Cross-Country Experience with Restructuring of Sovereign Debt and Restoring Debt Sustainability

The paper is organized as follows: Section II provides a cross-country overview of the economic conditions, including the composition of sovereign debt and its dynamics prevailing prior to the debt operations, and a discussion of the scope and outcomes—in terms of the debt relief provided by creditors—of the debt restructurings. Section III analyzes the debt profile, liquidity position, DSAs and accompanying stress tests, and an EWS to assess whether debt sustainability has been restored in each country case. Conclusions are set out in Section IV.

Abstract

The paper is organized as follows: Section II provides a cross-country overview of the economic conditions, including the composition of sovereign debt and its dynamics prevailing prior to the debt operations, and a discussion of the scope and outcomes—in terms of the debt relief provided by creditors—of the debt restructurings. Section III analyzes the debt profile, liquidity position, DSAs and accompanying stress tests, and an EWS to assess whether debt sustainability has been restored in each country case. Conclusions are set out in Section IV.

I. Introduction

1. In the extreme circumstances where a restructuring of sovereign debt becomes unavoidable, an overarching objective is to restore the country’s debt to a sustainable path.

Assessing whether debt sustainability—a situation in which a borrower is expected to be able to continue servicing its debt without an unrealistically large correction to the balance of income and expenditure—has been restored involves an evaluation of the nature of the crisis—including whether it is one of solvency or liquidity—and requires complex judgments, particularly regarding whether a debt restructuring would be sufficient to contribute to a credible and durable exit from the crisis that would enable the country to regain access to international capital markets.2

2. Analysis to date has focused on developing tools to improve debt sustainability assessments and the process for an orderly restructuring of sovereign debt. Less attention has been given to actual experience with debt-restructuring operations, notably on assessing whether debt sustainability was restored. This paper aims to fill this gap, with an emphasis on debt owed to private creditors.3 It reviews the experience of the past several years, where a number of countries—including Argentina (2001 and 2005),4 Dominican Republic (2005), Ecuador (1999), Moldova (2002 and 2004),5 Pakistan (1999), Russia (1998-2000), Ukraine (1998-2000), and Uruguay (2003)—have restructured their sovereign debt in the context of efforts to resolve a crisis. These debt restructurings have either followed a sovereign default or been undertaken pre-emptively in an effort to avoid default.

3. Against this background, the paper examines the initial conditions that gave rise to the debt operations, discusses the impact that the restructurings had in each of these cases, and attempts an assessment of whether sustainability has been restored. The assessment of debt sustainability focuses on three aspects—the current level of debt and related vulnerabilities, as estimated by an early warning system (EWS); vulnerabilities stemming from the liquidity position; and medium-term debt-related vulnerabilities as assessed by debt sustainability analyses (DSAs), including stress testing to determine the effects of shocks to key variables. The assessment is based on data from staff reports that were issued through October 2005, most of which have been published. We recognize at the outset that the sample is too small to allow for firm generalizations.

4. The paper is organized as follows: Section II provides a cross-country overview of the economic conditions, including the composition of sovereign debt and its dynamics prevailing prior to the debt operations, and a discussion of the scope and outcomes—in terms of the debt relief provided by creditors—of the debt restructurings. Section III analyzes the debt profile, liquidity position, DSAs and accompanying stress tests, and an EWS to assess whether debt sustainability has been restored in each country case. Conclusions are set out in Section IV.

II. Cross-Country Experience with Recent Sovereign Debt Restructurings

5. To provide the context, the initial conditions that led to a restructuring of sovereign debt are examined; then the scope and salient features of the debt restructurings are reviewed; and finally the outcome of the debt restructurings and debt dynamics around the time of restructuring are presented.

A. initial Conditions

6. An overview of key debt indicators prior to the restructurings reveals substantial differences in the level and composition of public debt across the eight countries. In particular, two years prior to the restructuring of sovereign debt, total public debt as a share of GDP ranged from 30 percent in the case of Ukraine to 99 percent in the case of Moldova, and the average debt-GDP ratio was 62 percent (see Table 1).6 In countries that eventually restructured pre-emptively to avoid default, debt ratios were between 30 percent and 99 percent, whereas in Ecuador and Russia, which restructured after a sovereign default, debt ratios stood at 66 percent.7 In terms of the composition of debt, several countries had a very large share of foreign-currency denominated debt (Argentina, Ecuador, and Russia had shares of above 95 percent in their respective central years of restructuring, labeled ‘Y’ in Table 1; Uruguay’s ratio was 91 percent), whereas in the Dominican Republic, Moldova, Ukraine, and Pakistan, the shares were somewhat lower (below 85 percent).

Table 1.

Public Debt in Recent Cases of Sovereign Debt Restructuring

(in percent of GDP)

article image
Sources: IMF staff reports; WEO database; and own calculations.

Central year of debt restructuring episode.

As reported in first staff report after restructuring; or inferred (bold Italics).

3/ As reported in last staff report before restructuring; or inferred (bold Italics).4/ Actual or latest estimate.

Including Ukraine, which was in technical default for a short period; Moldova, which incurred arrears towards Gazprom but remained current on its Eurobond obligations; and the Dominican Republic, which was in arrears to its London Club creditors but remained current on its international bonds.

Pakistani fiscal year, July to June; 2004 column refers to 2003/04. Latest estimates are lower than previous estimates partly due to an upward revision in the GDP series.

Combined public sector.

8/ 2005 projections are based on the assumption of full creditor participation in the debt exchange; data cover the federal government only.

7. Notwithstanding considerable divergence in pre-restructuring levels and composition of debt, the majority of countries experienced a surge in debt ratios in the run-up to the restructurings. Total debt grew rapidly in the Dominican Republic, Ecuador, Russia, Ukraine, and Uruguay.8 A combination of factors was responsible, including high and increasing public interest obligations (Ecuador and Uruguay), rapid currency depreciation (Dominican Republic, Ecuador, Russia, Ukraine, and Uruguay), declining economic activity (Ecuador, Russia, Ukraine, and Uruguay), and the fiscal cost of supporting a troubled financial sector (Dominican Republic, Ecuador, and Uruguay). By contrast, Argentina’s and Pakistan’s debt levels did not increase much before the crisis and Moldova’s debt ratio was in decline.9 In the case of Argentina, prior to the abandonment of the convertibility regime, overvaluation of the exchange rate may have contributed to mask debt-related vulnerabilities. Moreover, the subsequent overshooting of the exchange rate may have overstated near-term debt ratios and hence vulnerabilities.

8. High and/or increasing debt levels gave rise to debt-servicing difficulties in all cases, despite differing economic circumstances and backgrounds.

  • Market confidence in Russia declined in 1998, when the oil-producing economy suffered from low petroleum prices, and a weakening ability to implement domestic policies that would address the fiscal imbalances exacerbated investor concerns in the wake of the Asian crisis. This led to a capital account crisis culminating in the devaluation of the ruble, considerable problems rolling over the large stock of treasury bills held by foreign investors, and ultimately to default.10 Gross reserves fell to 15 percent of short-term debt in 1998, as compared to 39 percent in 1996.11

  • Neighboring Ukraine was pulled into crisis partly by the problems in Russia, which worsened Ukraine’s market access at a time when high debt-service payments were falling due. While attracting liquidity was the immediate concern, at a debt level of 42 percent of GDP in 1998, there were few solvency concerns. The authorities adjusted the exchange rate band several times in 1998, and finally floated the currency in March 1999. In the event, liquidity became tight, with gross reserves fallen to 14 percent of short-term debt at end-1998 as compared with 148 percent in 1996. While Ukraine remained current on its debt-service obligations to private creditors,12 it incurred arrears to official bilateral creditors from 2000.

  • In 1998, Ecuador, like Russia, was adversely affected by low oil prices and impaired investor confidence in the emerging market asset class. In addition, the country was hit by a banking crisis. These factors combined led to a devaluation of the sucre in early 1999 and the subsequent default on debt to private creditors, as the public sector debt service burden had increased rapidly (from 8.3 percent of GDP in 1998 to 18.1 percent of GDP in the first quarter of 2000).13 Ecuador had already been in arrears to Paris Club creditors since 1996. In the run up to the crisis, gross reserves declined from 50 percent of short-term debt (1996) to 30 percent in 2000.

  • In the 1990s, Pakistan’s external debt-service payments increased faster than export earnings (debt service as a share of export receipts rose from 26 percent in 1992 to 34 percent in 1998). Pakistan entered a liquidity crisis when following its nuclear tests in 1998 it was subject to international sanctions, affecting both its current and capital accounts. In the event, debt-service payments to official bilateral creditors were suspended in 1998, but Pakistan remained current with respect to its obligations to private creditors. By June 1998, gross reserves dropped to below 10 percent of short-term debt, as compared with 27 percent in June 1996.

  • In Argentina, three years of recession and difficulties in containing the fiscal deficit led to a confidence crisis, leading to soaring sovereign spreads and doubts about the sustainability of the convertibility regime in 2001. At end-2001, the ratio of gross reserves to short-term debt had declined to 31 percent as compared with 48 percent in 1999. Public sector debt service increased from 38 percent of exports in 1998 to 66 percent in 2001. In December 2001, after having undergone two debt operations, Argentina defaulted on government debt owed to private and official bilateral creditors.

