (2003), Honohan and Klingebiel (2000), and Lindgren and others (1999).
- 19 -
• As long as banks’ financial conditions—including their capital, liquidity, asset quality, and
exposure to market risk—are strong enough to absorb the additional government paper,
increasing bank exposure to a weak sovereign may not be a major source of concern. In fact,
in such circumstances, domestic banks can become an important source of financing for the
sovereign when access to other alternatives dries up in the run—up to a crisis (see
Section II.C).
• However, in situations where the health of the banking system has already been weakened by
crisis dynamics and claims on the government represent a high share of total bank assets, the
stability of the sector may become closely tied to a continued performance of public sector
assets.27 When a high bank exposure to the sovereign—which is often encouraged by
regulatory frameworks that consider such claims to be riskless assets—coincides with an
unsustainable debt level of the sovereign requiring a debt restructuring operation, a tradeoff
may arise between two competing objectives: restoring a viable debt and debt-service profile
and preserving the soundness of the banking system. The intensity of the tradeoff will likely
depend on the state of the banking sector, its exposure to the sovereign relative to other
creditor groups, and the required size of the haircut to restore debt sustainability.
Although banking sector exposure to the sovereign increased considerably in some of the cases
considered, the associated risks were reduced by efforts to limit the impact on the domestic
banking sector in several countries where a debt restructuring became unavoidable.
For example, the Argentine sovereign was successful in restructuring most of its liabilities to the
domestic banking sector on a voluntary basis in November 2001, which provided
the basis for excluding about 85 percent of the banking sector’s total sovereign exposure from
the sovereign default that occurred only one month later. Despite the NPV loss due to coupon
reduction and maturity extension, the direct effect of the sovereign default was thus contained as
the restructured debt continued to be serviced and government debt was not marked to market.28
In Uruguay, the domestic banks’ voluntary participation in the 2003 debt exchange—which was
much smaller in scope than the Argentine operation—was facilitated by a menu of instruments
designed to meet their needs to avoid a principal haircut.
27
This pattern emerges, for instance, in the cases of Argentina and Indonesia, where claims of banks on the central
government as a share of their total assets increased to 58 percent and 53 percent at the end of t+1, respectively,
from 20 percent and 17 percent at the end of t–1. For a discussion of the impact of the Argentine crisis on the
banking sector, and cross-bank variation in vulnerability, see also Barajas et al. (2006).
28
Banks, however, incurred large losses from the policy response to the economic crisis—including the asymmetric
pesoization of banks’ assets and liabilities and the impact of judicial injunctions—which affected banking sector
soundness and led to the issuance of US$21 billion in sovereign debt to the banking sector. As a consequence,
banking sector exposure to the sovereign increased to 55 percent of assets by end-2004.
- 20 -
Maturity and Currency Composition
No clear cross-country trends can be established regarding the evolution of domestic public debt
maturities in the context of crises (Figure 6, top panel).29
• Available evidence suggests that the maturity composition of domestic debt did not change
significantly for most countries in the run-up to crises. In fact, Uruguay is the only case out of
the five countries for which data are available (Indonesia, Mexico, Thailand, Turkey, and
Uruguay) and where a modest growth in the share of short maturities can be detected.
• By contrast, three out of a total of five countries experienced a substantial increase in short-
term debt after the crisis. This pattern was most pronounced in the case of Turkey, where the
share of short maturities tripled through the end of the crisis year (representing a rise by
13 percentage points).30 Smaller increases can be observed for Thailand and Uruguay, while
Mexico succeeded in reducing temporarily the share of short-term debt in total debt before
experiencing a reversion to broadly pre-crisis levels in the second year after the crisis (1997).
The only case where any short-term exposure was avoided throughout the crisis episode was
Indonesia. The fact that short-term debt often continued to grow well past the event suggests
that investors were often not willing to accept a lengthening of maturities at costs acceptable
to the sovereign.
While foreign currency debt often increased pre-crisis, most countries were unable to reduce
their vulnerability to sudden exchange rate changes after the crisis event (Figure 6, lower panel):
• Governments typically made greater use of foreign currency-denominated debt in the runup to
a crises.31 The most prominent case is Mexico, where the government swapped large amounts
of peso-denominated treasury bills for tesobonos indexed to the U.S. dollar, both to reduce its
funding costs and to underscore its commitment to the exchange rate peg. On a smaller scale,
29
This finding partially reflects the heterogeneity in the countries’ pre-crisis debt structures (see also Borensztein
and others, 2004): Uruguay relied almost exclusively on instruments with a contractual maturity of less than one
year, and Mexico for about 40 percent of its total domestic debt stock. In Indonesia, Thailand, and Turkey, more
than 80 percent of the sovereign’s pre-crisis domestic debt was longer term.
30
The continuing decline in the average maturity of Turkey’s outstanding domestic debt stock mainly reflects the
fact that the large stock of so-called noncash debt (bank recapitalization bonds issued during the crisis), which
typically carried longer maturities than regular paper, was increasingly substituted with regular debt.
31
This finding is consistent with the results of a recent study by Frankel and Wei (2004), which finds that
U.S. dollar-denominated debt is a good crisis predictor in EMCs.
- 21 -
Figure 6. Domestic Public Sector Debt by Country: Maturity and Currency Composition
Maturity Composition 1/
(share of short-term debt in total domestic debt)
100% 100%
Uruguay
90% 90%
80% 80%
70% 70%
60% 60%
Mexico
50% 50%
40% 40%
30% 30%
Thailand
20% 20%
10% Indonesia
10%
Turkey
0% 0%
t–2 t–1 t t+1 t+2 t–2 t–1 t t+1 t+2
Currency Composition
(share of foreign currency debt in total domestic debt)
100% Uruguay
100%
90% 90%
80% 80%
70% 70%
60% 60%
50% 50%
40% 40%
Turkey
30% 30%
20% 20%
Brazil Mexico
Indonesia
10% 10%
0% 0%
t–2 t–1 t t+1 t+2 t–2 t–1 t t+1 t+2
Sources: World Bank, Global Development Finance; IMF, World Economic Outlook; national authorities; and IMF staff estimates.
1/ For Uruguay, data are only available through t+1.

