Journal Issue
Share
Article

Changing Patterns in Low-Income Country Financing and Implications for Fund Policies on External Financing and Debt

Author(s):
International Monetary Fund
Published Date:
February 2009
Share
  • ShareShare
Show Summary Details

Executive Summary

Low-income countries (LICs) face significant challenges in meeting their development objectives while maintaining a sustainable debt position. The international community’s main answer to this dilemma has been to promote recourse to concessional external resources. The Fund’s recommendations to LICs conform to this preference: the practice in Fund-supported programs in LICs has generally been to set zero limits on nonconcessional external borrowing while not restricting concessional financing, although flexibility has been applied on a case-by-case basis to allow some nonconcessional borrowing when warranted.

While the principle that programs need to address the terms as well as the amount of external borrowing remains valid, various factors warrant a review of the Fund’s policy on external debt limits in LICs. First and foremost, the situation of LICs has evolved and patterns of financing of LICs have changed substantially in recent years: a number of them have made good progress in strengthening macroeconomic management; debt burdens have been relieved; official financing has become available from a broader group of creditors; and until recently external private creditors’ interest in LICs was on the rise. Second, much work has been done to strengthen debt sustainability analyses (DSAs) and the joint Bank-Fund debt sustainability framework (DSF) was introduced since the last policy review of debt limits. Third, the current policy raises a number of implementation issues, such as the distinction between external and domestic debt, which require reconsideration as LICs become more integrated into the international financial system.

This paper proposes an approach that moves away from a single design for concessionality requirements towards a menu of options. Such an approach would allow to take greater account of the diversity of situations faced by LICs. Two aspects of diversity seem particularly relevant in this context: the extent of debt vulnerabilities and macroeconomic and public financial management capacity. The current practice, which is easy to implement and well known by all stakeholders, could continue to be applied to lower capacity countries, but with more flexibility for those with lower debt vulnerabilities. More flexible and sophisticated options, eschewing the debt-by-debt approach of the current policy, could be considered for higher capacity countries, including use of average concessionality requirements and targets on the present value of public external debt (or total public debt). For the most advanced LICs, consideration could be given to dropping concessionality requirements. Over time, an increasing number of LICs would be expected to move to the more flexible and sophisticated approaches as their macroeconomic and public financial management capacity improves. Consistent with the need for sound analytical underpinnings, all the options proposed in the paper would rely on DSAs.

The paper also discusses the issue of defining “external” debt (i.e., debt subject to concessionality requirements) as well as the institutional coverage of debt limits.

I. Introduction1

1. Low-income countries (LICs) face significant challenges in meeting their development objectives while maintaining a sustainable debt position.2 More external resources will be required in many of these countries to step up development spending, particularly investment in infrastructure and social areas. Yet, prudence is also required. There is a history of frustrated attempts at ramping up external borrowing to finance public investment, in the hope that the investment will yield additional income sufficient to service the additional debt. Such a strategy can work under the right conditions. But, as the many past debt crises in LICs have shown, it is risky for economies that have fragile fiscal and balance of payments positions, are vulnerable to unforeseen shocks, or have weak institutions.

2. The international community’s main answer to this dilemma has been to promote recourse to concessional external resources (including grants), which can allow countries to complement domestic savings and run higher levels of expenditures in a sustainable fashion. The preference for concessional finance was reaffirmed strongly after the delivery of MDRI relief.3 The Fund’s recommendations to LICs conform to this preference. In particular, Fund-supported programs in LICs generally preclude nonconcessional external borrowing, with exceptions occasionally made on a case-by-case basis based on country-specific circumstances.

3. While the principle that programs need to address the terms as well as the amount of external borrowing remains valid, various factors warrant a review of the Fund’s policy on external debt limits in LICs. First and foremost, the situation of LICs has evolved and patterns of financing of LICs have changed: a number of them have made good progress in strengthening macroeconomic management; debt burdens have been relieved; official financing has become available from a broader group of creditors; and until recently external private creditors’ interest in LICs was on the rise. Second, much work has been done to strengthen debt sustainability analyses (DSAs) and the joint Bank-Fund debt sustainability framework (DSF) was introduced since the last policy review of debt limits. Third, the current policy raises a number of implementation issues, such as the distinction between external and domestic debt, which require reconsideration as LICs become more integrated into the international financial system.

4. The paper is structured as follows: Section II briefly reviews the Fund’s current policy on external debt limits in LICs and its implementation. Section III discusses the developments that warrant a review of this policy. Section IV outlines possible reform options for the policy on debt concessionality and other aspects of debt limits with the broad aim to adapt the policy to the new financial realities. Section V raises issues for discussion. The goal of the paper is to solicit initial views from the Executive Board on these issues. Based on Directors’ views, a follow-up paper will lay out a specific policy proposal for Board consideration.

II. External Debt Limits in LICS: Policy and Practice

A. The current policy on external debt limits and its rationale

5. The Fund’s policy on external debt limits was established thirty years ago and amended three times since (Box 1). The guidelines on external debt limits (IMF Board Decision No. 6230–(79/140),4 as amended subsequently) apply to all members, not only LICs, with a Fund-supported program.5 Therefore, they were drafted in relatively general terms and stress the need for a flexible implementation, consistent with uniformity of treatment. They stipulate that “when the size and the rate of growth of external indebtedness is a relevant factor in the design of an adjustment program, a performance criterion relating to official and officially guaranteed foreign borrowing will be included in upper credit tranche arrangements.”

6. The debt limits policy has tried to meet a variety of objectives. Board discussions on this issue in the past show that the main objectives have been to: (i) prevent the accumulation of external debt during the arrangement period that could lead to unsustainable debt service obligations in the future, while allowing for adequate external financing; (ii) reduce other external vulnerabilities (e.g., related to the maturity structure of external debt); and (iii) ensure the overall consistency of financial programs so that domestic demand restraint is not threatened by unanticipated external borrowing. Other motivations for such ceilings have been advanced at times, such as encouraging member countries to monitor and control the accumulation of external debt obligations; and providing assurance to potential international creditors that sound external debt policies are being pursued, thereby facilitating access to appropriate external financing.

7. Concessional debt is not expected to be covered by external debt limits. The guidelines state that “flexibility will be exercised to ensure that the use of the performance criterion will not discourage capital flows of a concessional nature by excluding from the coverage of performance criteria debts defined as concessional”. This exclusion reflects the need to allow for adequate financing volumes while limiting the impact on debt ratios and external vulnerabilities. Concessionality is assessed from a creditor’s perspective, i.e., using as a reference the cost of funds to the creditor. The guidelines indeed define a debt as concessional “on the basis of currency-specific discount rates based on the OECD commercial interest reference rates (CIRRs), and including a grant element of at least 35 percent, provided that a higher grant element may be required in exceptional cases.”

Box 1.Evolution of IMF Guidelines on External Debt

The Guidelines on Performance Criteria With Respect to External Debt in Fund Arrangements (“guidelines on external debt”) were established in 1979 with the adoption of Decision No. 6230-(79/140).1 At the time, the limit applied to debt with maturities greater than one year and less than 10–12 years. Loans that were assessed as concessional using the OECD DAC’s definition (with a flat 10 percent discount rate, and a minimum grant element of 25 percent) were excluded from the performance criterion.

In 1983 amendments were made to encourage staff to include short-term external debt with a maturity of less than one year and, in some cases, to extend the limit to debt with maturities of as much as 15 years. In 1995 the method of calculating the level of concessionality was refined to use currency-specific discount rates derived from the OECD’s commercial interest reference rates. Also, the suggested minimum required grant element was increased from 25 percent to 35 percent. Moreover, the limits on the maturities of debt that fell within the performance criterion were abolished.

The last refinement of the guidelines was introduced in 2000, when the definition of debt was expanded and clarified in response to the rapid evolution of financial markets and instruments that had taken place since the mid-1990s. For the purpose of the guidelines, the term “debt” was defined to mean a current liability created under a contractual arrangement through the provision of value in the form of assets (including currency) or services, and which requires the obligor to make one or more payments in the form of assets (including currency) or services at some future point(s) in time. Such a clarification had the effect of separating the definition from any particular type of debt instrument.

1 Selected Decisions and Selected Documents of the IMF (http://www.imf.org/external/pubs/ft/sd/index.asp).

8. The institutional coverage of external debt limits is expected to be broad. According to the guidelines, “normally the performance criterion will relate to official and officially guaranteed foreign debt. The coverage will include official entities for which the government is financially responsible as well as private debt for which official guarantees have been extended and which, therefore, constitute a contingent liability of the government.” The purpose of this provision has been to ensure a broader coverage of the public sector than that generally allowed by the fiscal accounts, which often cover only the central government, thereby potentially missing public sector sources of external vulnerabilities and domestic demand.

9. Consistent with long-standing statistical principles, the definition of external debt has been based on the residency of the creditor. The guidelines are silent on this issue. When the 1979 guidelines on debt limits were adopted, there was probably little doubt that, in practice, external debt was simultaneously debt issued in foreign currency, under foreign law, and to a nonresident.

B. The implementation of the policy in LICs6

10. Most Fund-supported programs in LICs share the same overall design for external debt limits. Despite the flexibility allowed under the current guidelines on external debt, the general practice in PRGF arrangements, PSIs, EPCA-supported programs, and more recently under the high-access component of the ESF, has been to prohibit nonconcessional external borrowing and not to restrict concessional borrowing.7,8 Fiscal programs include targets that are consistent with the design of these external debt limits. About half of the programs in the past two years targeted a fiscal balance—measured, for data availability and reliability reasons, from “below the line”, i.e., the financing side—that either excluded foreign-financed capital expenditures, or could be adjusted in the event of additional concessional financing. The other half included targets on components of domestic budget financing, such as credit to the government.9

11. This “standard” practice applied to slightly less than half of Fund-supported programs in LICs as of mid-January, 2009. 17 programs (out of 37) used the standard minimum concessionality requirement of 35 percent and had no allowance for nonconcessional borrowing (Table 1).10

12. Debt limits have tended to be tighter in countries with higher debt vulnerabilities:

  • About one fourth of Fund-supported programs in LICs as of mid-January, 2009 had concessionality requirements above 35 percent, ranging from 45 to 100 percent. These higher requirements have applied in countries—although not all of them, see Section III.B—where DSAs concluded that there is a high risk of external debt distress (or a situation of debt distress) and where debt management capacity was limited.
  • In a few cases where debt sustainability was an issue, more comprehensive limits have been used. In recent years, a few programs have employed debt limits based on the present value (PV) of public and publicly guaranteed (PPG) external debt.11 Such (indicative) targets were used as a complement to the usual debt limits. PV limits encompass all external debt regardless of the level of concessionality.