  • In Moldova, public debt had built up in the late 1990s due to a combination of sustained fiscal deficits, negative GDP growth, and a sharp depreciation of the leu following the Russian crisis in 1998. The debt ratio improved somewhat after 1998, but debt-servicing difficulties intensified, partly as a result of increasing amortization payments falling due. In the event, public external debt-service obligations increased from 17 percent of exports in 1998 to 23 percent in 2002. Gross reserves as share of short-term debt were 39 percent in 2002. Moldova had been in external arrears on debt-service obligations and energy payments to official bilateral creditors as well as to Gazprom, some of which since 1994, but remained current on its Eurobond obligations.

  • Vulnerabilities in Uruguay had been building up long before the crisis, owing to a long recession the country had been enduring since 1999, persistent fiscal deficits, and an inability to deal with banking system weaknesses. Debt and debt-service problems surfaced after the 2002 banking crisis (triggered by massive withdrawals of Argentine and subsequently domestic foreign currency deposits) and the ensuing change in the exchange rate regime. External debt-service obligations increased from 36 percent of exports in 2000 to 56 percent in 2002. Gross reserves coverage of short-term debt fell from 34 percent in 2000 to 16 percent in 2002. Given the high level of public debt, there were some concerns about solvency. Uruguay did not default on its commercial or official debt service obligations.

  • In the Dominican Republic, the discovery of fraud and losses in the banking system triggered a banking crisis in 2002/03. Private deposits were withdrawn, prompting large official injections of liquidity. Inadequate fiscal management practices undermined the intended fiscal restraint, setting off a vicious cycle of high inflation, peso depreciation, growth of public debt, and capital flight. Gross reserves fell from 151 percent of short-term debt to 31 percent in 2003. Debt service obligations increased from 11.6 percent of goods exports in 2001 to 21.5 percent in 2005 (before restructuring). External arrears started to accrue to the Paris Club (since 2003) and commercial banks (2004). However, the Dominican Republic remained current on its external bond obligations. In the run of the crisis, public debt peaked at 56 percent of GDP in 2003. The debt restructuring has largely addressed liquidity rather than solvency concerns, resolving the bunching of debt service payments falling due.

B. The Scope and Outcome of Sovereign Debt Restructuring with the Private Sector

9. The scope and outcome of sovereign debt restructurings with the private sector varied quite considerably across the eight country cases, with outcomes differing according to whether the restructuring took place pre-emptively or following a default.14 Countries that restructured pre-emptively generally received less debt reduction than those that restructured post-default, but also experienced smaller output declines on average.15

  • The scope of debt restructuring depended on the share of debt owed to private creditors (see Tables 2 and 3). Argentina (2001 and 2005), Ecuador, and Uruguay each restructured approximately half of their public debt. By contrast, in countries where debt was largely owed to official creditors, the scope for debt restructuring with the private sector was more limited (parallel debt operations with the Paris Club are summarized in Table 4).

    Table 2.

    Summary: Scope of the Debt Restructuring

    article image
    Source: IMF staff reports.

    Calculations are based on inclusion of all past due interest.

    Table 3.

    The Scope of Debt Restructuring

    article image

    In the run-up of this exchange offer, Ukraine deferred selected principal payments to respect inter-creditor equity, and, as a consequence, was briefly in technical default during the period of the exchange offer.

    Other elements of Moldova’s strategy are described in Annex II.

    Table 4.

    Recent Paris Club Reschedulings

    article image
    Source: Paris Club.

  • In pre-emptive restructuring cases, debt relief was largely provided by extension of maturity with limited reduction in coupon payments. Debt reduction measured in terms of the decline in the net present value of the restructured debt was relatively small (see Table 5). With the exception of Argentina (2001), the pre-emptive cases received NPV reductions of no more than 8 percent, when evaluated at a common discount rate of 10 percent.16 In the case of the Argentine megaswap, the NPV value increased by 28 percent, while the subsequent Phase I restructuring had a NPV reduction of 32 percent. Jointly, the two restructurings resulted in an NPV reduction of 10 percent. The four post-default cases received NPV reductions of 25 percent (Ecuador), 44 percent (Russia), 58 percent (Moldova), and 75 percent (Argentina’s 2005 global debt exchange).17

Table 5.

Debt Reduction in Net Present Value of Recent Debt Restructurings

(in percent)

article image

Negative numbers indicate an increase in the net present value of debt.

Kozack (2005).

IMF (2001).

Various unpublished IMF studies.

“Uruguay: 2003 Article IV Consultation and Third Review Under the Stand-By Arrangement and Request for Modification .

Waiver of Applicability of Performance Criteria, “ available on the IMF public website at: http://www.imf.org/ external/pub s/ft/scr/2003/cr03247.pdf.

Based on exit yields.

Sturzenegger and Zettelmeyer (2005).

“Russian Federation: Staff Report for the 2000 Article IV Consultation, “ available on the IMF public website at: http://www.imf.org/external/pub s/ft/scr/2000/cr00145.pdf.

“Argentina—StaffReport for the 2005 Article IV Consultation, “ available on the IMF public website at: http://www.imf.org/external/pub s/ft/scr/2005/cr0 5236.pdf.

  • The distinction between pre-emptive and post-default cases becomes more pronounced when looking at debt reduction measured in terms of reduction in principal outstanding (see Tables 6 and 7). Among the pre-emptive cases, no country received more than 6 percent reduction in nominal principal (some countries even saw a small increase), while post-default debt operations yielded substantial reductions.

  • While post-default cases have tended to receive more debt reduction, they also experienced on average deeper economic contractions than the pre-emptive cases.18 On average, real GDP in the pre-emptive restructuring cases contracted by 3.6 percent in the year of lowest growth during the crises (year t in Figure 1). This compares to an average contraction of 7.5 percent in the post-default cases.19 20

    Figure 1.
    Figure 1.

    Evolution of GDP Growth Around the Crisis Episodes

    (in percent)

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    Source: WEO.

  • In cases that could be characterized as exhibiting solvency problems, the amount of debt relief was greater when the restructuring took place following a default. In this context, the decision to restructure pre-emptively may have had some impact on the incentives of the debtors to reach an agreement. In particular, failure to reach an agreement with creditors could subject the debtor to significant reputational, political, and economic costs in the event that default could not be avoided. Under these circumstances, sovereign debtors may acquiesce to debt restructuring terms that satisfy their creditors, but are not sufficient to restore sustainability. The factors that affect the negotiation strategies of creditors and debtors are, however, broad and complex. It is not possible to adequately disentangle the impact of the decision to default from the broader economic circumstances surrounding that decision, including the more severe recessions that were endured in post-default countries.

Table 7.

Results of the Debt Restructurings1

article image
Sources: IMF staff reports; and other documents.

Further details are presented in Annex II.

Evaluating the reduction at a higher discount rate, as done in IMF (2001), leads to a much higher debt reduction.

While, in Uruguay, the debt stock remained high after the restructuring (see Section III), the authorities nonetheless chose not to propose a deeper haircut given that their primary consideration was to improve the debt profile by lengthening maturities, avoid default, and minimize contractual interference, thus aiming at a collaborative process and a voluntary exchange in an effort to support market confidence.

A factor facilitating the high debt relief was the change in the obligor of many original claims (Vneshekonombank).

Principal arrears and PDI recognized through end-2001.

C. Debt Dynamics Around the Time of Restructuring

10. Debt dynamics around the time of restructuring were characterized by increases in debt in the run-up to the crisis, followed by a reduction in the aftermath, but the extent of variability has differed markedly among the cases. Figures 2 and 3 show the debt dynamics around the restructuring period for pre-emptive and post-default cases, respectively, comparing actual debt ratios to pre-restructuring, post-restructuring, and most recent projections. The figures show the build-up of public debt-GDP ratios through the crisis in each case. In most countries (Argentina, Ecuador, Moldova, Pakistan, Russia, Ukraine, and Uruguay), public debt ratios fell considerably following the crisis. The Dominican Republic’s debt ratio has also started to decline following its recent crisis.

Figure 2.
Figure 2.
Figure 2.

Evolution of Public Debt in Pre-Emptive Cases24

Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

Source: IMF staff reports.Source: IMF staff reports.1/ October 2005 data contains a substantial upward revision of historical GDP data, leading to reduced debt/GDP ratios.Source: IMF staff reports.Source: IMF staff reports.Source: IMF staff reports.
Figure 3.
Figure 3.

Evolution of Public Debt in Post-Default Cases

Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

Source: IMF staff reports.Source: IMF staff reports.Source: IMF staff reports.

11. The debt dynamics since the restructuring can be attributed to a number of key factors, including fiscal performance, economic growth, real interest, and exchange rates.21

In cases, where sufficient time has passed since the restructuring episodes, actual debt dynamics in the post-restructuring periods can be compared to staff’s projections elaborated in staff reports following the restructurings.22 Figure 4 shows a comparison of changes in debt to GDP ratios in the respective post-restructuring periods (evolution from the respective central year of restructuring until 2004) for Ecuador, Pakistan, Russia, Ukraine, and Uruguay. In a majority of cases (Pakistan, Russia, Ukraine, and Uruguay), debt dynamics were better than anticipated at the time of restructuring.23

Figure 4.
Figure 4.

Change in Debt/GDP Ratios, Post-Restructuring Period

(in percentage points)

Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

Source: IMF staff reports.