- 22 -
a rise in the share of foreign currency liabilities in total domestic public debt can also be
observed for Brazil, Turkey, and Uruguay (see Ghosh and others, 2002; and IMF 2004 for the
latter cases).32
• Following the crisis, only a minority of countries for which data are available were able to
reduce their reliance on foreign currency debt. In Mexico, the tesobonos were fully retired by
the end of 1995, which effectively eliminated the foreign currency component in domestic
borrowing (see Collyns and others, 2003). The Uruguayan government was also able to
marginally reduce its dependence on foreign currency debt, but at the expense of extremely
high interest rates. By contrast, the public sector had to rely on increased foreign currency
issuances as a means to retain market access following the crisis in the cases of Brazil,
Indonesia, and Turkey.33
• The evidence suggests that countries generally faced significant difficulties in reducing their
exposure to maturity or currency risk after a crisis. While this conclusion is supported for a
majority of countries for which relevant data are available with regard to either of the two risk
types, it is even more compelling when looking at the parallel evolution of the maturity and
currency composition of the countries’ domestic public debt: in no case did a sovereign
succeed in reducing both risks associated with the structure of its domestic debt at the same
time. In fact, in such situations characterized by a high degree of investor uncertainty,
countries may often have no choice but to rely on short-term and/or foreign
currency-denominated debt to induce the private sector to hold sovereign paper.
IV. CONCLUSIONS
The evidence for 12 capital account crises that took place since the mid-1990s indicates that
sovereigns typically emerged from an episode of financial distress with a weaker debt profile
compared to the pre-crisis situation. As a result, concerns over the public sector’s vulnerability
may have actually increased in some cases. While outcomes varied due to country-specific
circumstances, and strong conclusions are difficult to draw, the following stylized facts are
noteworthy.
32
A pre-crisis shift to foreign currency instruments did not take place in the case of Indonesia, which exclusively
relied on local currency instruments. Apparently, Indonesia (and the other Southeast Asian countries) enjoyed a
higher degree of investor confidence prior to 1997–98 compared to many of the other countries in the sample, based
on the credibility of the countries’ anti-inflationary monetary stance, prudent fiscal policies reflected by relatively
low public sector borrowing requirements, and well-developed domestic capital markets.
33
In the case of Turkey, the sharp increase in foreign currency-denominated domestic debt in year t was largely due
to a debt swap in June 2001, in which the government exchanged about US$5 billion worth of local currency paper
for dollar-indexed bonds to help domestic banks close the currency mismatches on their balance sheets. That said,
through May 2004, about one third of the total volume of domestic bonds issued was denominated in foreign
currency. In Brazil, the share of foreign currency-denominated instruments in total post-crisis domestic borrowing
was a more modest seven percent through May 2004, but the country also relied extensively on interest rate- and
inflation-linked instruments.
- 23 -
In general, financial crises were associated with higher sovereign debt levels in the emerging
market countries involved.
• Crises typically cause an increase in public sector debt ratios, driven in particular by the
impact of severe exchange rate depreciations. Specifically, the debt-to-GDP ratio for gross
(net) public debt increased on average by 36 (32) percentage points within two years from the
onset of the crisis.
• For most countries, the increase in debt levels persisted for several years. Only three out of
the twelve countries in the sample (Ecuador, Mexico, and Russia) were able to fully roll back
the crisis-induced increase in their debt-to-GDP ratios within the three-year horizon examined
with the available data. All of these countries benefited from favorable trends in their terms of
trade; Ecuador and Russia also benefited from a debt restructuring.
• Typically, the rise in (net) debt-to-GDP ratios over the crisis cycle was driven more by
domestic rather than external debt, although their relative shares in total debt differed at
different stages of the crisis cycle. New domestic borrowing in the context of crisis resolution
was often linked to financial sector restructuring, particularly in the case of the Asian crisis
countries.
Financial crises were also associated with changes in debt structures.
• In particular, there was an increase in sovereign liabilities that could contribute to greater
rigidity of the debt structure in all of the countries in the sample. This pattern resulted from
both an increase in the share of multilateral claims in total public debt associated with the
financial assistance needed in order to resolve the crisis; and a heightened exposure of a weak
domestic banking sector to government debt as a consequence of efforts aimed at resolving
problems in the banking sector.
• There is, however, little evidence to suggest that debt rigidity represented a persistent source
of significant vulnerability for the countries in the sample. In particular, in the case of 6 of the
12 countries in the sample, rigid liabilities exceeded 20 percent of GDP at the end of the first
post-crisis year, but they were all successful in reducing debt levels soon thereafter.34 This
helped these countries make progress towards debt sustainability, and thus rendered the issue
of debt rigidity—which could have complicated a debt restructuring operation in the event
that it was required to reduce unsustainably high debt levels—less critical.
34
This group of countries includes Uruguay (58), Argentina (56), Indonesia (46), Turkey (46), Ecuador (29), and the
Philippines (23). The figures in parentheses indicate the respective country’s ratio of rigid liabilities to GDP,
calculated as the sum of public debt owed to multilateral institutions and to the domestic banking sector.
- 24 -
• Evidence from a limited set of cases suggests that countries generally faced significant
difficulties in reducing their exposure to maturity or currency risk after a crisis. In no case did
a sovereign succeed in reducing both types of risk at the same time. Data limitations
notwithstanding, this finding confirms that improving the composition of sovereign debt at
times of heightened uncertainty is a serious challenge for policymakers.
- 25 - ANNEX
Data Issues and Methodology
For each country, the crisis year was determined on the basis of qualitative assessments and
available case studies. This follows from the fact that a quantitative definition, like the
familiar pressure index for currency crises, does not work well
due to the wide range of crisis types considered, including Crisis years (t)
currency, banking, and debt crises. It also reflects common Argentina 2001
practice, as, for example, in Kaminsky and Reinhart’s twin crisis Brazil 1998
paper (1999) or in Caprio and Klingebiel (2003). Reported results Ecuador 1999
are not very sensitive to changes in crisis dates, as a window Indonesia 1998
spanning from the end of the last pre-crisis year through the end Korea 1997
of the first post-crisis year is used for most analyses, and, at times, Mexico 1995
is even expanded further. Philippines 1998
Russia 1998
The country sample comprises the capital account crises cases Thailand 1997
included in Gosh and others (2002), augmented by five countries. Turkey 2001
First, Argentina, Turkey, and Uruguay are included as prominent Ukraine 1998
recent crises cases. Second, Russia and Ukraine were added to Uruguay 2002
broaden the sample of restructuring cases. A unifying feature of
all these crises is that they combined sharp reversals of capital inflows, triggering a currency
crisis, with other factors which were critical in negatively affecting investor confidence, like
adverse public debt dynamics (Argentina, Brazil, and Turkey), a risky public debt
management strategy (Mexico), and pervasive financial sector weaknesses
(Indonesia, Korea, and Thailand).
The study draws on several data sources. External debt data are mainly drawn from the
World Bank’s Global Development Finance CD-ROM (2003), and are updated with data
collected from IMF country economists. Data on Supplementary Data Sources
domestic public debt are based mostly on submissions Argentina EBS/03/130, Suppl.1;
of the country desks, and are completed with input SM/05/193; SM/05/107;
from the Fiscal Affairs Department’s debt database as and staff estimates
well as national sources. GDP and exchange rate data Brazil EBS/05/39
Ecuador EBS/03/2, Suppl. 1 and
are drawn from the World Economic Outlook (WEO) FAD database
database, and banking sector data from the IMF’s Indonesia EBS/03/132; EBS/03/35
International Financial Statistics. Korea SM/04/23 and FAD
database
For the decomposition of the increase in debt-to-GDP Philippines EBS/03/107 and FAD
ratios around crisis periods, the study uses, where database
available, macroeconomic data contained in IMF Russia SM/04/269 and FAD
database
Article IV staff reports. Missing data points were Thailand SM/03/266
filled by recourse to the IMF’s WEO database. Given Turkey EBS/05/66; Turkish
that the decomposition analyses use the debt data Treasury; and staff
newly constructed for this project, and are based on estimates
net debt, the results differ from the decompositions Ukraine SM/03/113
reported in the respective country reports. Uruguay EBS/05/23; EBS/05/78;
and staff estimates
- 26 - ANNEX
Figure A1. External Sovereign Debt by Country: Creditor Composition