13. Debt limits have also been looser than usual when the situation warranted it:

  • About a third of the Fund-supported programs in LICs in place as of mid-January, 2009 had a nonzero limit on nonconcessional borrowing. Including cases where waivers were granted for exceeding the limit, some nonconcessional borrowing has been permitted in close to 40 percent of programs.12 Nonzero ceilings have been included for two main reasons: (i) to finance critical large-scale projects, mostly infrastructure, for which concessional resources were not available; and (ii) to support, in a few cases, a gradual shift from concessional to market-based finance (Table 1).
  • In a few cases, concessionality was assessed using a “financing package” approach rather than debt by debt. A key requirement for this approach is that the various parts of a financing package (e.g., a nonconcessional loan and a grant) be sufficiently integrated so as to be considered one single debt for the purpose of assessing concessionality.13
Table 1.Concessionality Requirement for New External Borrowing for Countries with PRGF, PSI, ESF, SBA, and EPCA (as of January 15, 2009)
CountryRisk of Debt Distress (as of mid January, 2009)Concessionality requirementNon-zero limit on non-concessionalUse of and rationale for nonconcessional borrowing (X) or financing or waiver Waiver (*) 1/
ArmeniaLow35XGeneral budget financing and cofinancing of foreign financed projects
Cape Verde 3/Low35XInfrastructure projects agreed to as part of a plan to develop medium-term investment program, including for state enterprises
Georgia 4/Low35XGeneral budget financing for developing access to international capital markets
Honduras 4/7/Low35
MadagascarLow35
MaliLow35
MoldovaLow35
Mozambique 3/Low35XGeneral budget financing to align the program with authorities’ own procedures and support independent oversight of government borrowing; very small amount
Senegal 3/5/Low35X ,Exception for infrastructure projects which are considered to be financially viable; waiver due to a one-off French loan whose terms are comparable to those of the WAEMU financial market
Tanzania 3/Low35
Uganda 3/Low35XInfrastructure projects critical to economic growth (to address power shortages)
ZambiaLow35XInfrastructure projects critical to economic growth (to address power shortages)
BeninModerate35*Government guaranteed debt to Benin Telecoms carried over from the previous government
Kyrgyz Republic 5/Moderate35
Malawi 3/Moderate35
MauritaniaModerate35*Infrastructure projects considered crucial for economic development and authorities agreed to negotiate for more concessional terms
NicaraguaModerate35
NigerModerate35
RwandaModerate50XInfrastructure projects important for economic growth and financially viable projects
Sierra LeoneModerate35
AfghanistanHigh60
Burkina FasoHigh35
BurundiHigh50
Central African RepublicHigh35
Congo, Republic ofHigh50
DjiboutiHighn.a.
Gambia, TheHigh45
GrenadaHigh35XInfrastructure projects for which plans were already well advanced at the time of the program and to accommodate bilateral loans slightly below minimum concessionality
HaitiHigh35
Comoros 5/6/In debt distress50
Cote d’Ivoire 6/In debt distress35
GuineaIn debt distress35
Guinea Bissau 6/In debt distress50
LiberiaIn debt distress100
TogoIn debt distress35
Pakistan 4/n.a.35XGeneral budget financing; Pakistan tapped international capital markets in the past

Indicates the occurrence of a non-zero limit or granting of a waiver at any test date during the period of the current program.

PRGF and EFF

PSI

SBA

ESF

EPCA

Risk of debt distress not listed explicitly in the DSA, but derived from the text of the latest Fund update (EBS/08/39).

Indicates the occurrence of a non-zero limit or granting of a waiver at any test date during the period of the current program.

PRGF and EFF

PSI

SBA

ESF

EPCA

Risk of debt distress not listed explicitly in the DSA, but derived from the text of the latest Fund update (EBS/08/39).

14. The sectoral coverage of debt limits has varied significantly across countries, particularly for official entities other than the central government. In many countries, only the central government was included, reflecting various considerations: data availability; a desire to keep the same institutional coverage as for the fiscal accounts; and the fact that external borrowing by other official entities generally requires a central government guarantee, and therefore is covered indirectly by the performance criterion. In some countries, certain public enterprises are either explicitly part of the official sector or excluded from it. The latter case generally corresponds to entities (such as public enterprises) which are commercially viable and can borrow externally on nonconcessional terms without a guarantee from the central government.14 Other practical country-specific considerations have also affected coverage. For instance, in the West African Economic and Monetary Union (WAEMU) where the market for government securities is regional, market borrowing from residents of another WAEMU country is not considered as falling under the external debt limit.

III. Why do Some Aspects of the Debt Limits Policy need to be Reviewed?

A. Issues raised by changing external financing patterns and the debt outlook in LICs

15. This section presents some stylized facts on changing external financing patterns and the debt outlook in LICs, which have implications for the Fund’s debt policy. The main findings of Appendix I on external financing trends in LICs through 2007 and Appendix II on the debt outlook are summarized below. These findings should be interpreted cautiously given serious data limitations.15 There is no single, comprehensive, and fully reliable database that covers all aspects of external financing flows to LICs. Several databases have been used for this purpose. These are not necessarily consistent and suffer from a number of shortcomings. The findings are therefore stylized facts, rather than precise estimates.

16. The main stylized facts relevant for this paper are:

  • Total external financing flows to LICs have increased significantly (as a share of recipients’ GDP) in the last two decades. This is attributable to private sector-to-private sector flows—mostly foreign direct investment (FDI) and private transfers. Official flows remain, however, the main source of LIC government financing by far. Private external financing of the public sector is on average still limited, but was (at least through 2007) significant in some LICs.
  • The structure of official flows has changed considerably. Traditional (DAC) bilateral donors now provide mostly grants. The shift to grant financing is less marked for multilateral institutions, which remain a significant source of debt financing for LIC governments. There is ample anecdotal evidence that the share of nontraditional (non-DAC) bilateral donors/creditors in new debt financing is on the rise though hard data on this trend are scarce.
  • Debt vulnerabilities are still significant in many LICs, including in some post-MDRI countries. The latest available DSAs indicate that 40 percent of LICs are at a high risk of external debt distress or already in debt distress, and this share rises to 70 percent if countries at a moderate risk of debt distress are added.
  • Beyond these averages, LICs face very diverse situations, both in terms of the composition of external financing and in terms of debt vulnerabilities.

17. Some of these findings confirm that the focus of the debt limits policy on preventing the build up of unsustainable debts while allowing for adequate external financing remains appropriate. Many LICs continue to be in fragile debt situations, while their development and financing needs remain huge. In addition, many development expenses, particularly those in the health and education areas, may have a substantial impact on the population’s welfare in the long term, but do not contribute significantly in the short and medium term to the country’s repayment capacity. For these reasons, recourse to external concessional resources (including grants) remains highly desirable for many LICs, as it allows them to increase the amount of expenditures they can finance in a sustainable way. The focus on concessional resources, which come mostly from official donors, is also justified by the fact that official financing of LIC governments remains predominant.

18. However, the recent trends also suggest a need to review some aspects of the debt limits policy:

  • The LIC universe comprises countries with very different characteristics with regard to external financing and debt, ranging from the very poor, heavily indebted, and highly aid-dependent countries to countries that have established a strong track record of macroeconomic performance and have had market access. The debt situation of about 30 percent of LICs is favorable from a debt sustainability perspective—they have been assessed at a low risk of debt distress. For these countries, consideration of more accommodating debt limits is warranted.
  • The range of donors and creditors engaged and interested in LICs has expanded in the past few years, at a time when the scaling up of aid promised by traditional donors has not yet fully materialized. Many of these donors/creditors have less concessional lending policies or practices than traditional donors, but they are willing to finance potentially high-return, cash-generating infrastructure projects for which concessional resources from traditional donors may not be available.16 LICs are often keen to work with these creditors/donors, which constitute an additional source of financing and are viewed as fast-delivering and less administratively onerous than traditional donors.17
  • With sometimes sizable external private financing of the government, particularly in the form of nonresident purchases of domestically-issued bonds, the distinction between external and domestic debt (as traditionally understood) is blurring. One reason for applying concessionality requirements to external debt was to reduce the exchange rate risk associated with market financing in foreign currency. However, the increasing amount of domestic currency debt held by nonresidents (in the context of fully convertible currencies) has weakened this rationale and raises the question of whether the use of the residency criterion remains appropriate for determining the scope of debt concessionality requirements.
  • The ongoing financial crisis will likely affect significantly the size and composition of financing flows to LICs in the near future. For instance, private capital inflows to LICs are expected to decrease sharply in 2009. However, these short-term developments do not weaken the case for reviewing aspects of the debt limits policy.

B. Other analytical and operational issues

19. The current debt limits policy does not incorporate recent advances in the Fund’s analytical approach to assessing debt sustainability. The concessionality policy was designed before the introduction of the DSF, and more generally the debt limits policy was designed before the introduction of formal DSAs (see Box 2 for a description of the DSF). The DSF provides a comprehensive approach to assessing debt sustainability and therefore for guiding choices about the type and amount of debt that could be contracted in a prudent manner. However, DSA results do not yet seem to inform systematically the design of debt limits (Figure 1 shows limited correlation between the risk ratings and the level of minimum concessionality requirements).

Box 2.The Main Components of the Debt Sustainability Framework

Under the DSF, DSAs consist of:

  • An analysis of a country’s projected external and public sector debt burden and its vulnerability to external and policy shocks—baseline and shock scenarios are calculated; and
  • An assessment of the risk of debt distress based on indicative external debt burden thresholds that depend on the quality of the country’s policies and institutions. The assessment is made by comparing external debt burden indicators in the various scenarios against their thresholds. Breaches of thresholds, present or projected, play an important role in the assessment.

There are four possible ratings of external debt distress: Low, moderate, high, and in debt distress. Risk ratings are determined as follows:

  • Low risk, when all the debt burden indicators are well below the thresholds;
  • Moderate risk, when debt burden indicators are below the thresholds in the baseline scenario, but stress tests indicate that the thresholds are breached if there are external shocks or abrupt changes in macroeconomic policies;
  • High risk, when one or more debt burden indicators breach the thresholds under the baseline scenario; or
  • In debt distress, when the country is already having repayment difficulties.

The quality of policies and institutions is measured by the CPIA index, compiled annually by the World Bank. The DSF divides countries into three performance categories: strong (CPIA at or above 3.75), medium (CPIA between 3.25 and 3.75), and poor (CPIA at or below 3.25). To reduce undesirable volatility in the debt distress ratings from annual fluctuations in the CPIA, a three-year moving average CPIA score is used to determine a country’s policy performance under the DSF.

Figure 1.Debt Distress Rating and Minimum Concessionality Requirements

(as of January 15, 2009)

20. Related to this, the minimum concessionality requirement of 35 percent has sometimes been criticized as arbitrary. The Fund has used this threshold since 1995, as well as the methodology based on currency-specific discount rates, mostly with a view to harmonizing its requirements with those of the OECD’s agreement on officially-supported export credits and reducing compliance costs for creditors.18

21. It can also be argued that current practice may constrain excessively debt management operations and capacity development. The prohibition of nonconcessional external financing in countries that could afford it may lead them to enter into operations which are not necessarily optimal from a debt management perspective. For instance, it is not always preferable to borrow from the domestic market than to borrow (nonconcessionally) from international markets. Similarly, successful LICs can aspire to reduce their dependence on aid and to rely increasingly on market financing over time, as they acquire characteristics of emerging market economies. This requires early development of debt management capacity and opportunities to hone these skills through more diverse financing schemes than permitted under the debt limits policy.

22. The blurring of the distinction between domestic and external debt raises practical and analytical problems. For program monitoring purposes, nonresident purchases of domestically-issued debt create new external debt, which falls under the related ceiling on nonconcessional external borrowing. This raises implementation issues, as the authorities are held accountable for transactions over which they have only limited control and monitoring is often difficult given the limited availability of information on secondary market transactions. This practice is also hard to defend from an economic perspective. It builds into programs a bias against nonresident purchases of domestically-issued local-currency public debt, while such purchases do not, by themselves, change the level or currency composition of public debt. Furthermore, within a well-designed reform strategy, nonresident involvement in public debt markets may be desirable insofar as it fosters the development of domestic financial systems.