12. A decomposition of debt dynamics sheds light on the reasons for this overperformance. Debt dynamics as projected after the restructurings are decomposed and compared with an ex post decomposition from most recent debt sustainability assessments. Analytically, the decomposition—based on the public sector debt sustainability template for middle-income countries—helps explain the impact on the debt dynamics attributable to changes in the primary balance, the exchange rates, and the interest-growth differential (see Annex III for a detailed analysis).

13. The analysis shows that in their post-restructuring periods, two countries (Russia and Uruguay) outperformed the primary fiscal path projected by staff following the restructurings.25 Three countries had firmer exchange rates than implicit in Fund projections, helping to contain the debt-GDP ratio more quickly than anticipated (Pakistan, Ukraine, and Uruguay). Moreover, in three cases (Pakistan, Russia, and Ukraine) the real interest-growth differential turned out more favorable than expected. Overall, these factors contributed to better debt dynamics than anticipated in four out of the five cases.

III. Sovereign Debt Restructuring and Debt Sustainability

14. Given the different initial conditions, scopes, and outcomes of each of the debt operations, this section attempts to examine the extent to which debt sustainability has been restored. One needs to recognize, however, that assessing the impact of a debt operation per se on debt sustainability is difficult, as these operations took place in the context of changes in the external environment and shifts in economic policies that inter alia underpin the degree of market confidence. Against this background, this section asks whether the debt operations combined with the supporting economic policies were successful in contributing to a return to sustainability.

A. Criteria for Assessing Debt Sustainability

15. As noted above, debt sustainability is defined as a situation in which a borrower is expected to be able to continue servicing its debt without an unrealistically large correction to the balance of income and expenditure. Sustainability thereby encompasses the concepts of solvency and liquidity, without making a sharp distinction between them.26 Which of these two aspects of sustainability is more relevant in making the sustainability assessment depends on the individual country circumstances. From a solvency angle, debt sustainability implies that a debtor must be able to generate sufficient funds in future periods to cover its debt-service obligations without indefinitely accumulating debt.27 In other words, the sovereign must be able to produce a level of primary surpluses that over the medium term would maintain or lower the ratio of debt to GDP. From a liquidity angle, debt sustainability implies that the debtor must be able to find sufficient amounts of financing in each period to close any financing gaps without having to resort to disorderly adjustment.

16. Assessing debt sustainability is highly sensitive to the assumptions concerning the projections of growth, inflation, interest, and exchange rates. Moreover, difficulties in estimating contingent liabilities arising, for example, from the resolution of financial sector problems, add to the challenge. While some problems can be addressed by sensitivity analysis, including in the context of standardized alternative scenarios and stress tests, the results are necessarily subject to judgments28 In addition, it needs to be recognized that the analysis of debt sustainability focuses on countries’ underlying economic vulnerabilities, but does not attempt to comprehensively evaluate near-term crisis risks.

17. Recognizing that elements of judgment and uncertainty are unavoidable, the stock of debt following a restructuring can in principle be judged as being sustainable if the probability of a recurrence of debt crisis is judged as fairly low. At the other extreme, the debt level would be judged as unsustainable if the probability of a recurrent crisis was rather high. Between these extremes, there could be a “gray zone,” a range of debt levels that would be consistent with a medium range probability of a recurrence of debt problems.

18. To help to classify the cases into the three categories, we apply three sets of criteria, which are discussed further in the following subsections:

  • The current debt level and vulnerabilities as estimated by an EWS. The current level of debt measures the accumulated debt burden that needs to be serviced (see Box 1). With the help of EWS estimates, it is possible to gauge debt-related vulnerabilities based on the historical experience in a wide sample of countries. It has to be recognized, however, that EWS estimates, while useful as part of a broader analysis, should not be relied on as an exclusive metric for assessing the likelihood of debt distress (see Box 2). These estimates can shed some light on near-term vulnerabilities, but not on persistent ones that could materialize in the longer term.

    Debt Crises and the Level of Debt

    In order to analyze the relationship between debt levels and the probability of debt crises, we estimate probabilities for debt crises, using lagged values of the debt-to-GDP ratio, GDP growth and the ratio of short-term external debt to GDP as explanatory variables (see Annex IV). The sample consists of 55 low- and middle-income countries, during the period 1971-2002. The pooled probit model allows for an in-sample forecast of crisis probabilities (graphically shown as the dots in Figure 1.1). We fit a polynomial through the in-sample forecasts (thick line in Figure 1.1) and show a +/- two standard-deviation band around it. From this sample, a 50 percent probability of being in debt crisis is associated with a debt-to-GDP ratio of 80 percent in the previous year. However, the standard deviation is rather substantial (18 percentage points). There is also no obvious cut-off point for the debt ratio that would allow deciding between sustainable and unsustainable debt levels, although a clear correlation is evident.

    Figure 1.1.
    Figure 1.1.

    Crisis Probabilities and Debt Levels

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    Sources: WEO; and IMF staff calculations.

    From the same dataset, we isolate the episodes of entry into debt crisis and order them by their observed debt-to-GDP ratios in the last year before entry (see Figure 1.2). Again, there do not appear to be obvious cut-off points. However, 40 (60) percent of entries into crisis in the sample occurred when debt levels in the last year before the crisis had been above 59 (39) percent of GDP. The average debt-GDP ratio in a year before crisis entry was 57 percent (compared to 47 percent in tranquil times).

    Figure 1.2.
    Figure 1.2.

    Debt Levels in Year Before Crisis Entry

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    In addition to the debt-to-GDP ratios, other factors regarding the current indebtedness play a role in the assessment of sustainability. For countries holding substantial amounts of concessional debt, a NPV-based ratio arguably reflects more accurately the incidence of debt. Moreover, the composition of debt matters. A higher share of domestic debt tends to indicate a lower vulnerability to real exchange rate movements. Similarly, a high share of official loans in the debt portfolio is associated with lower vulnerability, as this creditor group may be more likely than others to roll over maturing claims. However, as these variables are not available for a large sample of countries and over a longer period of time, the effects on sustainability cannot easily be assessed quantitatively.

    An Early-Warning System Approach

    Building on Manasse, Roubini, and Schimmelpfennig (2003), Schimmelpfennig (2003) specifies an EWS model based exclusively on data available in the WEO database. The estimation method is a binary regression (logit) with robust variance estimates for pooled data. The sample comprises of 37 low- and middle-income countries over the period 1990-2002. A country is defined to be in a debt distress if it is classified to be in default by Standard & Poor’s or if it receives nonconcessional financing from the IMF in excess of 100 percent of quota. Countries can be in crisis for protracted periods.

    Independent variables in the model are total external debt in percent of GDP, short-term external debt by remaining maturity as a share of official reserves, the current account balance in percent of GDP, growth of foreign exchange reserves, GDP growth, inflation volatility, as well as a dummy for hyperinflation episodes (by contrast, public debt as share of GDP is not included).1 All coefficients show the expected sign and are statistically significant.

    The EWS model fits the data well, correctly predicting 62 percent of in-sample crisis entries while sending false alarms of looming distress (Type II errors) in 12 percent of the cases. The EWS model signals heightened debt vulnerabilities when the predicted vulnerability score is greater than the in-sample frequency of crisis episodes (24.8 percent).

    The model results should be interpreted with care. As noted in IMF (2003c), “identifying ‘danger zones’ is still more an art than a science, with a large element of judgment required.” In a recent survey of a variety of EWS models developed by Fund staff and by private institutions, Berg, Borensztein, and Pattillo (2005) explained that “the advantage of EWS models lies in their objective, systematic nature. The models process data in a mechanical way and are not clouded by conventional misperceptions or biases based on past experiences.” The survey found that EWS model forecasts can be statistically significant predictors of distress, with some models outperforming bond spreads, agency ratings, and analysts’ risk scores. This said, the authors also concluded that the results of their survey “are mixed ... and reinforce the view that EWS models are not accurate enough to be used as the sole method to anticipate crises. However, they can contribute to the analysis of vulnerability in conjunction with more traditional surveillance methods and other indicators.”

    1 Since the ratio of public debt to GDP does not enter as an independent variable, it is not possible to use this specification to relate public debt levels with crisis probabilities.

  • Vulnerabilities stemming from the liquidity position. From a liquidity perspective, a sovereign that is faced with large financing needs (including for rollovers of maturing obligations) and/or a small pool of available resources to satisfy its obligations stands an increased risk of incurring a debt service crisis.

  • Medium-term debt-related vulnerabilities. Medium-term debt dynamics and related vulnerabilities can be gauged based on debt sustainability analysis. This analysis calculates medium-term debt projections based on estimates for key macroeconomic variables. Stress tests and alternative scenarios bring to light related vulnerabilities.29

B. The Current Debt Level and Related Vulnerabilities

19. There are no clear-cut threshold levels regarding debt-to-GDP ratios that would allow classifying countries into groups of low, medium-range or high probability of crisis recurrence. However, cross-country experience can shed some light on the relationship between debt levels and sustainability. From a sample of 55 low- and middle-income countries over the period 1971-2002, a debt-to-GDP ratio equal to 80 percent is estimated to be associated with a 50 percent probability of a being in debt crisis the following year (see Box 1). Similarly, a debt level of 100 percent is associated with a probability of 63 percent, while a debt level of 40 percent corresponds to a rather low probability of 20 percent in the sample. Moreover, 40 percent of entries into crisis occurred when debt levels had been above 59 percent of GDP in the last year before crisis, while 60 percent of crisis episodes occurred at debt levels above 39 percent of GDP.30 However, there do not appear to be obvious cut-off points for the range of sustainable debt, as crises have occurred at a very large range of debt ratios.