(as a share of MLT debt)
Argentina Brazil
100% 100%
90% 90%
80% 80%
59% 57% 57% 55% 70% 57%
70% 62% 64%
69% 69% 68% 68%
74% 75%
60% 60%
50% 50%
40% 5% 6% 8% 8% 40% 8%
12% 9%
30% 7% 5% 30%
11% 11%
20% 36% 37% 36% 37% 20% 17% 14% 36%
24% 26% 26% 28%
10% 10% 21% 21%
10% 12%
0% 0%
t–2 t–1 t t+1 t+2 t+3 t+4 t–2 t–1 t t+1 t+2 t+3 t+4
Ecuador Indonesia
100% 100%
90% 90% 21% 18% 20% 17%
23% 23% 24%
80% 42% 39% 39% 38% 80%
55% 53% 52%
70% 70%
60% 60% 43%
44% 43% 47%
45% 41%
22% 48%
50% 21% 23% 24% 50%
40% 17% 18% 40%
17%
30% 30%
20% 37% 38% 37% 40% 20% 35% 36% 38% 37% 36%
28% 30% 30% 29% 32%
10% 10%
0% 0%
t–2 t–1 t t+1 t+2 t+3 t+4 t–2 t–1 t t+1 t+2 t+3 t+4
Korea Mexico
100% 100%
90% 90%
80% 80%
56% 56% 55% 54% 54% 57%
70% 60% 65%
58% 70% 61% 64%
70%
76% 60%
60% 82%
50% 4% 2%
50%
40% 8% 7% 40% 14%
19% 20% 14%
6% 12%
30% 30% 11%
10%
20% 11% 40% 38% 37% 35%
20%
9% 29% 32% 29% 27%
27% 26% 25%
10% 10% 21%
13% 9%
0% 0%
t–2 t–1 t t+1 t+2 t+3 t+4 t–2 t–1 t t+1 t+2 t+3 t+4
Multilaterals Bilaterals Private Sector
- 27 - ANNEX
Figure A1 (cont'd). External Sovereign Debt by Country: Creditor Composition