IV. Some Options for Reform

23. The previous section illustrates the need to review some aspects of debt limits in LICs. This section will focus first and foremost on concessionality requirements. It will then discuss more briefly the issue of defining external debt as well as the institutional coverage of debt limits.

A. Concessionality requirements

24. Concessionality requirements would be more effective in achieving their intended objectives, and would be perceived as such by all stakeholders, if they were based on a clearer analytical foundation. Linking concessionality requirements, whose main goal is to ensure debt sustainability, more closely to DSAs would be a natural step in this regard, particularly at a time when the DSF is becoming a tool of reference both for LICs and their donors.

25. A primary goal of any new approach should be to take greater account of the diversity of situations faced by LICs. Two aspects of diversity seem particularly relevant in this context:

  • The extent of debt vulnerabilities. A country where debt sustainability concerns are high should adopt tighter debt limits more systematically, possibly involving limits on total debt (not only nonconcessional debt) and/or a higher minimum concessionality requirement.19 Conversely, if debt vulnerabilities are low, looser limits should be considered, which could allow for some nonconcessional borrowing. DSAs would be the right instrument to assess the extent of debt vulnerabilities.
  • The country’s macroeconomic and public financial management capacity. One of the virtues of the current approach, which should not be underestimated, is that it requires limited capacity from country authorities. The methodology and information requirements are relatively simple. It is also easy to monitor—an important attribute in a program context, where the authorities need to know where they stand regarding the attainment of their objectives on a continuous basis. This approach seems broadly adequate for countries with low or moderate administrative capacity. However, countries with a strong track record of macroeconomic discipline and public financial management (including a strong capacity to identify and implement suitable projects), and where capacity is developed enough to handle directly the whole gamut of donors and creditors and their various financing instruments, could benefit from use of a more sophisticated approach.

26. Consistent with the above principles, the reform proposal below is based on a menu of options (see Table 2). Given the need for sound analytical underpinnings, all the options would rely on DSAs. For a given country, the choice of the appropriate option would be based on the two criteria discussed in the previous paragraph. Unless debt sustainability is a serious concern and capacity is limited, the options discussed below involve more flexibility for LICs, including those building on the current approach.20

27. For lower capacity countries, the current approach could continue to be applied, albeit with more flexibility and a more systematic link to DSAs. Countries with a low or moderate DSA risk rating would be in the lower vulnerability category, while those with a high risk rating (or in debt distress) would be in the higher vulnerability category:21

  • For countries with lower vulnerabilities, the concessionality level would be 35 percent and nonzero limits on nonconcessional borrowing could be considered more systematically—or set higher—than current practice and/or be more frequently untied.22, 23 Higher limits would allow countries to undertake more infrastructure investment, while untied limits would give the authorities more freedom in choosing projects and financing. The size of these limits could be based on the results of DSA tests.24
  • For countries with higher vulnerabilities, the concessionality requirement would generally be set at 50 percent or above. The presumption would be that there would be no nonconcessional borrowing, except in exceptional circumstances (e.g., financing with a grant element marginally below the minimum requirement, or critical and highly profitable project for which concessional financing is not available).
Table 2.Concessionality Options Matrix
Extent of debt vulnerabilities (as assessed by DSAs)
LowerHigher
Macroeconomic and public financial management capacityLowerMinimum concessionality requirement based on the current system, but with added flexibility on nonconcessional external debt (e.g., higher and untied nonzero limits, if consistent with maintenance of low debt vulnerabilities)Maintain minimum concessionality requirement based on current system, likely higher than 35 percent, with limited or no room for nonconcessional borrowing and possibly nominal limits on concessional borrowing too.
HigherMinimum average concessionality requirement over six months or the fiscal year applied to external or total public borrowing; for most advanced LICs, no concessionality requirements and overall nominal debt limit if neededOverall limit on the PV of external or total public debt or average concessionality at a higher level; for most advanced LICs, ceilings on nominal external or total public debt

28. For higher capacity countries, more flexible and sophisticated options, eschewing the debt-by-debt approach of the current policy, could be considered:25

  • For those with lower vulnerabilities, concessionality could be measured on average over six months (the period typically covered by a review in PRGF-supported programs) or the fiscal year.26 The target for the average level of concessionality would be informed by the DSA and should be consistent with an unchanged risk rating. It could therefore be met by mixing highly concessional borrowing with nonconcessional borrowing, leaving significant margin for maneuver to the authorities. This approach could also be applied to total (not only external) public debt.
  • For countries with higher vulnerabilities, a debt limit in PV terms may be more appropriate, because this would limit overall debt accumulation, while still leaving some margin for maneuver to the authorities. Should a PV target be considered too difficult to implement (see Box 3 for a description of operational issues with PV targets), an average minimum concessionality requirement could be used, but with a higher threshold derived from the DSA. In both cases, the target could in principle be applied to external or total public debt.
  • Finally, for the most advanced LICs, consideration could be given to dropping concessionality requirements. These countries would typically have higher per capita income, a strong track record of macroeconomic and public financial management, significant market access, and experience in dealing with nonconcessional financing.27 For such countries, strict concessionality requirements would likely be counterproductive, especially in the presence of a sound fiscal framework. If debt vulnerabilities remain high, or if the coverage of the public sector in the fiscal accounts is narrow, debt ceilings on nominal external or total public debt or in some cases sub-ceilings on specific types of debt could be contemplated, consistent with the fiscal framework. Such a treatment of the most advanced LICs could be applied to lower middle-income countries too, as has already been done on a case-by-case basis.28

Box 3.Operational Implications of a PV Target

The use of PV-based targets has increased in the past few years with the introduction of the DSF, but remains limited. This approach follows the methodology of the DSF to calculate and project the PV of debt. The appeal of this approach is that it is comprehensive and fully consistent with the DSA. A few countries have, or had, PV-based conditionality in their programs in addition to traditional fiscal targets and limits on nonconcessional external debt accumulation. Examples include Guyana, Rwanda, the Central African Republic, and The Gambia.

At first glance, the data requirements and capacity needed to implement PV-based conditionality appear manageable. The data required include projected disbursements, including from existing loans, on a loan-by-loan basis, in the original currency of disbursement, over several years. Projected disbursements are converted into U.S. dollars using the WEO exchange rates. The debt service is generated using the terms and conditions of each loan. A uniform 5 percent discount rate is applied to calculate the PV of debt service obligations, in line with the methodology of the DSF.

In practice, however, the implementation and monitoring of a PV target is much more challenging than the current concessionality requirements. In the latter, a new debt is assessed once and for all at the time of contracting. In a PV target approach, a loan whose disbursements are phased must be monitored on a continuous basis. This generally requires timely sharing of information among different government agencies, which is not always the rule. For instance, it is quite common for debt management units to be informed late of disbursements on project loans. All the information on disbursements must then be aggregated on a regular basis to ascertain that the target for the period is within reach. While all these tasks can be performed by countries with high administrative capacity, experience has shown that they can raise significant challenges for countries with more limited capacity.

PV targets may be more appropriate as indicative targets than as performance criteria (a point endorsed by the Board in the past). The PV of debt is highly sensitive to the timing of projected disbursements, which is difficult to predict and sometimes beyond the authorities’ control. Disbursements on project loans can be affected by donors’ internal procedures and requirements, limited domestic implementation capacities, or lack of domestic counterpart funds, while disbursement of budget support may be tied to donors’ conditionality. Also, the higher information-sharing requirements of a PV target increase the risk of misreporting. Some of these considerations also apply to the average concessionality approach, which could also be used for indicative targets rather than performance criteria.

29. Over time, an increasing number of LICs would be expected to move to the more flexible and sophisticated approaches as their macroeconomic and public financial management capacity improves. At this juncture, a majority of LICs would still be expected to be classified in the lower capacity category. They would nevertheless benefit from a more flexible application of the current approach, giving them more financing choices while limiting the risks to debt sustainability. Thanks to the substantial outreach efforts that have been made in the recent past, the current framework is well understood by donors and creditors. Keeping it in place for lower capacity countries will ensure that those donors and creditors that are committed to making their lending consistent with Fund (and Bank) concessionality requirements (e.g., OECD export credit agencies, other multilaterals, etc.) will continue to do so for a large group of LICs. As country capacity improves over time, including thanks to technical assistance efforts of the international community, LICs would move to the more flexible options for ensuring debt sustainability.29

30. Consistency of the financial program supported by a Fund arrangement can be achieved with any of the above options for debt limits. There may be cases where debt limits solely based on sustainability considerations would allow for too much borrowing from a demand management perspective. In such cases, any of the options could be combined with a limit on the overall fiscal deficit, which would then be the binding constraint. In cases where there are debt vulnerabilities other than long-term sustainability, nominal sub-ceilings on certain categories of debt could be considered (e.g., limits on debts with an original maturity below a certain threshold).

Challenges and possible drawbacks with the suggested approach

31. The framework outlined above would depend heavily on assessments of macroeconomic and public financial management capacity, which would inevitably have a subjective element. A possible starting point would be to look at the CPIA (or one of its sub-components). Use of the CPIA would be consistent with the role played by this index in determining the policy-dependent thresholds in the DSF. Another advantage is that the CPIA is published and available for almost all LICs. In addition to the CPIA, other indicators would be taken into account, such as the presence and successful implementation of a Medium-Term Debt Management Strategy (MTDS) and Public Expenditure and Financial Accountability (PEFA).30 Overall, an assessment of macroeconomic and public financial management capacity would rely on these indicators as well as on relevant qualitative information and therefore would involve judgment.

32. The proposed framework would put a greater onus on DSAs and could make them more contentious. For lower capacity countries, the risk rating (or an overall assessment based on total public debt) would systematically influence the minimum concessionality requirement, and the DSA would be used to determine how much nonconcessional borrowing could be contracted. For higher capacity countries, the reliance on DSAs could be even more extensive, particularly for the use of PV targets. This central role given to DSAs could strengthen the criticism that DSAs are too conservative, and in particular that they do not take adequately into account the impact of additional (debt-financed) public investment on growth and exports. While this issue goes well beyond the DSF itself—it is really about how to assess the relationship between investment and growth in LICs’ macroeconomic frameworks—more work in this area would be warranted in any case. In this regard, consideration could be given to building on the work done recently by staff in the context of aid scaling-up scenarios.31 However, it should be noted that DSAs already play a critical role in IDA’s policy on nonconcessional borrowing, with ratings indirectly determining whether a country is subject to this policy. Also, the greater role played by DSAs could be a strong incentive for country authorities and other stakeholders to pay even more attention to the framework.

33. Moving to targets on public debt, rather than PPG external debt, could raise new challenges. Public debt is a more comprehensive concept, and therefore a more relevant variable from the perspective of debt management and fiscal sustainability. Setting limits on public debt would also address the issue of the blurring distinction between external and domestic debt in some LICs. But such a move would also raise a number of operational issues, such as the definition of public debt (e.g., treatment of domestic arrears) and whether DSAs provide sufficient guidance on how to set limits on total public debt.32 The move to public debt targets should also not reduce the focus on external vulnerabilities. At this juncture, staff recommends that such targets could be tried in specific cases where the potential problems are considered manageable. An assessment of experience could be conducted after a period of time.