20. In 2004, two of the eight country cases had debt ratios in excess of 90 percent (see Table 8). However, Uruguay’s debt ratio (93 percent) has since declined (to 69 percent in 2005, somewhat faster than projected). Argentina’s debt ratio (129 percent at end-2004) has been reduced substantially (to 87 percent at end-2005) following the global debt exchange. The early repurchase of all its outstanding Fund obligations (SDR6.7 billion) in January 2006 has also helped to improve the debt structure. Debt ratios in four other countries ranged between 40 percent and 70 percent (Dominican Republic, Ecuador, Moldova, and Pakistan), respectively.31 The remaining two countries (Ukraine and Russia) had ratios at or below 30 percent.

Table 8.

Public Debt Ratios, 2004 1/

(in percent of GDP)

article image
Source: IMF staff reports.

2004 is the most recent year with actual data for all countries.

End of FY 2003/04.

21. EWS can be used to supplement the analysis of debt-related vulnerabilities.

Table 9 shows estimated vulnerability scores based on a model applied by Manasse, Roubini, and Schimmelpfennig (2003) and Schimmelpfennig (2003) (see Box 2 for a description of the methodology).32 The model correctly predicts 62 percent of in-sample crisis entries but also sends false alarms of looming distress in 12 percent of the cases (Type II errors), underscoring that results should be interpreted carefully, and in conjunction with a menu of other indicators. The results would assign significant vulnerability scores to Argentina and Uruguay. However, Argentina’s score drops considerably in 2006, as the post-exchange debt ratio enters into the projection.33 Vulnerability scores for the remaining countries are much lower.

Table 9.

EWS Vulnerability Scores _(in percent)_

article image
Source: IMF staff calculations.1/ 2005 and 2006 columns refer to FY 2004/05 and 2005/06 respectively.

C. Vulnerabilities Stemming From Liquidity Needs

22. Liquidity risks can stem from a country’s external situation as well as from its fiscal position. Projections for the ratio of foreign exchange reserves to short-term external debt can be used to show potential balance of payments vulnerabilities that could arise should market access deteriorate. A ratio below one implies that complete lack of market access for one year would put pressures on the balance of payments. From a fiscal perspective, gross financing needs projections show the amount of resources a sovereign needs to secure in addition to projected revenues, in order to be able to execute its planned expenditures.

23. Within the country sample, the degrees of vulnerability are widely dispersed (see Table 10). Ecuador, Uruguay, and to a smaller extent Argentina, show substantial degrees of vulnerability in their external liquidity positions, while risks are somewhat smaller in the Dominican Republic and Moldova. By contrast, Pakistan, Russia, and Ukraine maintain rather comfortable levels of reserves. Regarding the magnitude of financing needs, risks are highest among the group in Uruguay, although with significant bond issuance in early 2006, Uruguay’s near-term risks have decreased. Financing needs are in a medium range in the Dominican Republic, Ecuador, Moldova, Pakistan, Ukraine, as well as in Argentina, although in the latter case they have been reduced considerably with the full prepayment of Fund resources in early 2006. Financing needs are relatively low in Russia.

Table 10.

Liquidity Indicators, 2005-06

article image
Sources: WEO; IMF staff reports; and staff calculations.

Ratio of gross international reserves to short-term external debt at remaining maturity.

Defined as the overall fiscal deficit, plus amorization of medium- and long-term debt, plus short-term debt at end of previous period.

2005 and 2006 columns refer to FY 2004/05 and 2005/06, respectivel

Financing need projections exclude debt buybacks.

Gross fiscal financing need is defined as federal overall cash balance plus federal amortization due.

D. Medium-Term Debt-Related Vulnerabilities

24. Debt-related vulnerabilities can be analyzed based on projections of the ratio of debt to GDP over the medium term and associated stress tests. Under the Fund’s standard debt sustainability analysis, a central projection, based on a medium-term macroeconomic framework, is supplemented by a series of alternative scenarios and bound tests, showing the dispersion of debt paths under alternative sets of assumptions (see Box 3 for a description of the

The Fund’s Methodology for Debt Sustainability Analysis in Middle-Income Countries

The Fund has started to apply a standard methodology to debt sustainability analyses in middle-income countries. Based on medium-term projections for a country’s macroeconomic framework, a baseline for debt dynamics is projected. In addition, a set of standardized alternative scenarios and stress tests are calculated, in an attempt to provide sensitivity analysis which is broadly comparable across countries.

The Fund’s standard sensitivity tests comprise of the following: an alternative scenario uses historical averages of key variables (real GDP growth, real interest rate, primary balance), normally over a period of 10 years. The stress tests apply a two-standard deviation shock to each of these variables in turn, leaving the remaining variables as under the baseline scenario.1

Additional stress tests include a combined shock to all three variables of one standard deviation; a one-time 30 percent depreciation of the real exchange rate; and a 10 percent of GDP increase in debt on account of events such as realization of contingent liabilities from a financial sector restructuring. Other alternative scenarios have more recently been added, including one using a primary balance forecast assuming unchanged policies; a scenario based on forecasts for key variables obtained from the financial markets; and a country-specific shock.

Regardless of the standardization, the debt sustainability analysis still requires a considerable degree of judgment. The central projections and alternative scenarios are necessarily functions of projections for key variables, including GDP growth, real interest rates, exchange rates, and the primary fiscal balance, which cannot be made without involving a significant measure of judgment.

It should also be underscored that stress tests are normally based on 10-year historical averages and standard deviations, implying rather harsh shocks for those countries emerging from transition or having undergone large crises during that time. Against that background, judgment is necessary in evaluating the comparability of the stress tests across countries. Moreover, given the relatively small number of stress tests and their degree of dispersion (see Table 9), the empirical value of such stress tests in determining whether a case is sustainable is limited.

1 Based on the framework adapted in 2003 (See IMF (2003b)). In July 2005, certain changes were made to the framework, including the application of less severe shocks in the bound tests (see IMF (2005)).

25. In all eight country cases, medium-term baseline projections tend to be somewhat below 2004 levels (see Table 11).

Table 11.

Debt Sustainability Analysis Projections (percent of GDP)

article image
Sources: IMF staff reports; and staff calculations.

For 2010, except Ukraine (2009) and Pakistan (2009/10). Data is based on 2003 DSA framework, except for Pakistan, Russia, and Uruguay, which use the 2005 framework (see Box 3).

Mean and standard deviation (STD) of scenarios and bound tests.

Public debt in 2004 refers to 2003/04.

Moldova’s DSA is expressed in terms of net present value.

However, the mean of the projected debt levels of different scenarios and stress tests for each country at the end of the projection horizon is lower than the current level only in seven cases. By contrast, for Ecuador, the mean of the projections is considerably above the 2004 level. The dispersion of projected debt levels is also substantial (see last column of Table 11).

26. A closer look at the cases sheds further light on medium-term debt-related vulnerabilities:

  • Ukraine. Under the baseline scenario in the most recent debt sustainability analysis, debt is projected to decline from 27 percent of GDP in 2004 to 16 percent by 2009.34 The alternative scenarios and bound tests show that some vulnerabilities remain, but only one scenario (combining a shock to GDP growth with a relaxation of the primary balance) would imply a steadily increasing debt ratio, while all others see the debt ratio temporarily increasing to at most 39 percent (see Figure 5).

    Figure 5.
    Figure 5.

    Ukraine: Debt Sustainability Analysis, 2003–09 (percent of GDP)

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    Source: “ Ukraine-Staff Report for the 2004 Article IVConsultation . “

  • Pakistan. Figure 6 shows that under the baseline scenario, debt is projected to decline from 68 percent in 2003/04 (equivalent to 59 percent in terms of net present value) to 46 percent.35 36 This projection is based on the assumption that on average, primary deficits of 0.4 percent of GDP will be maintained over the projection horizon (2009/10), substantially below the average primary surplus achieved over the preceding four years (1.7 percent). The stress tests show that the applied shocks would not lead to explosive debt paths, with even the most adverse scenario (a one-time 30 percent real depreciation in 2005/06) leading, after a temporary increase to 69 percent, to a steady decline in the ratio to 60 percent by 2009/10.

    Figure 6.
    Figure 6.

    Pakistan: Debt Sustainability Analysis, 2001/02-09/10

    (in percent of GDP)

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    Source: “Pakistan–Staff Report for the 2005 Article IV Consultation .”

  • Moldova. Under the baseline scenario, debt is projected to decline from 54 percent in 2004 to 26 percent by 20 1 0.37 The projection is based on assumed primary deficits in the order of 1/2 percent of GDP, implying a loosening compared to a 2001-04 average surplus of 2.2 percent (see Figure 7).38 The alternative scenarios and bound tests point to limited risks, with 2010 projected debt substantially below 2004 levels in all simulations. Only one bound test, a one-time 30 percent real depreciation in 2005, would lead to a temporary increase in the debt ratio to above 60 percent.

    Figure 7.
    Figure 7.

    Moldova: Debt Sustainability Analysis, 2003-10

    (in percent of GDP)

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    Source: “Republic of Moldova—Staff Report for the 2004 Article IV Consultation .”