(as a share of MLT debt)
Philippines Russia
100% 100%
90% 24% 90%
28% 28% 31% 34% 38% 34% 36% 37% 36% 33% 35%
80% 40% 80% 39%
70% 70%
60% 60%
46% 42% 41%
50% 41% 50%
38% 35% 36% 52%
40% 40% 53% 49% 44% 48% 52%
48%
30% 30%
20% 20%
31% 30% 32% 28% 28% 27% 25%
10% 10% 15% 18% 16% 15%
13% 13% 13%
0% 0%
t–2 t–1 t t+1 t+2 t+3 t+4 t–2 t–1 t t+1 t+2 t+3 t+4
Thailand Turkey
100% 100%
90% 24% 23% 23% 90%
27% 28%
34% 37%
80% 80%
57% 53% 55% 58%
70% 70% 62%
68% 68%
60% 60%
50% 49% 50%
54% 54% 53% 50%
47% 11%
40% 47% 40% 12%
10%
10%
10%
30% 30%
17% 14%
20% 20% 36% 35%
31% 32% 28%
10% 24% 22% 22% 23% 23% 10%
19% 16% 15% 18%
0% 0%
t–2 t–1 t t+1 t+2 t+3 t+4 t–2 t–1 t t+1 t+2 t+3 t+4
Ukraine Uruguay
100% 100%
90% 24%
90%
28% 28% 25% 25%
31%
80% 36% 80% 40% 41%
42% 43%
49%
70% 58% 56%
70% 2%
3% 2% 2%
60% 30% 30% 60% 2%
38% 31% 30% 26%
23%
50% 50%
40% 40% 6%
7%
30% 30% 54% 58% 57% 55%
45% 49%
42% 42% 44% 44%
20% 38% 41% 20% 35% 38%
10% 10%
0% 0%
t–2 t–1 t t+1 t+2 t+3 t+4 t–2 t–1 t t+1 t+2 t+3 t+4
Multilaterals Bilaterals Private Sector
- 28 - ANNEX
Table A1. Public Sector Debt by Country: Evolution from t–2 to t+2
(in U.S. dollar billion)
t–2 t–1 t t+1 t+2
Argentina (2001) 1999 2000 2001 2002 2003
Total public sector debt 121.9 128.1 143.8 137.0 177.8