34. The increased flexibility under the more advanced options reduces the Fund’s traditional “gate keeper” role and may weaken donors’ incentives to provide concessional resources. This gate keeping function has led to the perception that the Fund is an obstacle to financing for development. Under these options, the gate-keeping (or creditor coordinating) function would entirely be performed by country authorities. This is why it is critical that these options be used only in countries with high capacity. The issue of donors’ incentives is a real one, particularly for those for which concessionality requirements have proved binding. Minimum concessionality requirements can, in some circumstances, strengthen the bargaining power of borrowing countries. However, many traditional donors provide financing (including grants) on terms much more favorable than required by concessionality requirements, suggesting that the latter is not the main factor influencing their financing terms. In addition, traditional donors’ concerns about free riding should be mitigated by the increased focus of the concessionality policy on debt sustainability, which gives them stronger assurances that their efforts to provide development finance will not be jeopardized by excessive borrowing from other sources.

B. Use of the residency criterion to define external debt

35. The increasing role of nonresidents in domestic debt markets in a number of LICs warrants a reconsideration of the use of the residency criterion in the debt limits policy. As mentioned earlier, in addition to long-standing statistical principles defining external debt by reference to the residency of the creditor, external debt in the past also normally meant debt in foreign currency and contracted under foreign law. The limits on nonconcessional external debt under the guidelines, so defined, therefore pushed donors to provide resources with a minimum grant element, but also contributed to reducing the exchange rate and rollover risk associated with market financing in foreign currency. In countries with an opening capital account, the nature of the vulnerabilities change: nonresidents can buy domestically-issued debt in local currency (classifying it as external debt), and residents can buy foreign-currency debt issued abroad (classifying it as domestic debt). It is not obvious that the latter is less risky than the former.

36. For the more advanced LICs, the proposed debt limits would partly address this issue. The proposed options for these countries would either drop any debt limits or rely on total public debt, eliminating the need to distinguish between external and domestic debt in this context. However, a distinction of debt based on its characteristics (e.g., currency of denomination) may still be desirable from an external vulnerability analysis standpoint, even if not for concessionality purposes.

37. For LICs with still relatively closed capital accounts or very limited financial integration with the rest of the world, the use of the residency criterion would still be relevant with some amendments. To address monitorability issues, consideration could be given to excluding systematically from the concessionality requirements nonresident acquisitions of domestically-issued debt in the secondary market, which would be expected to be very limited.33 This exclusion could be complemented with some additional safeguards, to make sure that any additional vulnerabilities stemming from transactions with nonresidents are addressed.

38. The extent of safeguards would be determined on a case-by-case basis, taking into account country circumstances. Safeguards could include ensuring that: (i) the program relies on an appropriately broad concept for the fiscal deficit performance criterion, to close any definitional loopholes; (ii) all new borrowing on the domestic market should normally be in local currency; (iii) relevant changes are made in the program’s design (e.g., higher NIR targets), if needed to mitigate vulnerabilities associated with significant nonresident holdings of domestic debt; (iv) any such transactions are fully reflected, to the extent possible, in the external debt sustainability analysis, including with an explicit assessment of the vulnerabilities potentially associated with them (e.g., higher rollover risk in the case of short-term borrowing, a potential threat to the exchange rate and/or reserves in the event of sudden withdrawals); and (v) the authorities report to the Fund the terms of new domestic borrowing, including the currency composition, and take steps over time to improve their monitoring of secondary market transactions.

39. The most difficult case is that of LICs in an intermediate situation. For these countries, a system based on residency with exclusions may lose its internal coherence if the debts excluded from the application of the residency criterion become large. One option would be to use currency denomination as a criterion. Concessionality requirements could be applied only to foreign-currency denominated debt. Use of this criterion would mitigate the exchange rate risk associated with foreign-currency nonconcessional borrowing. Additional measures and safeguards to address any risk arising from large nonresident inflows into domestic-currency debt might be needed (cf. previous paragraph). Use of this criterion could however pose serious problems in dollarized economies, where debt issuance in the domestic market may be largely or entirely in foreign currency.34

C. Concessionality and public enterprises

40. While the policy allows and has been applied with considerable flexibility, there would be benefits in clarifying the institutional coverage of debt limits. Creditors and donors committed to observing concessionality requirements (e.g., export credit agencies in OECD countries) often query staff about the inclusion or exclusion of specific public entities, which suggests a need for clarification in this area.

41. Public enterprises and other official sector entities should be covered by debt limits, unless an explicit selective exclusion was made.35 A case for selective exclusion could be made for public enterprises and other official sector entities that are in a situation to borrow without a guarantee of the government and whose operations pose limited fiscal risk to the government.36 The objective would be to avoid constraining inappropriately their operations and potentially hampering investment. The following criteria could be used to decide on such exclusion:37 (i) managerial independence, including pricing policy and employment policy; (ii) relations with the government, including existence of subsidies and transfers, quasi-fiscal activities, and the nature of the regulatory and tax regime; (iii) governance structure, including periodic outside audits, publication of comprehensive annual reports, and shareholders’ rights; (iv) financial conditions and sustainability, including market access, less than full leveraging (debt-to-asset ratio comparable to the industry average), profitability, and record of past investment; and (v) other risk factors, including vulnerabilities stemming from contingent liabilities, and the importance of the public enterprise.

42. These criteria should permit an adequate distinction between high and low fiscal risks, although experience suggests that the application of the criteria must remain a matter of judgment.

Annex I. Other Options for Concessionality Requirements Explored by Staff but not Recommended in this Paper

Concessionality based on a lender-by-lender approach.

43. In a lender-by-lender approach, concessionality would be determined on average, based on the overall resource flows from a lender to a particular country over a given period (e.g., a year). This approach would help those lenders that can meet concessionality requirements on average, but not on a debt-by-debt basis, for instance because their policies do not allow them to lend concessionally for infrastructure projects, while they can do it for social sector projects. This approach raises a number of operational challenges raised by a PV target, without providing as much flexibility from a borrower’s perspective:

  • The timing of the contracting of the various loans from a given lender may not be entirely under the control of the authorities. This approach also raises informational and aggregation issues, potentially making it difficult for the authorities to monitor it.
  • It would require a precise definition of a lender. Bilateral donors/lenders often have several separate agencies that provide financing to the same country. These donors/lenders could argue that they represent one government, which could entail a coordination problem.
  • The definition of a period for which all the loans from one lender would need to satisfy the minimum grant element requirement may be problematic. This period may not correspond to all lenders’ fiscal years, raising planning and commitment issues.

Concessionality requirements based on type of expenditure.

44. A concessionality-by-expenditure approach would better reflect the growth and productivity impact of debt-financed expenditures. In theory, any project where the social return is higher than the cost of borrowing would be beneficial and could be pursued. But this approach also raises a number of operational issues. Because staff cannot assess the profitability of each project/expenditure, such an approach would need to be based on broad expenditure categories, reflecting their nature or purpose. Higher requirements could apply to social projects (expected to have low immediate returns), while lower requirements could be contemplated for infrastructure investments. In practice, however, the classification of a given project in a pre-established expenditure category may be challenging. It is also unclear how general budget financing should be handled. This approach could also lead to severe distortions in the structure of externally-financed expenditures.

Annex II. Survey of Donors Conducted for this Paper

45. Staff surveyed bilateral creditors and multilateral development banks regarding their lending practices and solicited their views on current Fund policies pertaining to external debt and concessionality. This appendix provides a brief summary of the responses to the survey. However, because the response rate was low (about 35 percent), the results should be interpreted with caution.

46. Creditor lending practices: 62 percent of respondents provide both loans and grants, 23 percent provide loans only, and 14 percent provide grants only. When asked which factors they take into account in deciding the level of concessionality of their loans, about one third of respondents pointed to IMF/IDA minimum concessionality requirements. The debt sustainability situation was the next most frequently cited consideration, while the type of expenditure being financed and per-capita income of the country were less frequently cited. Many creditors indicated that they consider a combination of these factors.

47. Lenders’ experience with Fund debt recommendations: About two thirds of respondents indicated that IMF recommendations were important in determining financing terms. Of these respondents, 35 percent cited the Fund’s minimum concessionality requirements. The next most frequently cited factors were: (i) the Fund’s general advice (in staff reports or DSAs); and (ii) the risk rating included in the DSA. More than one quarter of all respondents noted that, on occasion, Fund recommendations have led them not to finance a project they were previously considering. A bit more than half of respondents were aware of the possibility of a financing package approach for concessionality calculation purposes but of those the majority has only employed this approach occasionally.

48. Creditors’ views on the IMF concessionality policy About 80 percent of respondents indicated that the Fund’s concessionality policy serves a useful purpose. Nonetheless, somewhat fewer than half believed that the policy should be kept as is while one fifth of respondents thought the policy should be changed. Just over one third of respondents believed that the policy should allow for more nonconcessional borrowing. Close to 40 percent of respondents thought that the requirement of a 35 percent grant element for concessional loans should be reconsidered.

Questionnaire on Debt Policies

Terms of your agency’s financial assistance to LICs

1. Does your agency provide:

AllBilateral creditorsMultilateral creditors
Loans only23%8%50%
Grants only14%15%12.5%
Loans and grants62%77%37.5%

2. If your agency can vary the financing terms it provides to LICs, what are the factors taken into account in deciding the level of concessionality:

AllBilateral creditorsMultilateral creditors
Per-capita income of the country8%8%8%
Type of expenditure which is financed13%8%23%
Debt sustainability situation21%28%8%
IMF or IDA minimum concessionality requirements31%28%23%
Other reasons (please specify)18%20%23%
N/A10%8%15%

Your agency’s experience with IMF recommendations in the debt area

1. Are IMF recommendations an important consideration in determining your financing terms?

AllBilateral creditorsMultilateral creditors
Yes68%69%67%
No32%31%33%

2. If yes, please specify what specific recommendations are taken into account:

AllBilateral creditorsMultilateral creditors
Minimum concessionality requirements (if any)35%29%50%
General advice provided in staff reports or debt sustainability analyses19%24%10%
Risk rating included in debt sustainability analyses16%14%20%
Other (please specify)13%14%10%
n/a16%29%10%

3. Have IMF recommendations led you not to finance a specific project you were earlier considering?

AllBilateral creditorsMultilateral creditors
Yes29%15%50%
No71%85%50%

4. If yes, how often has this occurred in the last three years?

AllBilateral creditorsMultilateral creditors
More than 2/3 of your projects0%0%0%
Between 1/3 and 2/3 of your projects17%50%0%
Less than 1/3 of your projects83%50%100%

5. Are you aware that instruments from various donors can be assessed together for concessionality calculations purposes, provided they are sufficiently integrated into one single package?