  • Uruguay. The baseline scenario projects a steady decrease in the debt ratio from 93 percent in 2004 to 54 percent by 2010, contingent on sustained primary surpluses of 3.5-4 percent of GDP. While this would not imply substantial tightening relative to the 2004 outcome (3.8 percent), it would nonetheless require sustained fiscal discipline, against a background of substantially lower surpluses in recent years (0.1 percent during 2000-03).39 The scenario also does not factor in the projected fiscal contingent liabilities associated with restructuring of the public banks, which could add up to 5 percentage points to the debt ratio by 2010.40 The stress tests confirm that significant vulnerabilities remain. A one time 30 percent real depreciation would raise the debt ratio to 104 percent by end-2006, from where it is projected to recover slowly, while a scenario based on 10-year historical averages would set the debt ratio on a steadily increasing path (see Figure 8).41 However, vulnerabilities are mitigated by the reduced rollover risk brought by the lengthening of maturities in connection with the debt restructuring.42

    Figure 8.
    Figure 8.

    Uruguay: Debt Sustainability Analysis, 2003-10

    (in percent of GDP)

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

  • Dominican Republic. Under the baseline scenario, public debt is projected to fall from 54 percent in 2004 to 38 percent in 2010 (see Figure 9). The scenario is conditional on achievement of primary surpluses of between 0.8 percent and 2.5 percent of GDP, implying a substantial fiscal consolidation from recent years: during 2001-04, the Dominican Republic posted an average primary deficit of 1.8 percent of GDP. The sensitivity tests show that some vulnerabilities remain: two scenarios, featuring a one-time real depreciation of 30 percent and a 10 percent increase in other debt-creating flows would push the debt ratio again temporarily to above 55 percent. However, most other scenarios project roughly steady or slightly falling debt ratios.

    Figure 9.
    Figure 9.

    Dominican Republic: Debt Sustainability Analysis, 2002-09

    (in percent of GDP)

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    Source: IMF staff reports.

  • Ecuador. Figure 10 shows that under a baseline scenario the debt ratio would decline from 53 percent in 2003 to 42 percent by 2010. The scenario is based on the continuation of sizeable primary surpluses (2-5 percent range), which would be consistent with recent experience (from 2000-03, the primary surplus averaged 5.3 percent), although recent political developments have increased uncertainty to the outlook.43 The sensitivity tests show that vulnerabilities are pronounced. Under a number of the standardized stress tests, debt would stabilize only at a substantially higher ratio of around 70 percent. Moreover, a shock to GDP growth would lead to a sharply rising debt path.

    Figure 10.
    Figure 10.

    Ecuador: Debt Sustainability Analysis, 2002-10

    (in percent of GDP)

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    Source: IMF staff reports.

  • Russia. Russia’s debt as a share of GDP is projected to fall from 22 percent in 2004 to 7 percent in 2010 under a baseline scenario, conditional upon primary surpluses in the 5-9 percent range, facilitated by favorable petroleum prices (during 2001-04, the primary surplus averaged 4.3 percent).44 Limited vulnerabilities remain, as evidenced by the stress tests (see Figure 11). A marked fall in oil prices as assumed under one of the stress tests would lead to steadily increasing debt ratios, reaching 34 percent by 2010. Two other applied shocks would each lead to a debt to GDP ratio of above 40 percent by end-2006, from where it would start falling gradually. However, as the stress tests are based on 10-year historical averages and standard deviations, they are based on a historical period which comprises the difficult transition years. This results in unusually harsh shocks: for instance, a shock to GDP growth using the historical average less two standard deviations implies considering an annual decline in real GDP of 8.1 percent for two consecutive years (compared to, for example, a fall of 4.2 percent in Ecuador). Despite such harsh shocks, Russia’s debt would nevertheless only exceed 30 percent in 2010 under two scenarios.

    Figure 11.
    Figure 11.

    Russia: Debt Sustainability Analysis, 2003-10

    (in percent of GDP)

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    Source: “Russian Federation—Staff Report for the 2005 Article IVConsultation. “

  • Argentina. The baseline scenario projects a decline in the debt-to-GDP ratio from 78 percent in 2005 to 57 percent by 2010, and does not yet factor in the favorable effects of the early repurchase of all outstanding Fund obligations in January 2006 on debt service obligations.45 46 While Argentina still needs to resolve its arrears to nonparticipating creditors, projections are based on staffs reform scenario, assuming full participation in the 2005 debt exchange, and anchored on federal primary surpluses between 3.3 percent and 3.7 percent in 2005-10.47 These surpluses are consistent with the 2004 outcome (3.9 percent), but historically Argentina’s primary surpluses have been lower (1.6 percent average in 2002-03), pointing to the sustained policy effort that will be necessary to achieve the assumed primary surpluses over the medium term. The stress tests point to remaining risks to debt sustainability (see Figure 12). However, it should be recognized that the standardized calibration method of the DSA stress tests, which is based on 10-year historical averages and standard deviations (see Box 3), may lead to the simulation of very large shocks in the case of Argentina. Given the high volatility of growth during that period and the very severe contraction of output experienced in 2001/02, the calibration, for instance, implies a simulation of a decline in growth by 12.3 percent in 2006 and 2007, compared to robust growth projections in the baseline scenario. This large GDP shock as well as a pronounced joint shock on the interest rate, the GDP and the primary balance would imply an increase in debt to more than 100 percent by 2007.48 A scenario based on historical averages as well as the GDP shock scenario would imply continuously rising debt ratios through 2010.49

    Figure 12.
    Figure 12.

    Argentina: Debt Sustainability Analysis, 2003-10

    (in percent of GDP)

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    Source: “Argentina-Staff Report for the 2005 Article IV Consultation.”

E. Summary for the Eight Countries

27. Based on the above analysis, medium-term vulnerabilities appear low in Pakistan, Russia, and Ukraine, and in the medium range in Argentina, Dominican Republic, Ecuador, and Moldova. A considerable degree of vulnerability remains in Uruguay (see Table 12).

Table 12.

Debt Sustainability Indicators

article image
Sources: IMF staff reports; and staff calculations.

For 2010, except Ukraine (2009) and Pakistan (2009/10).

Mean and standard deviation (STD) of scenarios and bound tests in final year.

Net present value of debt-to-GDP ratio in 2003/04: 59 percent.

Debt levels and DSA projections are expressed in terms of net present value.

Post restructuring debt level is 78.2 percent (end-2005 projection based on assumption of full creditor participation in the debt exchange).

  • In Russia and Ukraine, relatively low debt levels and EWS vulnerability scores, comfortable reserve coverage, manageable financing needs, and favorable DSA projections point to overall low solvency and liquidity crisis vulnerabilities. Pakistan’s crisis vulnerability can also be characterized as relatively low: while its debt level is substantially higher than Russia’s or Ukraine’s, its composition of debt—of which only a small fraction is owed to commercial creditors—makes it less vulnerable to swings in market confidence, and the relatively low risk implied in the DSA projections, low EWS vulnerability scores, and relatively favorable liquidity indicators suggest that overall, Pakistan’s debt vulnerabilities are limited.

  • Vulnerabilities in Argentina, the Dominican Republic, Ecuador, and Moldova can be characterized as in the medium range. In this group, debt levels, EWS vulnerability scores, and financing needs are generally higher than in the low vulnerability group, reserve coverage is lower, and DSA projections tend to indicate higher risks. In Argentina, nonparticipating creditor claims are still in arrears.

  • In Uruguay, while short-term risks have already been significantly lessened, considerable vulnerabilities remain. While on a declining trajectory, the debt level is still high, and liquidity indicators point to remaining challenges. However, near-term risks have been reduced as a result of generally favorable external financing conditions and of the Fund arrangement remaining on track.50

28. Developments in sovereign debt markets and credit ratings lend some support to the above analysis.

Liquidity conditions in international bond markets have been very favorable in recent years—the overall EMBI Global spread fell from 1000 basis points (bps) in late 2001 to below 240 bps in December 2005—making it somewhat difficult to extract inferences from country spread movements. However, spreads on sovereign debt issuance by Russia, Ukraine, and Pakistan indicate that market confidence has returned following the resolution of their respective crises, with spreads having fallen below the EMBI Global in these countries (see Figure 13). Spreads in the Dominican Republic and Uruguay have also fallen considerably since the height of the crises, while improvements in sovereign spreads have been less systematic in Ecuador. The spread on Argentina’s sovereign debt remained extraordinarily high until mid-2005, but fell to below 500 bp shortly after the Global Debt Exchange, and has declined further since then.51

Figure 13.
Figure 13.

Sovereign Spreads of Restructured Bonds, 2000-05

(Spreads over comparable U.S. Treasuries, in percentage points)

Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

Source: Bloomberg.
  • Credit ratings have improved considerably in Argentina, Moldova, Pakistan, Russia, and Ukraine, while improvements in Ecuador were less pronounced.52 Moody’s has assigned a B3 rating to Argentina’s foreign-currency long-term debt and applied this rating to the newly issued bonds. Moody’s ratings for the Dominican Republic and Uruguay have not yet improved since the respective debt restructurings (see Figure 14). Standard and Poor’s upgraded Uruguay in 2004 to B, five notches below investment grade, and raised the outlook from stable to positive in early 2006. Moreover, Standard and Poor’s downgraded Ecuador in 2005 to CCC+, and assigned a B rating to the Dominican Republic after the recent bond exchange.