External public and publicly guaranteed debt 85.0 85.2 87.5 83.8 86.3
Official creditors 26.2 26.4 36.1 35.7 37.5
Multilaterals 20.3 21.8 31.6 30.8 31.0
IMF 4.5 5.1 14.0 14.3 15.5
IBRD / IDA 8.6 9.1 9.4 8.5 7.5
Others 7.2 7.6 8.2 8.0 8.0
Bilaterals 5.9 4.6 4.5 4.9 6.5
Private sector creditors 58.8 58.8 51.4 48.1 48.8
Debt to commercial banks 0.9 0.8 0.6 0.5 0.4
Bonds and Others 57.9 58.0 50.8 47.5 48.4
Domestic Debt 36.9 42.9 56.3 53.2 91.5
Of which: Foreign currency denominated debt n.a. n.a. n.a. n.a. n.a.
Of which: Short-term debt n.a. n.a. n.a. n.a. n.a.
Brazil (1998) 1996 1997 1998 1999 2000

Multilaterals 9.5 10.1 21.1 27.8 20.7
IMF 0.1 0.0 4.8 8.8 1.7
IBRD / IDA 5.9 5.7 6.3 6.8 7.4
Others 3.5 4.3 10.0 12.2 11.6
Bilaterals 15.9 12.1 11.5 12.1 11.0
Bonds and Others 60.3 53.9 53.1 62.8 64.1
Domestic Debt 197.0 269.9 385.3 283.9 349.5
Of which: Foreign currency denominated debt 18.5 41.5 80.9 68.8 77.9
Ecuador (1999) 1997 1998 1999 2000 2001

Multilaterals 3.5 3.9 4.0 4.1 4.3
IMF 0.1 0.1 0.0 0.1 0.2
IBRD / IDA 0.9 0.9 0.9 0.9 0.9
Others 2.4 2.9 3.1 3.1 3.1
Bilaterals 2.2 2.3 2.4 2.3 2.6
Bonds and Others 6.4 6.5 6.6 4.2 4.2
Domestic Debt 2.1 2.3 3.0 2.1 2.3
Of which: Short-term debt n.a. n.a. n.a. 0.1 0.1
- 29 - ANNEX
t–2 t–1 t t+1 t+2
Indonesia (1998) 1996 1997 1998 1999 2000

Multilaterals 17.2 18.8 27.0 30.0 30.7
IMF 0.0 3.0 9.1 10.2 10.8
IBRD / IDA 11.9 10.7 11.4 12.1 12.4
Others 5.4 5.1 6.5 7.6 7.5
Bilaterals 28.9 26.7 31.2 36.3 35.2
Bonds and Others 7.9 7.5 7.2 5.8 4.8
Domestic Debt 0.0 0.0 14.6 61.7 75.4
Of which: Short-term debt 0.0 0.0 0.0 0.0 0.0
Korea (1997) 1995 1996 1997 1998 1999