AllBilateral creditorsMultilateral creditors
Yes57%46%75%
No43%54%25%

6. If yes, how often have you used this option in the last three years?

AllBilateral creditorsMultilateral creditors
More than 2/3 of your projects18%40%0%
Between 1/3 and 2/3 of your projects9%0%20%
Less than 1/3 of your projects73%60%80%

Your agency’s views on the IMF concessionality policy

1. This policy serves a useful purpose.

AllBilateral creditorsMultilateral creditors
Agree81%75%89%
Disagree5%0%11%
Neutral14%25%0%

2. This policy should be kept as is.

AllBilateral creditorsMultilateral creditors
Agree48%54%37.5%
Disagree19%0%50%
Neutral33%46%12.5%

3. The policy should be relaxed to allow for more nonconcessional borrowing by LICs

AllBilateral creditorsMultilateral creditors
Agree33%23%50%
Disagree43%46%37.5%
Neutral24%31%12.5%

4. The requirement of a 35 percent grant element for a loan to be considered concessional should be reconsidered

AllBilateral creditorsMultilateral creditors
Agree38%31%50%
Disagree29%23%37.5%
Neutral33%46%12.5%
Appendix I. Stylized Facts on Trends in LICS’ External Financing

49. This appendix provides a broad picture of external financing flows to LICs through 2007. Given significant data limitations, the approach followed here has not been to seek to produce accurate and comprehensive quantitative estimates. Rather, the objective has been to identify stylized facts, robust across data sources, which are relevant for this paper. The possible implications of the ongoing financial crisis on these flows are discussed in a separate paper.

50. There is no single, comprehensive, and fully reliable database that covers all aspects of external financing flows to LICs. While this appendix relies first and foremost on the database of Dorsey, Thomas William, Tadesse, Helaway, Singh, Sukhwinder and Brixiova, Zuzana (2008)”The Landscape of Capital Flows to Low-Income Countries” IMF Working Paper No. 08/51 (henceforth Dorsey and others 2008), updated for 2007, several other databases have been used to illustrate certain specific points (e.g., GDF, DAC and BIS). These databases are not necessarily consistent, in addition to suffering from a number of shortcomings (see Box A1). Nine PRGF-eligible countries were excluded from the sample, either because of data limitations (Afghanistan, Kiribati, Liberia, Somalia, Timor-Leste and Yemen) or because their large size and often atypical characteristics affected significantly the aggregate results (India, Nigeria, and Pakistan).

51. Total external financing flows to LICs have increased substantially (as a share of recipients’ GDP) in the last two decades (Figure A.I.1). Total net flows to LICs, including official and private lending, transfers, and FDI more than doubled in the 1990s compared with their level from the previous decade. They peaked in 2007 at about 15 percent of GDP on average, almost three times their 1980s level.

Figure A.I.1.Total Official and Private Net Flows to LICs’ Public and Private Sectors

Source: based on Dorsey and others (2008).

Box A1.Data and Methodological Issues

There is no single and comprehensive database that covers all aspects of external financing flows to LICs. This appendix uses therefore several databases. The primary source for the overall flows is Dorsey and others (2008), which is based on the WEO and has a broad country coverage and few missing observations. The main focus here is on net financing flows, because data for gross flows are not always available, at a relatively highly aggregated level. Other sources are used to discuss trends at a more disaggregated level, such as the databases of GDF, and DAC for official lending and grants and BIS for private banks’ lending (Table 1). Most of these sources present some shortcomings, either in terms of coverage, productions lags, or data accuracy and availability. In addition, as shown by Dorsey and others (2008), these sources are not always consistent.

Table 1.Decomposition of External Financing Flows to LICs and Sources of Data
LICso/w Public Sector
Total financing flows to LICsWEO/Dorsey and others
PrivateWEO/Dorsey and othersBIS
OfficialWEO/Dorsey and others
MultilateralDACGDF
BilateralDACGDF
non-OECDDRS
Table 2.Overview of Data Sources
WEOGDFOECD DACBISDRS
Capital flows coveredAll capital flows: Income credits, private transfers, official transfers, debt forgiveness and other capital account flows, FDI, portfolio flows, other investments and reserve movementsAll capital flows except income credits, official transfers reserve assetsOnly grants, concessional loans, debt reliefOnly private bank loansStocks and flows of public and publicly guaranteed debt, private non-guaranteed debt, short-term debt, debt forgiveness
General featuresMost comprehensive source; uses staff estimates as neededComprehensive source on stocks and flows of debt; uses official statistics and estimatesData from creditor and donor countriesData from private commercial banksData reported by debtors is combined with information obtained from creditors and estimates; extensive gaps in series
Data availabilityVery few missing data pointsVery few data missing pointsVery few data missing pointsVery few data missing pointsFew missing data points
Private vs. official flowsAvailableAvailableOfficial flows onlyPrivate flows onlyAvailable
Note: Some debt components of these data sources are reported in the Quarterly External Debt Statistics (QEDS). Source: Dorsey and others (2008) and staff estimates.
Note: Some debt components of these data sources are reported in the Quarterly External Debt Statistics (QEDS). Source: Dorsey and others (2008) and staff estimates.

52. This massive increase is attributable to private flows, which constitute by far the main source of external financing. 38 These flows, which had fluctuated between 1 and 3 percent of GDP in the 1980s, started to increase rapidly in the early 1990s to reach 11 percent of GDP on average in LICs. Trends on official flows are more difficult to interpret over the recent period, as they are affected by the recording of major debt relief operations. 39 However, net official flows (as share of recipients’ GDP) seem to have been on a downward trend in recent year.

53. This strong increase in private capital flows has been driven mainly by FDI and private transfers (Figure A.I.2). Both types of flows have increased steadily and each now accounts for over a third of total private flows on average. Portfolio investment, which was negligible in the 1980s, also increased significantly in the past few years, particularly in 2007.

Figure A.I.2.Private Net Flows to LICs

Source: based on Dorsey and others (2008).

54. The decline of official flows (as a share of recipients’ GDP) reflects a decrease in bilateral net lending (Figure A.I.3). According to GDF data, bilateral official net lending has almost disappeared (Figure A.I.4).40 The survey of donors confirmed that most traditional donors do not lend any more, a trend which is also confirmed by DAC data (Figures A.I.56). However, the existing data sources which provide detail on the origin of official lending—something the WEO does not do—probably do not capture the full extent of lending by nontraditional donors.41 As a result, the size of bilateral official lending, as well as the importance of nontraditional donors in new lending, are likely significantly underestimated.

Figure A.I.3.Net Official Flows to LICs

Source: based on Dorsey and others (2008).

Figure A.I.4.Net Lending to LICs’ Public Sectors

Source: GDF.

Figure A.I.5.Bilateral ODA Net Lending

Source: DAC.

Figure A.I.6.Share of Grants in Multilateral and Bilateral ODA

Source: DAC.

Figure A.I.7.Share of Concessional Lending in Total Lending to LICs

Source: GDF.

55. Bilateral official financing not provided in the form of lending has held up (Figure A.I.6). Such financing corresponds to grants and debt relief operations. The latter have increased significantly in the past few years, reflecting some large operations (e.g., the Paris Club’s cancellation of a large part of Nigeria’s external debt).

56. Multilateral institutions remain a significant source of official lending (Figure A.I.4). According to GDF data, such net lending to LICs’ official sectors increased significantly in the early 1990s and exceeded on average 1 percent of GDP in the past few years. According to this same source, multilateral lending has become increasingly concessional in past decades (Figure A.I.7). However, it should be noted that this is according to these institutions’ definition of concessionality, which may not be identical across institutions, may have changed over time, and may not be identical to the Fund’s. DAC data (Figure A.I.6) indicate that the share of grants provided by multilateral institutions decreased in the 1990s and recovered in the past few years (the 2006 surge reflecting the implementation of the MDRI).

57. Official flows remain the main source of LIC government financing by far. Beyond the financing provided in the form of transfers (grants) and debt relief, official donors and creditors remain the main sources of LIC governments’ gross borrowing (Figure A.I.8). Borrowing from private external creditors is still limited on average but was (at least through 2007) significant in some LICs.42Figure A.I.9 focuses on foreign banks’ net flows to LICs and show that these flows, including to official sectors, have been quite volatile.

Figure A.I.8.Gross Borrowing of LICs’ Public Sectors

Source: GDF.

Figure A.I.9.Foreign Banks’ Net Capital Flows to LICs by Sector

Source: BIS.

58. Reflecting debt relief and the composition of new financing flows, the structure of external debt has changed significantly in post-MDRI countries. The share of debt owed to non-Paris Club creditors (official or private) has increased, while that of debt owed to Paris Club creditors is close to zero. The share of multilateral debt remains high.

Appendix II. The Debt Sustainability Outlook in LICS

59. Debt Sustainability Analyses (DSAs) performed under the Debt Sustainability Framework (DSF) provide a comprehensive view of the debt outlook of LICs. 194 DSAs were completed between the introduction of the DSF in 2005 and December 2008 covering 68 LICs; 160 DSAs were published.43 All recent DSAs but three include both an external and a public debt DSA. Using data from the latest available DSAs, this appendix provides an overview of the current external debt situation of LICs, the external debt sustainability outlook, and the domestic and total public debt position.44

External debt situation of LICs at end-2007

60. External debt ratios in LICs remain, on average, sizeable. LICs’ total external debt—the sum of public and publicly-guaranteed (PPG) external debt and private-sector external debt— amounted to 60 percent of GDP on average at end-2007 (Table 1). PPG external debt is still by far the main component of total external debt. When expressed in PV terms, which capture its concessional nature, PPG external debt stood on average at about 43 percent of GDP at end-2007 (or 161 percent of exports), and total external debt slightly below 50 percent of GDP (Figure A.II.1).45 Debt service payments on PPG external debt absorbed 9 percent of export earnings and 12 percent of fiscal revenue in 2007. The remainder of this section will focus on the ratio of the PV of PPG external debt to exports (henceforth the PPG external debt ratio), which is the most critical ratio for a large majority of LICs in external DSAs.46

Figure A.II.1.Breakdown of LICs’ Present Value of External Debt-to-GDP

(average; in percent)

Sources: Latest DSAs available for LICs.

61. The averages mask, however, wide disparities. The standard deviation of the PPG external debt ratio was 327 percent on average at end-2007. This ratio ranges 8 percent for Nigeria to 2,503 percent for Liberia.47 Excluding pre-completion point HIPCs, whose debt ratios are very high, the average debt ratio falls by more than half to 77 percent (Figure A.II.2).48

Figure A.II.2.LICs’ Present Value of External Debt-to-Exports

(average; in percent)

Sources: Latest DSAs available for LICs.

62. Post-MDRI countries have lower debt ratios than non-HIPCs. The PPG external debt ratio for post-MDRI countries averaged 65 percent at end-2007 against 91 percent in non-HIPCs. The distribution of ratios across countries is also narrower in post-MDRI point countries (Figure A.II.3). The higher average debt ratio in non-HIPCs does not seem to be related to weaker export performance. The latter has been stronger over the past 10 years, and their current exports-to-GDP ratios are higher, than those of post-MDRI countries (Table 2).

Figure A.II.3.Average Present Value of External PPG Debt-to-Export Ratio

(In percent; in 2007)

Sources: Latest available DSAs for LICs.

63. The most indebted countries appear to be those facing greater economic and institutional challenges. Countries tend to have higher PPG external debt ratios when they have: (i) no or limited natural endowments (such as hydrocarbons and minerals); (ii) lower GNI per capita (defined as below the IDA cutoff of $1,095); (iii) weaker quality of policies and institutions measured by the World Bank’s Country Policy and Institutional Assessment (CPIA); and (iv) less friendly business environment measured by the World Bank ease of doing business indicators.