    Figure 14.
    Figure 14.

    Moody’s Sovereign Credit Ratings, 1998-2005

    (Ratings for long-term foreign currency borrowing)

    Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

    Source: Bloomberg.

IV. Conclusions

29. The purpose of this paper was to review recent experience with the restructuring of sovereign debt owed to private creditors. The focus has been to try to determine whether the objective of restoring sustainability has been achieved. Three broad criteria have been applied in evaluating whether countries have succeeded in this objective: (i) the debt profile and debt vulnerability scores as estimated by an EWS model; (ii) vulnerabilities stemming from the liquidity position, as indicated by the financing needs and the level of international reserves; and (iii) medium-term debt-related vulnerabilities as presented in the sensitivity analyses of the respective DSAs.

30. Countries that restructured pre-emptively have had diverse experiences since restructuring, with specific factors playing a prominent role.

  • Liquidity more than solvency was the issue in Ukraine and the Dominican Republic. Consequently, the focus was on debt service relief rather than debt reduction. In the case of Ukraine (classified as low vulnerability, implying that the country is currently regarded as having a low medium-term risk of a recurrent debt crisis), this appears to have been sufficient to ensure a durable exit from crisis, while in the Dominican Republic (medium vulnerability) risks regarding liquidity and debt dynamics remain.

  • The scope of restructuring of private sector-held claims was too limited to have a significant impact on debt sustainability in Pakistan and Moldova. However, reschedulings of debt owed to official creditors yielded significant debt relief for Pakistan, and moderate debt-service relief for Moldova.53 In Pakistan (low vulnerability), this was sufficient to significantly reduce the probability of a future crisis, while vulnerabilities persist in Moldova (medium vulnerability).

  • In the remaining cases, there was evidence pointing to solvency problems, yet sufficient debt relief could not be secured in Argentina (2001) and Uruguay. Argentina defaulted shortly thereafter. In Uruguay, considerable vulnerabilities remain in the aftermath of the crisis: while on a declining trend, debt remains high and the liquidity position relatively tight. However, near-term risks have been significantly reduced as a result of generally favorable external financing conditions, the build-up in international reserves, and the Fund arrangement remaining on track.

31. None of the post-default restructuring cases is currently ranked as highly vulnerable.

  • In Russia, favorable external conditions played a key role, and the country is now classified as having a low vulnerability.

  • Ecuador has been classified as a medium-vulnerability country as setbacks in domestic policy implementation have had a substantial impact on the debt dynamics, in spite of high oil prices.

  • In Argentina, vulnerabilities also appear to be in the medium range, given the large stock of debt in arrears remaining after the closing of the 2005 debt exchange. Dealing with creditors that did not participate in the exchange is key to further reduce the debt burden and facilitate that access to international capital markets can be sustained in times of less favorable global liquidity conditions.

32. The small sample of restructuring cases implies that broad conclusions are highly tentative. Nevertheless, the recent experience suggests some differences between cases of pre-emptive and post-default restructuring:

  • In all three cases where sovereign bonds were restructured after a default, there was clear evidence of solvency problems, and debt restructuring led to a sizeable reduction in principal payments and in the NPV of debt. However, the evidence in this paper has not disentangled the impact of the decision to default on incentives in negotiations from the broader economic circumstances surrounding that decision, including the more severe recessions that were endured in post-default countries.

  • In contrast, the NPV reduction in the six countries that restructured pre-emptively was considerably smaller. The debt operations appear to have been geared to address liquidity rather than solvency concerns, even in the two cases where the evidence pointed to solvency problems. While the sample is very small, in these two cases, either the debt restructuring left significant debt vulnerabilities (Uruguay), or a crisis indeed reoccurred (Argentina).

Annex I: The Role of the Fund

1. The examined restructurings generally occurred while Fund arrangements were in place with the countries, but frequently the Fund-supported programs were off-track. All eight countries had a Fund arrangement when they restructured their public debt to external private creditors. However, Fund-supported programs with Pakistan, Ukraine, Russia, and Moldova were off-track during at least part of these restructuring episodes. Ecuador did not have a Fund-supported program prior to the restructuring of its domestic debt, but a program was agreed prior to the restructuring of external public debt. Argentina’s program was on-track during the June 2001 megaswap, but went off-track soon afterwards, before the Phase I restructuring of November/December 2001. Similarly, during the 2005 global debt exchange, Argentina had a Fund-supported program which was off-track. Among the eight cases, only Uruguay and the Dominican Republic had Fund-supported programs that were on-track during the entire restructuring process (see Table I.1).

Table I.1.

Fund Arrangements During Recent Sovereign Debt Restructuring Episodes

article image
article image

2. While the degree of detail in the fiscal and external sector projections varied, in nearly all cases the policy framework preceding the restructuring included medium-term assumptions for macroeconomic policies and the balance of payments related to the “envelope” of resources that could support an eventual agreement between sovereign debtors and their private creditors. In pre-emptive cases avoiding default, the Fund-supported programs that were in place before the restructuring were generally more specific in defining the medium-term macroeconomic framework than in post-default cases.

  • For Ukraine’s August 1998 EFF, the MEFP sketched a reserve target and the intended direction for the fiscal deficit (rather than the primary deficit) until 2001. The staff report supported this with detailed BoP assumptions through 2004 and fiscal projections through 1999, amended by an overall fiscal deficit path through 2001. Throughout the restructuring period, letters of intent were published, but not the staff reports. During the exchange offer of Eurobonds and Gazprom bonds, the Fund-supported program was off-track, and Fund management sent a letter in February 2000, describing Ukraine’s macroeconomic performance and judging that the terms of the exchange offer would be consistent with financing needs of a program that could be supported by the Fund.

  • Pakistan’s December 1998 policy framework paper specified the primary balance path and reserve targets through 2002/03, while details on external financing projections were given through 2000/01. The policy framework paper was published, but the subsequent May 1999 staff report and letter of intent were not. The Fund-supported program went off-track in mid-1999, and Fund management sent a letter in November 1999, describing the macroeconomic program underlying the bond exchange and asking for support by the international community.

  • In the case of Argentina, for the May 2001 SBA review, the MEFP specified a fiscal path by referencing the Fiscal Responsibility Law, which required the federal budget to be balanced by 2003 but did not set out specific primary balance targets. The staff report presented the full fiscal table through 2001 as well as a fiscal below-the-line presentation and full balance of payments table through 2006. The May 2001 staff report was published, but for the August 2001 review, only the MEFP was made public. In September, the Fund approved an augmentation of Argentina’s Stand-By Arrangement by approximately US$8 billion to about US$22 billion. US$3 billion of this augmentation could be used to support a voluntary and market-base operation by Argentina to increase the viability of its debt profile. The Fund-supported program went off-track before the Phase I restructuring. The Fund approved and fully disbursed an eight-month SBA in January-August 2003 (SDR 2.2 billion or 103 percent of quota), following which a three-year SBA (SDR 9 billion or 424 percent of quota) was approved in September 2003. Subsequent to the second review, which was completed in March 2004, no further reviews were completed under that program. The staff report for the second review was published, and contained fiscal and balance of payments projections through 2004, but not for the medium term. Fund management did not send letters supporting the Argentine debt operations.

  • For Moldova’s July 2002 PRGF review, the MEFP specified primary balance targets for 2002 and 2003. The staff report, which was published, presented the full fiscal table through 2003 and balance of payments projections through 2010. Fund management sent a letter in April 2002, describing performance under the PRGF, which was off-track at the time, but stating that financial support, including from the private sector, would be essential for the future success of Moldova’s economic program. Moldova did not have a Fund-supported program during its 2004 Gazprom debt regularization.

  • In the case of Uruguay, the March 2002 two-year SBA contained commitments for the primary balance through 2003, supported by full fiscal and external sector projections through 2003. For the March 2003 review, the MEFP contained primary balance projections until 2005, and the staff report (published) showed full fiscal and external sector projections through 2004. Fund management sent a letter in March 2003, describing the macroeconomic program, including the primary surplus target for 2003 and the medium term, and encouraging a high participation in the exchange offer.

  • For the Dominican Republic’s 2005 debt restructuring, the MEFP contained primary balance and external reserves projections through 2006. The accompanying staff report contained full fiscal and balance of payments projections through 2007 and 2009, respectively. The MEFP was published, but the staff report was not.

  • For Russia’s 1999 17-month SBA, the MEFP did not specify a primary balance target beyond 1999, and similarly, detailed fiscal and external sector projections were only given until 1999 in the staff report. However, an overview table containing a medium-term framework until 2005 was presented. The staff report was not published, but a press release showed the primary balance target and an overview of BoP projections for 1999. At the time of the restructuring of PRINs and IANs, Russia’s SBA was off-track, and Fund management sent a letter in July 2000, describing economic developments, indicating that discussions were under way that could lead to a new SBA, and judging the proposed debt exchange as warranting the support of the international community.

  • Ecuador’s 12-month SBA was approved in April 2000, thus after the rescheduling of domestic debt but before the exchange offer on external debt of July/August 2000. The MEFP (published) did not specify quantified fiscal targets beyond 2000. The staff report (not published) provided detailed fiscal and external sector tables only through 2000 but a debt sustainability analysis laid out medium-term fiscal and BoP projections. For the restructuring of external debt, Fund management sent a letter in July 2000, describing the macroeconomic program (including the fiscal deficit target for 2000), and urging the private sector to support the exchange.