Multilaterals 2.8 2.4 18.0 28.5 18.6
IMF 0.0 0.0 11.1 16.9 6.1
IBRD / IDA 2.2 1.9 4.6 7.5 8.4
Others 0.6 0.4 2.3 4.1 4.1
Bilaterals 2.5 2.2 1.9 1.2 4.1
Bonds and Others 12.3 16.2 22.1 24.6 24.6
Domestic Debt 39.1 39.5 44.3 61.5 108.9
Mexico (1995) 1993 1994 1995 1996 1997

Multilaterals 21.1 21.3 34.6 31.2 25.8
IMF 4.8 3.8 15.8 13.3 9.1
IBRD / IDA 12.3 13.0 13.8 12.7 11.5
Others 3.9 4.4 5.0 5.3 5.2
Bilaterals 15.0 16.7 15.4 14.6 11.2
Bonds and Others 36.5 38.5 53.0 54.6 51.3
Domestic Debt 50.3 91.0 38.4 27.8 37.3
- 30 - ANNEX
t–2 t–1 t t+1 t+2
Philippines (1998) 1996 1997 1998 1999 2000

Multilaterals 8.3 8.2 9.5 9.5 9.1
IMF 0.4 0.9 1.6 1.8 2.0
IBRD / IDA 4.9 4.4 4.5 4.3 3.4
Others 3.1 2.9 3.5 3.4 3.7
Bilaterals 12.4 11.3 12.4 13.6 12.4
Bonds and Others 6.1 6.8 6.8 8.2 9.2
Domestic Debt 40.2 40.3 31.1 41.5 42.5
Russia (1998) 1996 1997 1998 1999 2000

Multilaterals 15.3 18.5 25.9 22.2 18.7
IMF 12.5 13.2 19.3 15.2 11.6
IBRD / IDA 2.5 5.1 6.3 6.7 6.8
Others 0.3 0.2 0.2 0.2 0.2
Bilaterals 61.0 58.9 62.8 64.9 63.7
Bonds and Others 23.0 13.5 23.1 20.5 40.0
Domestic Debt 0.0 89.8 56.9 23.2 19.8
Thailand (1997) 1995 1996 1997 1998 1999

Multilaterals 3.2 2.7 6.0 7.7 8.9
IMF 0.0 0.0 2.4 3.2 3.4
IBRD / IDA 1.9 1.7 1.8 2.2 2.8
Others 1.3 1.0 1.8 2.3 2.6
Bilaterals 8.0 8.0 12.2 17.3 21.7
Bonds and Others 2.7 2.9 3.6 3.9 3.6
Domestic Debt 12.5 12.7 10.4 17.6 28.4
- 31 - ANNEX
t–2 t–1 t t+1 t+2
Turkey (2001) 1999 2000 2001 2002 2003

Multilaterals 7.8 11.5 22.1 30.8 33.5
IMF 0.9 4.2 14.1 22.1 24.1
IBRD / IDA 3.0 3.7 4.8 5.3 5.8
Others 3.9 3.6 3.1 3.4 3.6
Bilaterals 9.1 8.7 8.6 9.3 9.5
Bonds 16.7 22.1 21.3 23.8 27.6
Domestic Debt 54.9 58.4 99.7 99.6 130.0
Ukraine (1998) 1996 1997 1998 1999 2000

Multilaterals 3.4 3.9 4.8 5.2 4.5
IMF 2.3 2.4 2.8 2.8 2.1
IBRD / IDA 0.9 1.1 1.5 2.0 2.0
Others 0.3 0.4 0.5 0.4 0.4
Bilaterals 3.4 2.9 2.7 3.8 2.6
Bonds and Others 2.0 2.0 3.5 3.0 3.0
Domestic Debt 2.0 5.5 4.7 3.6 3.8
Uruguay (2002) 2000 2001 2002 2003 2004