Debt Distress Rating and Vulnerability

64. According to recent DSAs and their ratings, about 30 percent of LICs have a low risk of external debt distress (Figure A.II.4). This share is higher for non-HIPCs (39 percent, or 11 countries) and post-MDRI countries (43 percent, or 10 countries), while no pre-completion point HIPC has a low risk rating. The slightly better performance of post-MDRI countries compared with non-HIPCs reflects to a large extent the provision of debt relief, which has decreased considerably their external PPG debt ratio. The quality of policies and institutions across the two groups, however, is comparable.

Figure A.II.4.Risk of Debt Distress in LICs

(Number of countries)

Sources: Latest DSAs available for LICs.

65. Another 30 percent of LICs have a moderate risk rating. (Figure A.II.5). This share is again higher for post-MDRI countries (43 percent, or 10 countries) than for non-HIPCs (25 percent, or 7 countries) and pre-completion point HIPCs (about 12 percent, or 2 countries). In these countries debt dynamics appear particularly sensitive to shocks to exports, leading to a breach of the DSA threshold in about 40 percent of cases (Table 3).

Figure A.II.5.Moderate-Rated Debt Distress Countries

(Number of countries)

Sources: Latest DSAs available for LICs.

66. Debt sustainability is a major concern for the 40 percent of LICs rated at high risk or in debt distress. Of these 28 countries, 15 are pre-completion point HIPCs (88 percent of this country group), 10 are non-HIPCs (36 percent), and 3 are post-MDRI countries (13 percent). Among the latter three, Burkina Faso and São Tomé and Príncipe were reclassified from moderate to high risk in the latest DSA. The downgrading reflects a decline in the CPIA for Burkina Faso and revisions to the timing and level of oil production for São Tomé and Príncipe. The Gambia has had a high risk rating since its completion point. Among the non-HIPCs, two countries are in debt distress and eight countries have a high risk of debt distress. Countries with higher risk ratings generally had debt ratios well above the DSA thresholds at end-2007. This was the case of all pre-completion-point countries (but Chad and Congo, Rep.) and nine out of 13 of the post-MDRI and non-HIPC countries.49 Only four countries had debt ratios below DSA thresholds at end-2007, but expected to exceed them in the future. For these countries, high initial debt ratios make debt dynamics explosive when sensitivity analyses are conducted. The export shock is the most extreme stress test for all countries but two. These countries appear also quite vulnerable to changes in the volume and terms of external financing.

67. Higher-risk countries share a number of characteristics that limit their capacity to carry debt, but are also a diverse group from an economic perspective. Excluding the pre-completion-point countries, the countries that experienced a breach of the debt thresholds at end-2007 had lower average export performance in the past 10 years than those with low or moderate ratings. Their export base is generally narrower, mainly concentrated on commodities. Lastly, the quality of their policies and institutions is significantly lower, with an average CPIA of 3 (making them, on average, “poor performers”), against 3.6 for the other countries. However, there is also a broad diversity of economic situations. Some of these countries are very poor (pre-completion-point countries have an average GNI per capita of $650 excluding those countries, the GNI per capita ranges from $320 in The Gambia to $4,670 in Grenada). Some countries benefit from large foreign direct investment (such as São Tomé and Príncipe with the current oil exploration) while others do not (e.g., Yemen). There are even large variations within this group in terms of CPIA (with Dominica being a strong performer) and business environment (Tonga ranked 43rd and Lao PDR 165th on the World Bank’s doing business indicators).

Public Government Debt

68. DSAs are used for the first time here to analyze domestic government debt across LICs. The availability of data on public domestic debt in LICs has been so far quite limited. The most recent analysis presented to the Executive Board dates from 2006 and was based on an ad-hoc database not derived from DSAs.50 The following analysis is based on gross domestic debt data.51

69. The use of DSAs promotes cross-country comparability but aggregation of data continues to be challenging. Due to data limitations and difficulties in record keeping, the definition and coverage of domestic debt may differ across LICs. Data limitation may limit the coverage to the central or general government in many countries. To reflect the difficulty of applying the residency criterion in many LICs, domestic debt is often defined as domestically-issued debt in DSAs.

70. Domestic government debt in LICs appears, on average, moderate but the average conceals again wide disparities. The distribution of the domestic debt-to-GDP ratio at end-2007 had a mean of 16 percent and a median of 13 percent (Table 4). The dispersion was quite wide, with the debt-to-GDP ratio ranging from zero to 92 percent. About 30 percent of LICs had domestic debt above 20 percent of GDP. Across country groups, pre-completion point countries had, on average, the largest domestic debt (Figure A.II.6).

Figure A.II.6.Public Domestic Debt-to-GDP

(Average; in percent)

Sources: Latest DSAs available for LICs.

71. Domestic debt represented, on average, less than 30 percent of total public debt in 2007. This share was higher in post-MDRI countries—reflecting external debt relief—than in pre-completion-point countries, which faced an external debt overhang, or in non-HIPCs (Figure A.II.6). Reflecting domestic debt’s higher costs and shorter maturities, domestic debt service payments represented about 40 percent of total debt service obligations on public debt.

72. On average, domestic debt has remained broadly stable in the past few years. The average ratio of domestic debt to GDP marginally increased from 15 percent in 2005 to 16 percent in 2007. 19 out of 65 countries have experienced an increase in their debt ratios with a mean of 2.6 percentage points. Four countries recorded much larger increases during this period: Ghana (+8 percentage points), Sudan (+7 percentage points), St. Lucia (+5 percentage points), and Togo (+3.5 percentage points).

73. Countries with higher level of domestic debt tend to share common attributes. These include (i) higher debt distress risk rating; (ii) higher per-capita income; (iii) weaker policies and institutions; and (iv) more developed financial system (as measured by the ratio of broad money to GDP). However, the differences among country groups are not as striking as for external debt.

74. The characteristics of the LICs with the highest levels of domestic debt are diverse, reflecting that the accumulation of domestic debt may fulfill different policy objectives (financing government spending, mopping up liquidity, developing the financial market). Non-HIPCs with high domestic debt tend to be richer, the quality of their policies and institutions stronger, and their financial intermediation greater than the average of their group.52 Conversely, HIPCs with high domestic debt tend to be poorest of their group with weaker quality of policies and institutions (Table 5).

75. At 74 percent of GDP, nominal total public debt remains high on average (Table 6). In PV terms, the ratio averaged almost 60 percent. The distribution of public debt ratios across country groups reflects largely that of external debt. Post-MDRI countries exhibited a lower ratio (32 percent) compared to non-HIPCs (47 percent). Pre-completion point countries are those with the largest public debt ratios (Figure A.II.7). Countries with low risk ratings had, on average, lower debt ratio than moderate-rated countries (by 18 percentage points) and countries with high and in debt distress ratings (by 67 percentage points).

Figure A.II.7.The Average Public Debt-to-GDP

(in percent)

Sources: Latest DSAs available for LICs.

Table A.II.1.Debt Indicators by Group of Countries at end-2007
Nominal external total debt-to-GDPPresent value (PV) of PPG debt-to-GDPPV of PPG debt-to-exports of goods and services (G&S)PV of PPG debt-to-revenue 1/2/Debt service on PPG debt-to-exports of G&SDebt service on PPG debt-to-revenue 1/2/
RatioStd dev.RatioStd dev.RatioStd dev.RatioStd dev.RatioStd dev.RatioStd dev.
LICs59.891.643.286.5161.2326.9180.6191.49.011.711.812.1
MACs 3/40.925.1..............................
HIPCs vs. Non-HIPCs
HIPC70.9117.350.6111.4210.7417.6204.7216.410.814.513.114.1
Non-HIPC44.022.532.520.790.666.9145.5147.36.14.39.67.9
HIPC status
Post-MDRI33.826.916.915.764.944.690.068.56.17.28.411.3
Pre-completion point121.0167.096.2161.5408.0591.9357.7257.217.219.019.415.4
Debt distress category
Low and moderate33.320.420.616.455.832.296.576.84.53.77.99.0
Low29.016.214.66.745.623.268.938.54.44.17.910.8
Moderate38.123.727.321.067.237.2128.995.84.63.57.96.4
High and in debt distress97.7133.075.3127.9311.8473.0322.1242.915.816.019.014.1
High53.726.834.919.1171.6129.2186.190.011.78.712.16.1
In debt distress165.7195.5137.7191.1528.5701.6518.5266.121.521.727.416.7
Natural resources 4/
Rich countries46.236.330.026.587.597.7166.5195.66.87.611.912.6
Hydrocarbon-rich43.336.429.125.788.7101.6136.1136.36.98.78.38.5
Mineral rich50.138.431.429.485.898.9193.1242.66.76.315.015.2
Non-rich countries65.1105.448.2100.4189.8377.6185.4192.19.813.011.712.1
Income Classification 5/
above $1,09545.821.633.522.184.060.0113.487.17.65.013.012.9
below $1,09564.6108.046.1102.1189.2384.0205.9216.29.313.511.312.1
Experience with bond issuance on international market
Yes38.123.828.421.163.347.297.167.45.73.79.26.3
No63.197.645.492.4176.2348.4195.7203.59.512.512.212.9
Quality of Institutions and policies 6/
Strong37.517.927.922.369.042.0113.295.46.85.213.414.1
Medium36.223.622.318.372.446.1101.373.44.73.46.14.7
Weak71.162.147.646.3210.9220.1281.5242.912.310.416.814.3
Ease of doing business 7/
Higher ranking (15–129)45.129.129.622.887.868.7135.9148.36.55.910.08.7
Weaker ranking (130–180)72.9122.255.2116.4226.5437.1218.4218.111.214.913.314.4
Sources: The latest DSAs; or latest data reported in country staff reports.

Revenue excluding grants.

Indicator was not reported in the DSA tables for 27 countries.

MACs stands for Market-Access Countries. The average excluding Iraq, Lebanon, Libya, and Venezuela.

According to the Guide to Resource Revenue Transparency.

Reflects the IDA income threshold using the GNI per capita.

Reflects the 2005–07 average of the World Banks’ CPIA.

Sources: The latest DSAs; or latest data reported in country staff reports.

Revenue excluding grants.

Indicator was not reported in the DSA tables for 27 countries.

MACs stands for Market-Access Countries. The average excluding Iraq, Lebanon, Libya, and Venezuela.

According to the Guide to Resource Revenue Transparency.

Reflects the IDA income threshold using the GNI per capita.

Reflects the 2005–07 average of the World Banks’ CPIA.

Table A.II.2.Macroeconomic and Institutional Indicators by Type of LIC and Debt Distress Categories
Number of countriesDistribution of hydrocarbon-and mineral-rich countriesExports of goods and services (G&S)-to-GDPGrowth rates of exports of G&SGrowth rate of GDPForeign direct investment-to-GDPCPIA 1/
in percent1998–2007 averages; in percent2005–07 averages; index
Average for all LICs68100.035.311.54.94.43.3
Non-HIPCs2831.642.711.75.25.03.4
Post-MDRI2336.829.710.55.44.03.5
Pre-completion point1731.630.712.43.74.12.8
Low2131.635.014.05.84.13.7
Non-HIPCs1115.838.515.26.34.43.8
Post-MDRI1015.831.212.75.33.83.7
Moderate1931.642.910.95.24.03.4
Non-HIPCs710.560.613.35.45.53.3
Post-MDRI1015.831.89.95.43.43.5
Pre-completion point25.336.57.14.01.63.5
High1726.329.39.44.56.03.1
Non-HIPCs85.331.96.94.16.23.4
Post-MDRI35.318.05.35.56.33.3
Pre-completion point615.831.514.94.65.52.6
In debt distress1110.532.010.83.13.42.6
Non-HIPCs20.046.05.83.31.71.7
Post-MDRI00.0...............
Pre-completion point910.528.911.93.13.72.7
Sources: The latest DSAs available; or latest data reported in country staff reports.