Annex II. Outcomes of Recent Restructurings

Ukraine restructured treasury bill debt held by nonresidents equivalent to approximately US$300 million into a two-year zero-coupon Eurobond with an annual yield of 20 percent in September and October 1998. Subsequently, it renegotiated its debt owed to Chase Manhattan clients (US$109 million) in October 1998. The deal involved up-front cash payments of 25 percent and converted the remainder into a two-year loan carrying 16.75 percent interest and involving graduated principal payments. This was followed in August 1999 by the renegotiation of debt held by ING and Merrill Lynch, together covering US$405 million. The deal involved the negotiated rollover into other instruments, including the augmentation of existing DM Eurobonds. After this piecemeal approach proved insufficient, Ukraine launched a more comprehensive restructuring in February 2000, involving Eurobonds and Gazprom bonds with a face value of US$3.3 billion (including the recognition of past due interest) and coupons ranging from 8.5 percent to 16.75 percent. Investors were given a choice of converting their claims into either a EUR-denominated Eurobond with 10 percent coupon or a US$-denominated Eurobond with 11 percent coupon. Both bonds carried a seven-year maturity and one-year grace period. The operation was sweetened by an up-front cash payment (see Table 6). None of the Ukrainian debt treatments involved principal reductions.

Table 6.

Nominal Principal Reduction in Recent Debt Restructuring Cases

article image
Source: IMF staff reports.

Negative numbers indicate an increase in principal.

Past due interest.

Not applicable; case did not involve arrears.

London Club agreement included upfront clearance of US$30 million in arrears.

In the offer, only PDI through end-2001 was recognized.

Pakistan restructured US$929 million of commercial loans in 1998/99, of which US$777 million involved the rolling over of trade finance facilities on an annual basis for three years. In November 1999, Pakistan exchanged its outstanding Eurobond obligations (US$608 million) into a US$-denominated six-year Eurobond with three years grace and a 10 percent coupon. The restructuring involved a nominal increase in principal outstanding.

Argentina’s June 2001 megaswap attracted about US$30 billion in outstanding government bonds. The exchange featured a reduction in near-term government debt service, however at the expense of steeply increasing liabilities in later years. It also involved a small increase in principal outstanding. Subsequently, the government announced a two-stage approach to a more comprehensive debt restructuring. During Phase I in December 2001, approximately US$41 billion in sovereign debt and US$9 billion in provincial debt was exchanged into new government-guaranteed loans featuring a reduction of interest rates to 70 percent of the contractual level (up to a maximum of 7 percent), a 17-month grace period for interest payments, and a three-year extension of maturities for those original claims maturing up to 2010. The exchange involved no reduction in principal. Before Phase II could be addressed, Argentina defaulted in late December 2001.

In January and February 2005, after three years in default, Argentina launched a global debt exchange offer to restructure defaulted bonds to private creditors residing both inside and outside of Argentina. Eligible for exchange were 152 different securities, amounting to a total of US$81.8 billion, including US$2.1 billion past due interest accrued through end-2001. Past due interest accrued since 2002 (around US$20 billion) was not recognized. Eleven new securities were offered in exchange, each with a detachable GDP warrant. The new securities include par bonds, which are not subject to a haircut on nominal principal, quasi par bonds with a principal haircut of 30 percent and discount bonds with a principal reduction of 66 percent. The new bonds differ in their repayment structures (grace periods between 21 years and 33 years, maturity between 30 years and 42 years), coupon structures (step-up structures on the par and discount bonds, fixed rate of 5.57 percent on the quasi par bonds), interest capitalization provisions, and cash sweeteners, which were designed to yield approximately equal net present value reductions.

Moldova employed a piecemeal debt restructuring strategy. In March 2000, agreements were reached with Dresdner Bank to cancel government-guaranteed debt, and with Hewlett-Packard to reschedule US$22.6 million in government guarantees over six years. In May 2000, a government guarantee to the Italian company TMCI Padovan (US$1.9 million) was restructured with a nine-year amortization schedule, and in August 2000, government guarantees worth US$4.6 million to the German company AKA were restructured over 15 years, with five years grace period. Arrears to Gazprom were cleared by issuance of promissory notes and by agreement on a debt-equity swap in early 2001. Moldova’s Eurobond exchange was the smallest exchange of the cases considered, covering US$39.7 million in outstanding principal of a single Eurobond. The restructuring was unique given that 78 percent of the outstanding principal was held by a single asset management company. Restructuring negotiations started in June 2002, an agreement in principle was reached in August 2002, and the agreement became effective in October 2002. Under the agreement, there was a US$2.55 million principal reduction and an upfront cash payment of US$3.97 million. The remaining obligation was restructured into a 61/2-year amortizing U.S. dollar-denominated bond with back-loaded principal payments. The bond paid U.S. dollar six-month LIBOR plus 426.5 bp. In September 2003, a rescheduling agreement was reached with Dresdner Bank regarding a called guarantee of EUR 2.3 million. Principal payments were stretched over three years, and there was a NPV reduction of 3 percent (using a 10 percent discount rate). In April 2004, an agreement was reached with Gazprom for cash settlement of defaulted promissory notes. The notes with a face value of US$111 million were settled for a cash payment of US$47 million, implying a NPV reduction of 58 percent (discounted at 10 percent). In June 2004, Moldova settled refinanced lease payments (US$11.1 million) that were in default, with Hewlett-Packard. The cash settlement implied a NPV reduction of 55 percent (discounted at 10 percent).

The Uruguayan bond exchange in April and May 2003 involved US$5 billion in outstanding debt. Investors were given the choice between two options: the first option exchanged each existing bond for a new obligation carrying a similar coupon and a longer maturity (five years, generally, blended in some cases with a 30-year bond). The second option would offer investors longer-dated bonds, which would serve as benchmark bonds and would consequently be more liquid than bonds offered under the first option. In some cases, also under this option these bonds would be blended with 30-year bonds. The exchange resulted in a reduction of principal of US$49 million, equivalent to 1.0 percent of the exchanged bonds.

The Dominican Republic exchanged in April/May 2005 US$1.4 billion in two outstanding bonds into two new amortizing bonds with longer maturities (2007-11 and 2013-18). The exchange did not involve a reduction in principal and reduced the net present value of claims only marginally (1 percent), as the case was largely seen as confined to the realm of liquidity needs. In October 2005, the Dominican Republic concluded an agreement with its London Club creditors to reschedule US$125 million in principal falling due in 2005-06. The agreement featured a maturity of five years and a grace period of three years, and the average interest rate was reduced by 2 percentage points. In the same month, an agreement was finalized with a Dutch bank to restructure US$50 million on similar terms.

In Ecuador, domestic debt (earning about 10.4 percent interest before the restructuring) was rolled over without a reduction of principal into seven-year bonds with two years grace period, paying LIBOR plus two percent interest rate. The defaulted external bonds were swapped into a single global US$-denominated 30-year step-up bond carrying an initial 4 percent interest rate that would increase by one percentage point per year to a maximum of 10 percent. This operation involved a principal reduction of between 0 percent and 60 percent on the different types of bonds. Bondholders were offered an option to convert this latter bond into a 12-year US$-denominated bond with 12 percent interest rate, at the price of accepting a further principal reduction of 35 percent. All in all, Ecuador received a principal reduction of nearly 40 percent on its defaulted external bonds. The deal also involved the cash payment of overdue interest obligations on the Brady bonds.

In Russia, the restructuring of domestic ruble debt (10.8 percent of GDP) in May 1999 involved the discounting of outstanding claims using a 50 percent discount rate. The resulting claim was exchanged into a menu of new assets: 10 percent in cash as well as three- and six-month treasury bills, 70 percent into four- and five-year bonds with coupons yielding 30 percent per annum in the first year, declining steadily to 10 percent in the final year, and 20 percent in government paper that could be used to settle tax arrears or invest in Russian enterprises. Regarding the defaulted MinFin-3 bonds, investors were allowed to exchange their claims for a combination of new foreign currency denominated eight-year bonds and four-year ruble denominated bonds at an interest rate of 15 percent in the first year and 10 percent thereafter. The Soviet-era debt owed to the London Club (US$29 billion in PRINs and IANs) was exchanged into US$21.2 billion worth of new 30-year Eurobonds with seven years grace period, paying interest of initially 2.3 percent, increasing over time to 7.6 percent. Simultaneously, past due interest on PRINs and IANs was capitalized into a ten-year Eurobond with six years grace, carrying coupons of 8.25 percent. The operation was sweetened by an up-front cash element (see Table 5). Overall, the Russian restructuring operations involved a principal reduction of approximately 4.1 percent of GDP.

Annex III. Debt Decompositions in the Post-Restructuring Phases

This Annex provides an analysis of the comparison between actual debt dynamics of the country cases in their respective post-restructuring periods with staff’s projections made in the staff reports following the restructurings. Since this analysis requires at least one year of post-restructuring period, it is confined to cases that completed their restructurings prior to 2004 (Ecuador, Pakistan, Russia, Ukraine, and Uruguay).54

The comparison of the evolution of the debt-to-GDP ratio in the post-restructuring period (from the respective central year of restructuring until 2004) relative to Fund staff projections shows that in a majority of cases (Pakistan, Russia, Ukraine, and Uruguay), debt dynamics fared better then anticipated (see Figure III.1). Only in the case of Ecuador did debt decline by less than had been anticipated in the aftermath of the restructuring.