Multilaterals 2.1 2.2 4.5 5.3 5.6
IMF 0.1 0.1 1.8 2.3 2.6
IBRD / IDA 0.5 0.6 0.7 0.7 0.8
Others 1.4 1.5 2.0 2.2 2.2
Bilaterals 0.4 0.3 0.3 0.2 0.2
Bonds and Others 2.9 2.8 3.1 3.6 3.9
Domestic Debt 3.3 5.0 4.0 3.1 1.2
Of which: Short-term debt 2.9 4.6 3.7 2.9 n.a.
- 32 -
References
Barajas, Adolfo, Emiliano Basco, V. Hugo Juan-Ramon, and Carlos Quarracino, 2006,
“Banks During the Argentine Crisis: Were They All Hurt Equally? Did They All
Behave Equally?” IMF Working Paper 06/42 (Washington: International Monetary
Fund.
Bordo, Michael, and others, 2004, “Keeping Capital Flowing: The Role of the IMF,” IMF
Working Paper 04/197 (Washington: International Monetary Fund).
Borensztein, Eduardo, and others, 2004, Sovereign Debt Structure for Crisis Prevention, IMF
Occasional Paper No. 237 (Washington: International Monetary Fund).
Caprio, Gerard, and Daniela Klingebiel, 2003, “Episodes of Systemic and Borderline
Financial Crises” (unpublished; Washington: World Bank).
Collyns, Charles, and G. Russell Kincaid, eds., 2003, Managing Financial Crises: Recent
Experience and Lessons for Latin America, IMF Occasional Paper No. 217
(Washington: International Monetary Fund).
Eichengreen, Barry, and others, 2005, “The IMF in a World of Private Capital Market,” IMF
Working Paper 05/84 (Washington: International Monetary Fund).
European Central Bank, 2005, Managing Financial Crises in Emerging Market Economies –
Experience with the Involvement of Private Sector Creditors (Frankfurt: European
Central Bank).
Frankel, Jeffrey, and Shang-Jin Wei, 2004, “Managing Macroeconomic Crises: Policy
Lessons” (unpublished; Washington: World Bank).
Gelpern, Anna, 2005, “After Argentina,” Policy Briefs in International Economics PB05–2
(Washington: Institute for International Economics).
Ghosh, Atish, and others, 2002, IMF-Supported Programs in Capital Account Crises, IMF
Goldstein, Morris, 2003, “Debt Sustainability, Brazil, and the IMF,” IIE Working Paper 03–1
(Washington: Institute for International Economics).
Hemming, Richard, and others, 2003, Fiscal Vulnerability and Financial Crises in Emerging
Market Economies, IMF Occasional Paper No. 218 (Washington: International
Monetary Fund).
Hoelscher, David S., and Marc Quintyn, 2003, Managing Systemic Banking Crises, IMF
Honohan, Patrick, and Daniela Klingebiel, 2000, “Controlling Fiscal Costs of Banking
Crises” (unpublished; Washington: World Bank).
International Monetary and Financial Committee, 2005, Progress Report to the International
Monetary and Financial Committee on Crisis Resolution, IMFC/Doc/11/05/4.
Available via the Internet: http://www.imf.org/external/np/pp/eng/2005/041205.htm.
International Monetary Fund, 2002, Assessing Sustainability, SM/02/166, 2/28/02. Available
via the Internet: http://www.imf.org/external/np/pdr/sus/2002/eng/052802.pdf.
- 33 -
——2003, World Economic Outlook, September 2003: Public Debt in Emerging Markets: Is
it Too High? World Economic and Financial Surveys (Washington: International
Monetary Fund).
——2004, Debt-Related Vulnerabilities and Financial Crises–An Application of the Balance
Sheet Approach to Emerging Market Countries, SM/04/210, 7/2/04. Available via the
Internet: http://www.imf.org/external/np/pdr/bal/2004/eng/070104.pdf
——2005, Assessing the Determinants and Prospects for the Pace of Market Access By
Countries Emerging From Crisis – Further Considerations, SM/05/76, 3/2/05.
Available via the Internet: http://www.imf.org/external/np/pp/eng/2005/030105a.pdf.
——2006, Cross-Country Experience with Restructuring of Sovereign Debt and Restoring
Debt Sustainability, SM/06/26, 1/23/06.
Kaminsky, Graciela, and Carmen M. Reinhart, 1999, “The Twin Crises: The Causes of
Banking and Balance-of-Payments Problems,” American Economic Review, Vol. 89,
No. 3, pp. 473–500.
Lindgren, Carl-Johan, and others, 1999, Financial Sector Crisis and Restructuring Lessons
from Asia, IMF Occasional Paper No.188 (Washington: International Monetary
Fund).
Mody, Ashoka, 2004, “Is FDI Integrating the World Economy?” World Economy, Vol. 27,
No. 8, pp. 1195–1222.
Reinhart, Carmen M., Kenneth S. Rogoff, and Miguel A. Savastano, 2003, “Debt
Intolerance,” NBER Working Paper No. 9908 (Cambridge, Massachusetts: National
Bureau of Economic Research).
Reinhart, Carmen M., and Kenneth S. Rogoff, 2004, “Serial Default and the ‘Paradox’ of
Rich to Poor Capital Flows,” NBER Working Paper No. 0296 (Cambridge,
Massachusetts: National Bureau of Economic Research).
Roubini, Nouriel, and Brad Setzer, 2004, Bailouts or Bail-Ins? Responding to Financial
Crises in Emerging Markets (Washington: Institute of International Economics).
Sturzenegger, Federico, 2003, “Default Episodes in the 90s: Factbook, Toolkit and
Preliminary Lessons” (unpublished; Buenos Aires: Universidad Torcuato Di Tella).