Reflects the World Bank’s Country Policy and Institutional Assessment (CPIA).

Sources: The latest DSAs available; or latest data reported in country staff reports.

Reflects the World Bank’s Country Policy and Institutional Assessment (CPIA).

Table A.II.3.Performance of Moderate, High, and in Debt Distress Countries under Selected Stress Tests
Export shock 1/Financing shock 2/Financing shock 3/
Debt Distress CategoryCategory of countriesShare of countries that breached the threshold (in percent)Average deviation in PV of debt-to-export ratio with respect to the baseline (in percentage points)Share of countries that breached the threshold (in percent)Average deviation in PV of debt-to-export ratio with respect to the baseline (in percentage points)Share of countries that breached the threshold (in percent)Average deviation in PV of debt-to-export ratio with respect to the baseline (in percentage points)
ModerateOverall3754.215.824.61122.2
Non-HIPCs1451.7012.8-15.0
Post-MDRI5054.42027.91023.9
Pre-completion point5062.15039.75030.4
HighOverall76118.35339.74730.4
Non-HIPCs7583.93841.33836.0
Post-MDRI6783.26720.16741.6
Pre-completion point83181.66747.25017.2
In debt distressOverall91150.19128.98219.1
Non-HIPCs10043.610012.11002.1
Post-MDRI.........
Pre-completion point89173.88932.67822.9
Sources: Latest DSAs available for LICs.

Export value growth at historical average minus one standard deviation in the 2nd and 3rd years of projections.

Net non-debt creating flows at historical average minus one standard deviation in the 2nd and 3rd years of projections.

New public sector loans on less favorable terms for all the projected years.

Sources: Latest DSAs available for LICs.

Export value growth at historical average minus one standard deviation in the 2nd and 3rd years of projections.

Net non-debt creating flows at historical average minus one standard deviation in the 2nd and 3rd years of projections.

New public sector loans on less favorable terms for all the projected years.

Table A.II.4.Domestic Debt Indicators by Group of Countries (in percent)
Domestic debt-to-GDP (2005)Domestic debt-to-total public debtDomestic debt-to-GDP (2007)Domestic debt-to-total public debtDomestic debt-service-to-total public debt service (2007)
AverageMedianStd dev.AverageMedianStd dev.AverageStd dev.
LICs15.012.111.718.715.712.714.828.139.223.6
HIPCs vs. Non-HIPCs
HIPC14.711.012.314.216.112.216.628.339.724.2
Non-HIPC15.315.311.224.415.015.611.527.838.023.0
HIPC status
Post-MDRI15.613.610.615.814.912.710.536.751.122.5
Pre-completion point13.38.115.311.517.811.123.116.324.818.0
Debt distress category
Low and moderate13.510.59.419.213.913.79.933.146.623.8
Low11.210.07.819.312.18.89.434.151.521.6
Moderate16.317.910.619.015.814.610.332.141.325.9
High and in debt distress17.313.114.618.018.212.419.721.128.419.3
High16.413.113.619.719.712.423.326.037.618.4
In debt distress19.415.618.213.815.913.613.113.218.115.2
Natural resources 2/
Rich countries12.18.58.617.912.212.19.528.846.928.6
Hydrocarbon-rich9.07.06.116.711.77.09.428.761.824.6
Mineral rich16.314.410.019.412.814.410.328.938.428.8
Non-rich countries16.012.912.619.017.013.216.227.936.521.5
Income Classification 3/
above $1,09517.111.612.621.717.416.013.429.942.518.0
below $1,09513.711.511.317.414.712.515.326.938.825.3
Broad money-to-GDP
Above the average23.220.112.427.022.619.319.231.035.319.2
below the average10.27.98.313.911.48.89.226.341.926.3
Experience with bond issuance on international market
Yes15.913.113.422.722.326.115.038.746.119.2
No14.812.111.618.014.912.214.726.838.324.2
Quality of Institutions and policies 4/
Strong15.212.69.622.715.215.59.332.031.110.0
Medium14.010.511.018.115.513.211.833.449.723.3
Weak15.811.514.316.916.412.119.921.929.721.3
Ease of doing business 5/
Higher ranking (15–129)16.215.310.523.216.115.011.431.939.520.7
Weaker ranking (130–180)13.89.412.914.615.310.617.524.738.926.5
Sources: Latest available DSAs; or latest data reported in country staff reports.

Domestic debt is defined as domestically-issued debt

According to the Guide to Resource Revenue Transparency.

Reflects the GNI per capita according to IDA income threshold.

Reflects the 2005–07 average of the World Banks’ CPIA.

Reflects the World Bank Doing Business Indicators.

Sources: Latest available DSAs; or latest data reported in country staff reports.

Domestic debt is defined as domestically-issued debt

According to the Guide to Resource Revenue Transparency.

Reflects the GNI per capita according to IDA income threshold.

Reflects the 2005–07 average of the World Banks’ CPIA.

Reflects the World Bank Doing Business Indicators.

Table A.II.5.Domestic Debt and other Indicators for Countries with High Domestic Debt(in percent, unless otherwise indicated)
Domestic debt-to-GDP (median in 2007)GNI per capita (In U.S. dollars)Financial intermediation 2/CPIA 3/(Index)GDP growth rates (1998–2007 average)
Most indebted countries 4/Average of groupMost indebted countries 4/Average of groupMost indebted countries 4/Average of groupMost indebted countries 4/Average of groupMost indebted countries 4/Average of group
Non-HIPC31.115.63395183756.244.43.73.53.75.2
HIPC26.112.254258343.033.03.13.23.74.6
Post-MDRI27.212.772163738.731.23.53.55.15.3
Post-decision point23.810.026344031.329.92.62.71.64.1
Pre-decision point92.212.0230617123.844.82.52.80.33.0
Sources: Latest available DSAs; or latest data reported in country staff reports.

Domestic debt is defined as domestically-issued debt.

Measured as the broad money-to-GDP ratio.

Reflects the 2005–07 average of the World Banks’ CPIA.

Defined as countries that have a domestic debt above 20 percent of GDP.

Sources: Latest available DSAs; or latest data reported in country staff reports.

Domestic debt is defined as domestically-issued debt.

Measured as the broad money-to-GDP ratio.

Reflects the 2005–07 average of the World Banks’ CPIA.

Defined as countries that have a domestic debt above 20 percent of GDP.

Table A.II.6.Public Debt Indicators by Group of Countries (in percent, 2007)
Public debt-to-GDPPresent value of public debt-to-GDP
AverageMedianStd dev.AverageMedianStd dev.
LICs73.649.4104.059.136.991.8
MACs 2/51.639.938.6
HIPCs vs. Non-HIPCs
HIPC87.349.8129.766.433.9114.9
Non-HIPC51.448.023.347.138.225.3
HIPC status
Post-MDRI44.032.228.531.822.922.5
Pre-completion point149.579.5185.4116.267.0168.0
Debt distress category
Low and moderate42.340.621.735.233.320.7
Low34.232.213.026.624.012.0
Moderate50.949.625.844.235.724.2
High and in debt distress118.177.7150.493.164.7135.0
High72.969.943.051.643.736.5
In debt distress190.492.3224.4159.588.9201.0
Natural resources 3/
Rich countries55.243.434.142.034.427.0
Hydrocarbon-rich51.140.935.740.27.030.8
Mineral rich59.953.634.044.214.424.0
Non-rich countries80.450.9119.665.438.1105.8
Income Classification 4/
above $1,09555.049.230.549.643.530.0
below $1,09580.448.1120.062.134.4106.1
Financial sophistication
Broad money-to-GDP
Above the average82.666.569.067.451.356.0
below the average68.040.9121.253.929.2108.5
Experience with bond issuance on international market
Yes54.449.824.249.745.726.9
No76.048.1109.960.235.097.0
Quality of Institutions and policies 5/
Strong47.744.124.743.137.227.9
Medium46.340.326.638.231.225.9
Weak86.266.673.862.147.257.4
Ease of doing business 6/
Higher ranking (15–129)56.648.129.046.737.126.0
Weaker ranking (130–180)89.157.3140.370.433.9124.2
Sources: Latest available DSAs; or latest data reported in country staff reports.

Domestic debt is defined as domestically-issued debt

MACs stands for Market-Access countries.

According to the Guide to Resource Revenue Transparency.

Reflects the GNI per capita according to IDA income threshold.

Reflects the 2005–07 average of the World Banks’ CPIA.

Reflects the World Bank Doing Business Indicators.

Sources: Latest available DSAs; or latest data reported in country staff reports.

Domestic debt is defined as domestically-issued debt

MACs stands for Market-Access countries.

According to the Guide to Resource Revenue Transparency.

Reflects the GNI per capita according to IDA income threshold.

Reflects the 2005–07 average of the World Banks’ CPIA.

Reflects the World Bank Doing Business Indicators.