Figure III.1.
Figure III.1.

Change in Debt/GDP Ratios, Post-Restructuring Period

(in percentage points)

Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

Source: IMF staff reports.

The reasons for deviations from the projected paths can be analyzed by decomposing the debt dynamics into contributions by the primary balance, the interest-growth differential, the exchange rate, and other identified debt-creating flows, capturing any changes in public sector liabilities that are not reflected in the fiscal balance.55 56 Comparisons of the decomposition of the debt dynamics are made difficult by the fact that around the time of the examined restructurings, many staff analyses and reports did not show the full set of assumptions necessary to compute the decomposition. In these cases, reasonable assumptions were added to the set of published variables to complement the dataset.

Table III.1 shows such decompositions for the five-country cases during their respective post-restructuring periods (showing cumulative contributions from the year after the restructuring until 2004), comparing projections made in the staff reports following the restructurings (“Post-” columns) with actual outcomes or latest estimates (“Latest”) columns.

Table III.1.

Evolution of Debt/GDP Ratios Since the Crises

(in percent of GDP; unless otherwise stated)

article image
Sources: IMF staff reports; and staff calculations.

Change in debt ratio from end-1999 to end-2004, and contributions during 2000-04.

Change in debt ratio from end-1999/00 to end-2003/04, and contributions during 2000/01-03/04.

Change in debt ratio from end-2003 to end-2004, and contributions during 2004.

As reported or implicit in the respective staff report; numbers in bold Italics indicate that additional assumptions had to be made.

Change in debt ratio from end-2000 to end-2003, and contributions during 2001-03.

From the decomposition, it appears that only a minority of countries (Russia and Uruguay) outperformed the primary fiscal path projected by staff following the restructurings, whereas in the case of Ecuador, and to a smaller extent also in Pakistan and Ukraine, the primary balance was worse than projected (see Figure III.2).57 58 Individual country circumstances differed.59

Figure III.2.
Figure III.2.

Contribution of Primary Balance to Change in Debt/GDP Ratio, Post-Restructuring Period

(in percentage points)

Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

Sources: IMF staff reports; and staff calculations
  • In Russia—where the outcome exceeded projections by the widest margin—reasons for rapid improvements in the primary balance can be traced to higher-than-projected revenue, owing largely to favorable developments in oil prices.

  • Uruguay’s 2004 primary surplus (3.8 percent of GDP) was higher-than-had been anticipated after the restructuring largely because of expenditure restraint. Lower than anticipated spending on social security benefits and on wages, coupled with lower capital expenditure largely explain the better outcome.

  • In Pakistan, fiscal performance fell marginally short of the ambitious projections, in spite of a surge in grants given Pakistan’s increased geopolitical importance after the events of September 11, 2001.60

  • Ecuador showed underperformance relative to projections by the widest margin. Despite having posted substantial primary surpluses in the post-crisis period, levels projected after the restructuring could not be attained, as the country was unable to fully capitalize on higher oil prices against the backdrop of large increases in the wage bill and social security benefits, discretionary tax cuts, and a rapid decline in oil output owing to inefficiencies at the state oil company.

In the post-restructuring period, the contribution of the exchange rate to debt dynamics was generally better than anticipated in post-restructuring Fund projections (see Figure III.3). Programs generally anticipated a debt-increasing effect of the exchange rate, and actual performance in most cases shows a smaller-than-expected increase. This was the case in Pakistan, Ukraine, and Uruguay, while the contribution was slightly above expectations in Russia. Due to full dollarization, there was no exchange rate effect in Ecuador.

Figure III.3.
Figure III.3.

Contribution of Exchange Rate to Change in Debt/GDP Ratio, Post-Restructuring Period

(in percentage points)

Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

Sources: IMF staff reports; and staff calculations.

In a slight majority of countries (Pakistan, Russia, and Ukraine), the real interest-growth differential turned out more favorable than assumed after the restructurings (see Figure III.4). This was brought by higher-than-expected growth (Russia and Ukraine) as well as lower-than-expected real interest rates (Pakistan and Russia).

Figure III.4.
Figure III.4.

Contribution of Interest-Growth Differential to Change in Debt/GDP Ratio, Post-Restructuring Period

(in percentage points)

Citation: Policy Papers 2006, 049; 10.5089/9781498332279.007.A001

Sources: IMF staff reports; and staff calculations.

Annex IV. Public Debt Ratios and Probabilities of Debt Crisis

We analyze the relationship between crisis probabilities and the ratio of public debt to GDP using a pooled probit model on an unbalanced sample of 55 low- and middle-income countries over the period 1971-2002. The countries in the sample are: Algeria, Argentina, Bangladesh, Bolivia, Brazil, Burkina Faso, Cameroon, Chile, Colombia, Costa Rica, Côte d’Ivoire, Czech Republic, Dominican Republic, Ecuador, Egypt, El Salvador, Ethiopia, Gabon, Ghana, Guatemala, Haiti, Honduras, Hungary, India, Indonesia, Jamaica, Jordan, Kenya, Korea, Malaysia, Mexico, Morocco, Nepal, Nicaragua, Niger, Nigeria, Pakistan, Panama, Paraguay, Peru, Philippines, Poland, Russia, South Africa, Senegal, Sri Lanka, Tanzania, Thailand, Togo, Tunisia, Turkey, Ukraine, Uruguay, Venezuela, and Zimbabwe.

The binary dependent variable (CRISIS) represents the occurrence of public debt crisis. It takes the value of 1 when a country is classified as in default by Standard and Poor’s. Independent variables are the lagged ratio of public debt to GDP (DEBT1), lagged GDP growth (GGDP1), and the lagged ratio of short-term external debt to GDP (ST_EXTDEBT_GDP1). All coefficients have the expected sign. DEBT1 and GGDP1 are significant at the one percent level, while ST_EXTDEBT_GDP1 is significant at the 10 percent level (see Table IV.1). The goodness of fit, as measured by the McFadden R-squared, is fairly low at 0.17, not untypical for this type of pooled probit regression.

Table IV.1.

Probit Estimation of Crisis Probability

Dependent Variable: CRISIS

article image
Source: IMF staff calculations.

We compute in-sample forecasts from this equation to yield predicted crisis probabilities associated with the sample data points. From this, we plot inferred crisis probabilities against lagged debt ratios (see Figure 1.1). In order to be able to associate a single probability with any given level of debt, we fit a polynomial through the data (see Table IV.2).

Table IV.2.

Polynomial Fit Through In-Sample Forecasts

Dependent Variable: DEBT1

article image
Source: IMF staff calculations.

References

  • Barnhill, Theodore M., and George Kopits, 2003, “Assessing Fiscal Sustainability Under Uncertainty,IMF Working Paper 03/79 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Berg, Andrew, Eduardo Borensztein, and Catherine Pattillo, 2005, “Assessing Early Warning Systems: How Have They Worked in Practice?”, IMF Staff Papers Vol. 52/3 (Washington: International Monetary Fund)

    • Search Google Scholar
    • Export Citation
  • Burnside, Craig, Martin Eichenbaum, and Sergio Rebelo, 2003, “Government Finance in the Wake of Currency Crisis,NBER Working Paper No. 9786 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Chalk, Nigel, and Richard Hemming, 2000, “Assessing Fiscal Sustainability in Theory and Practice,IMF Working Paper 00/81 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Croce, Enzo, and V. Hugo Juan-Ramón, 2003, “Assessing Fiscal Sustainability: A Cross-Country Comparison,IMF Working Paper 03/145 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Detragiache, Enrica, and Antonio Spilimbergo, 2001, “Crises and Liquidity: Evidence and Interpretation,IMF Working Paper 01/2 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Ghosh, Atish, and others, 2002, IMF-Supported Programs in Capital Account Crises, IMF Occasional Paper No. 210 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Goldstein, Morris, 2003, “Debt Sustainability, Brazil, and the IMF,Working Paper WP 03-1 (Washington: Institute of International Economics).

    • Search Google Scholar
    • Export Citation
  • Guidotti, Pablo, and Manmohan Kumar, 1991, Domestic Public Debt of Externally Indebted Countries, IMF Occasional Paper No. 80 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Hemming, Richard, and Petrie Murray, 2000, “A Framework for Assessing Fiscal ” IMF Working Paper 00/52 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Hemming, Richard, and Teresa Ter-Minassian, 2003, Public Debt Dynamics and Fiscal Adjustment, in Managing Financial Crises—Recent Experience and Lessons for Latin America, IMF Occasional Paper No. 217 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Hemming, Richard, Michael Kell and Axel Schimmelpfennig, 2003, Fiscal Vulnerability and Financial Crises in Emerging Market Economies, IMF Occasional Paper No. 218 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2001, “Involving the Private Sector in the Resolution of Financial Crises—Restructuring International Sovereign Bonds.” Available on the IMF public website at http://www.imf.org/external/pubs/ft/series/03/IPS.pdf.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2002a, “Assessing Sustainability.” Available on the IMF public website at http://www.imf.org/external/np/pdr/sus/2002/eng/052802.pdf.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2002b, “Sovereign De