The Level and Composition of Public Sector Debt in Emerging Market Crises

Uploaded by

Copyright:

The Level and Composition of Public Sector Debt in Emerging Market Crises

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

The Level and Composition of Public Sector Debt in Emerging Market Crises

Uploaded by

Copyright:

WP/06/186

The Level and Composition of

IMF Working Paper

Policy Development and Review Department

Prepared by Monica de Bolle, Björn Rother, and Ivetta Hakobyan1

Authorized for distribution by Mauro Mecagni

JEL Classification Numbers: F30 and F34

Keywords: Financial crises, debt distress, sovereign debt

Author(s) E-Mail Address: [email protected]

II. Evolution of Sovereign Debt Levels .....................................................................................4

III. Changes in the Structure of Sovereign Debt......................................................................10

II. EVOLUTION OF SOVEREIGN DEBT LEVELS

A. Evolution of Debt Levels in the Immediate Post-Crisis Period

Figure 1. Public Sector Debt by Country as a Share of GDP

Gross debt Net debt (gross debt minus international reserves)

Pre-crisis (t–1) Post-crisis (t+1) Pre-crisis (t–1) Post-crisis (t+1)

Table 1. Decomposition of Changes in Debt-to-GDP Ratio, from t–1 to t+1 1/

Individual Contributions to Variations

Change in public Real interest Real GDP Exchange rate

Argentina (2001) 88.5 –1.1 3.2 6.2 63.4 16.7

Average 32.1 –3.2 –7.6 0.6 36.4 5.9

1/ Crisis years in parenthesis.

• With the exception of Argentina and Uruguay—where deep recessions contributed to a

B. Evidence on Debt Reversals

1/ Net debt (gross debt minus international reserves).

C. External Debt versus Domestic Debt

t–2 through t+4 t–2 and t–1 t t+1 through t+4

1/Net debt (gross debt minus international reserves).

III. CHANGES IN THE STRUCTURE OF SOVEREIGN DEBT

A. External Public Debt13

Figure 4. Debt to Multilateral Institutions

Evolution in Percent of Pre-Crisis GDP by Country Change in U.S. dollar terms

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

Composition of Multilateral Debt, Cross-Country Averages

20% 42% 43%

Figure 5. External Sovereign Debt: Creditor Composition, Cross-Country Averages 1/

Exposure as a Share of Total Debt

80% 43% 44% 43% 43%

Multilaterals Bilaterals Private sector

Change in U.S. Dollar Exposure

Multilaterals Bilaterals Private sector

1/ Medium- and long-term sovereign debt.

increase relative to 1/ Between t–1 and t+1.

Box 2. Evolution of Foreign Direct Investment Flows in Crisis Episodes

t+1. This pattern is confirmed by data on 1.5% 30

aggregate FDI flows into the 12 crisis 1.0% 20

fact, Indonesia and Russia were the only -0.5% -10

countries in the sample that experienced -1.0% -20

B. Domestic Public Debt

Domestic Debt Rigidity

Table 2. Domestic Banking Sector Claims on Public Sector 1/ 2/

Sources: IMF, International Financial Statistics and World Economic Outlook.

1/ Includes only countries with banking sector exposure to government exceeding

Maturity and Currency Composition

1/ For Uruguay, data are only available through t+1.

Financial crises were also associated with changes in debt structures.

Data Issues and Methodology

Figure A1. External Sovereign Debt by Country: Creditor Composition

Multilaterals Bilaterals Private Sector

Figure A1 (cont'd). External Sovereign Debt by Country: Creditor Composition

Multilaterals Bilaterals Private Sector

t–2 t–1 t t+1 t+2