1This paper was prepared by a team led by Hervé Joly consisting of Christian Beddies, Kenji Hosono, Eteri Kvindraze, Marie-Hélène Le Manchec, Shannon Mockler, and Perry Perone. Overall guidance was provided by Hugh Bredenkamp and Dominique Desruelle.
2In the remainder of this paper, LICs are defined as PRGF-eligible countries, unless indicated otherwise.
3See “Applying the Debt Sustainability Framework for Low-Income Countries Post Debt Relief” (http://www.imf.org/external/pp/longres.aspx?id=3959).
4Selected Decisions and Selected Documents of the IMF (http://www.imf.org/external/pubs/ft/sd/index.asp).
5There are no equivalent guidelines on domestic debt limits. Accumulation of domestic debt is generally addressed through the monetary and fiscal programs, based on macroeconomic stability and crowding out considerations.
6The remainder of this paper will focus on debt limits in LICs, because this is where there is a particular need to review policy and practice. Practice in non-LICs will be discussed only when relevant to issues faced in LICs. The reform proposals made in Section IV could have implications for the policy and/or practice in non-LICs. Some of these implications are discussed in Section IV. Others could be addressed in the follow-up paper.
7A minimum concessionality requirement of 35 percent was included in recent Stand-by arrangements (SBAs) in LICs (Georgia, Honduras and Pakistan in 2008). The SBA with Honduras did not allow any nonconcessional borrowing, while Georgia’s had a limit of $250 million for such borrowing, and Pakistan’s a very high ceiling of $9.5 billion. Recent SBAs with lower middle-income countries included a variety of debt limits. Some had a minimum concessionality requirement of 35 percent, together with a nonzero limit on nonconcessional borrowing (e.g., Iraq and Paraguay). Nonzero limits were often set quite high, like in Pakistan’s case (e.g., Peru and Turkey). Other SBAs had no concessionality requirements, but limits on nominal external borrowing by the government (e.g., Gabon, and Seychelles). Finally, Ukraine’s SBA had no external debt limit at all, but a ceiling on the cash deficit of the government.
8In the case of PSIs, external debt limits are not performance criteria but assessment criteria. For EPCAs, external debt limits are part of the member’s quantitative macroeconomic framework.
9Such a design could theoretically lead to demand management issues, as there is no upper bound to the fiscal deficit. According to the 2005 review of PRGF program design in mature stabilizers fiscal deficits have, however, been smaller than envisaged, because of shortfalls in foreign project implementation. In addition, the import content of such projects is generally high, posing limited risks to macroeconomic stability. See “Monetary and Fiscal Policy Design Issues in Low-Income Countries” (http://www.imf.org/external/np/pp/eng/2005/080805m.pdf).
10The grant element embedded in PRGF loans has decreased since the introduction of the PRGF, due to the decline in interest rates on currencies comprising the SDR basket. It should be noted, however, that PRGF resources are provided to address balance of payments needs, not to finance development.
11The Guyana program in 2004 was the first to employ an indicative limit on the PV of PPG external debt.
12The discussion on “Applying the Debt Sustainability Framework for Low-Income Countries Post Debt Relief” (http://www.imf.org/external/pp/longres.aspx?id=3959) indicated that “consideration should continue to be given—on a case-by-case basis—to nonconcessional finance depending on the impact on debt sustainability and the overall strength of a debtor country’s policies and institutions, as well as of the quality of the investment to be financed and of the overall public expenditure program. While the availability of concessional financing will clearly be a consideration in this assessment, many Directors emphasized the need for prudence, implying that the lack of such financing should not be the only justification for supporting recourse to nonconcessional resources.” Since then, the proportion of programs with nonzero limits has remained broadly stable (at about one-third), indicating both the flexible treatment of concessionality limits as endorsed by the Board, as well as prudence in cases where such flexibility was not warranted. The size of nonzero debt limits has varied from very small (e.g., Mozambique) to several percentage points of GDP (e.g., Pakistan).
13Integration is assessed by staff, taking into account a number of considerations (e.g., the various parts of the financing package have the same intended use and purpose, are provided to the same borrower, and have interrelated disbursement schedules and other relevant cross-conditions). The overall concessionality of an integrated package of debt is then calculated using the weighted average of the grant elements of its various components.
14A concrete illustration of many of these issues can be found in the Republic of Congo’s Technical Memorandum of Understanding (TMU). (http://www.imf.org/external/pubs/cat/longres.cfm?sk=22747.0).
15These issues are discussed in detail in Dorsey, Thomas William, Tadesse, Helaway, Singh, Sukhwinder and Brixiova, Zuzana (2008), “The Landscape of Capital Flows to Low-Income Countries” IMF Working Paper No. 08/51( http://www.imf.org/external/pubs/ft/wp/2008/wp0851.pdf ).
16This issue has been raised forcefully in the past 2–3 years by LIC authorities, especially in post-MDRI countries, but also by creditors in the case of highly-profitable (and possibly cash-generating) infrastructure investments.
17Some of these donors, however, sometimes offer financing deals which may challenge LICs’ assessment capacity and raise transparency issues.
18Another possible critique of this approach is that the concessionality of a loan is measured based on the opportunity cost to the creditor, not the borrower. While a borrower perspective would in principle be more relevant for debt sustainability, it would raise significant issues in practice. The latter include the definition of appropriate benchmarks in countries with shallow domestic markets; how to address the exchange rate risk; and potentially unintended outcomes (e.g., for countries with high interest rates reflecting bad policies, meeting concessionality requirements would be easier for donors than in countries with prudent policies and hence low interest rates).
19One could argue here that debt limits were too loose in the past in a number of cases. This allowed some countries to build up excessive amounts of multilateral (concessional) debt, which had subsequently to be relieved under the HIPC Initiative. As noted in “Monetary and Fiscal Policy Design Issues in Low-Income Countries,” (http://www.imf.org/external/np/pp/eng/2005/080805m.pdf), insufficient attention had been paid to external viability in program design. The closer integration of debt limits with DSAs could in principle allow tailoring better debt limits to that kind of situation.
20The authorities could choose to opt for tighter debt limits than suggested by this approach, however, if they wish to use these limits as a commitment device.
21While this broad mapping of ratings and debt vulnerabilities would be expected to be applied, judgment would be used in some cases. For instance, ratings are based on ratios for external PPG debt. In cases where domestic public debt is assessed to increase the risk of debt distress, consideration could be given to shifting a country with a lower rating to the higher debt vulnerability category. Some judgment could also be used for countries in the moderate risk rating category to decide whether they have lower or higher debt vulnerabilities.
22“Untied” means here that the nonzero limits would not correspond to any specific project discussed ex ante with Fund staff. The limits could therefore be used freely by the authorities, including for general budget financing through borrowing on international markets. Higher “tied” limits would still require project assessments, preferably by the World Bank. Countries with no experience of nonconcessional borrowing would start with relatively small amounts to gain experience and build capacity.
23While the 35 percent threshold may be perceived as arbitrary, any other minimum threshold would be as arbitrary. Therefore, it is proposed here to keep the existing minimum threshold, which has the advantage of being well known and used by other creditors.
24For instance, the DSA could help determine how much annual nonconcessional borrowing would be consistent with keeping the risk rating unchanged.
25Other options than those presented in this paragraph have been explored as well but were deemed impractical. See Annex I.
26The averaging system, although simpler than a PV target, still requires much more capacity to implement than the current approach. In particular, it requires careful recording and planning of borrowing operations if the year-end target is to be met. Also, it requires a much more active handling of creditors by the authorities, to make up for less “gate keeping” by the Fund.
27These countries would likely be classified as “gap” or “blend” countries by IDA and as such would not be subject to IDA’s Non-Concessional Borrowing Policy. Gap countries are IDA-eligible countries with a Gross National Income (GNI) per capita above the operational cutoff for more than two consecutive years; they are eligible for “hardened” IDA terms (less favorable than regular IDA terns). Blend countries are IDA-eligible countries that are creditworthy enough to borrow from IBRD.
28See footnote 7 for a description of debt limits in recent Fund-supported programs in lower middle-income countries.
29These efforts include the Fund’s, particularly in the area of design and implementation of Medium-Term Debt Management Strategies (jointly with the World Bank).
30PEFA is a partnership between the World Bank, the European Commission, the UK’s Department for International Development, the Swiss State Secretariat for Economic Affairs, the French Ministry of Foreign Affairs, the Royal Norwegian Ministry of Foreign Affairs, and the International Monetary Fund. PEFA aims to support integrated and harmonized approaches to assessment and reform in the field of public expenditure, procurement and financial accountability (http://www.pefa.org).
31A recent study on “The Macroeconomics of Scaling-up Aid—the Cases of Benin, Niger, and Togo (http://www.imf.org/external/np/pp/eng/2008/091908a.pdf)shows that (i) the degree of scaling-up to meet the Gleneagles commitments varies considerably across countries; and (ii) the impact on growth could be significant, depending on absorptive and administrative capacity. The magnitude of scaling-up in some of these cases illustrates the potential implications for debt sustainability. In Niger for example, the conclusion was that the risk of debt distress could increase substantially if new aid is not sufficiently concessional, despite the substantial growth impact of scaled-up investment. This study has used both a traditional macroeconomic model (Benin and Niger) and a dynamic stochastic general equilibrium model (Togo).
32For instance, DSAs do not include thresholds for public debt, unlike for PPG external debt. This issue was discussed in detail in “Applying the Debt Sustainability Framework for Low-Income Countries Post Debt Relief” (http://www.imf.org/external/pp/longres.aspx?id=3959). An empirical analysis conducted at the time found that domestic debt clearly mattered for the risk of external debt distress. However, the integration of domestic debt into the DSF poses many conceptual and practical challenges.
33Essentially, excluding systematically nonresident purchases of domestic debt implies that domestic debt, for concessionality purposes, is defined as domestically-issued debt. This approach was adopted recently in the TMU for the PSI with Senegal (http://www.imf.org/external/pubs/cat/longres.cfm?sk=22596.0).
34In highly dollarized economies, consideration could be given to using the jurisdiction of issuance as a criterion.
35The proposed approach for debt limits does not modify the approach under other Fund policies (e.g., the lending into arrears policy and exchange restrictions under Article VIII), wherein case-specific determinations are made as to whether official entities, including public enterprises, are part of the public sector.
36Obviously this would apply only to public enterprises for which information is actually available.
37These criteria were proposed in staff papers discussing the issue of the coverage of the fiscal accounts (see for example “Public Investment and Fiscal Policy—Lessons from the Pilot Country Studies,” http://www.imf.org/external/np/pp/eng/2005/040105a.htm).
38In Dorsey and others (2008), official flows consist of liabilities to official creditors, official transfers, and debt forgiveness. Private flows are the sum of FDI, portfolio investment, liability to non-official creditors, and private transfer. “Un-allocated capital account transfers” reflects the residual in the capital account.
39Issues related to the recording of debt relief operations probably explains why Dorsey and others (2008) and WEO data show negative net official flows in 2006, the year the MDRI was implemented
40Figures 4–7 show official flows to LICs’ public sectors, since there is no database comprehensive enough to capture the overall trends of official flows to private sectors.
41For instance, the DRS, which relies on reporting by debtors, does not include data on, or significantly underestimate, major financing operations by nontraditional donors in some African countries.
42For example, in countries like Nigeria, Malawi, Zambia and Ghana, foreign holdings of domestic government debt are substantial, and some countries (Gabon, Ghana, and Seychelles) have recently issued international sovereign bonds (Regional Economic Outlook: Sub-Saharan Africa, April 2008, http://www.imf.org/external/pubs/ft/reo/2008/AFR/eng/sreo0408.htm).
43LIC DSAs are conducted for all PRGF-eligible countries, but nine (Azerbaijan, India, Kiribati, Maldives, Pakistan, St. Vincent and the Grenadines, Timor Leste, Uzbekistan, and Vanuatu). As these countries have greater access to external market financing (or in the case of Uzbekistan ample resources through commodity exports), market-access country DSAs are completed (these DSAs are more focused on liquidity considerations and key debt indicators are in nominal terms).
44The latest DSAs used are those that were issued and sent to the Board by December 31, 2008. For the few LIC DSAs that did not include a rating, a rating was assigned to facilitate groupings and comparisons. This assignment was done following the principles laid out in the guidance note on the application of the DSF.
45All the averages in this section are simple averages.
46This ratio is generally the one that drives ratings in DSAs, as it tends to be the one closest to its policy-dependent threshold.
47Out of the 68 LICs covered in the external debt database, there are 28 non-HIPC LICs and 40 HIPC LICs.
48Includes the post-decision point and the pre-decision point countries. For these countries, the debt ratios at end-2007 reflected only the impact of interim debt relief.
49This is the debt-to-GDP ratio which is breached in the case of Congo, Rep.
50See Appendix I of “Applying the Debt Sustainability Framework for Low-Income Countries Post Debt Relief” (http://www.imf.org/external/pp/longres.aspx?id=3959).
51Public DSAs can be presented on a gross or net basis depending on the countries’ circumstances. There are two countries for which DSAs were completed on a net basis, to reflect better the government’s financial assets associated with oil revenues (the Republic of Congo and Nigeria). They are therefore not covered in this section.
52Also, the three countries located in Central America and the Caribbean belong to this group.

Other Resources Citing This Publication