Back to cover

The World Bank’s Role in and Use of the Low-Income Country Debt Sustainability Framework

Chapter 2 | The Low-Income Country Debt Sustainability Framework

Introduced in 2005 and most recently updated in 2017, the joint World Bank–IMF LIC-DSF has been a cornerstone of debt sustainability analysis in IDA-eligible countries.1 The framework classifies countries based on their assessed debt-carrying capacity, estimates thresholds for selected debt burden indicators, formulates baseline projections and stress test scenarios relative to these thresholds, and then combines indicative rules and staff judgment to assign ratings for the risk of debt distress.

The economic context of many LIC-DSF countries had changed significantly between 2005 and 2017, which contributed to important gaps in the LIC-DSF. The financial landscape was evolving, with financing from non–Paris Club creditors, domestic markets, and international bond markets (particularly for “frontier” emerging economies) increasing in importance. As a result, LIC-DSF countries were increasingly exposed to a wider set of vulnerabilities, including from market volatility. As such, the 2017 review of the LIC-DSF maintained the basic structure of the Debt Sustainability Framework (DSF) but made modifications to ensure that the DSF remained relevant to the rapidly changing financing landscape facing low-income countries (LICs) and to improve the insights produced into debt vulnerabilities.

Reforms introduced in 2017 were intended to make the framework comprehensive, more transparent, and simpler to use, while enabling the DSF to better capture risks of debt distress (figure 2.1). As a result of the review, World Bank and IMF management (i) introduced realism tools; (ii) transitioned to a composite measure for debt-carrying capacity; (iii) improved the identification of debt distress episodes; (iv) introduced tailored scenario tests; (v) simplified debt indicators, thresholds, and standardized tests; (vi) expanded the assessment of risks; and (vii) enhanced guidance for the application of staff judgment (appendix B). A discussion of the aspects where the World Bank plays a role appears below.

Figure 2.1. Structure of the Low-Income Country Debt Sustainability Framework after 2017 Reforms

A Flow chart showing various changes made for the 2017 reforms of the Low-Income Country Debt Sustainability Framework.

Figure 2.1. Structure of the Low-Income Country Debt Sustainability Framework after 2017 Reforms

A Flow chart showing various changes made for the 2017 reforms of the Low-Income Country Debt Sustainability Framework.


Source: International Development Association and International Monetary Fund 2017a.

Note: CPIA = Country Policy and Institutional Assessment.* New features.

Process for Low-Income Country Debt Sustainability Analysis Preparation

The LIC-DSF Guidance Note describes the process for producing a DSA (IDA and IMF 2017a). The Guidance Note indicates that all LIC-DSAs should be produced jointly by IMF and World Bank staff and are expected to be submitted to both the IMF and the IDA Executive Boards, either for discussion or for information. A full LIC-DSA should be produced at least once every calendar year. For the World Bank, a DSA is needed for each IDA-eligible country every year to determine the IDA credit-grant allocation and to inform the application of the SDFP.2

A DSA is also required in some additional situations (even when an annual DSA has already been prepared). These circumstances include when an IMF-supported arrangement is prepared or when a modification is proposed to an associated performance criterion or waiver for noncompliance related to debt limits. For World Bank financing requests, an LIC-DSA is required when a country that is subject to IDA’s Non-Concessional Borrowing Policy (replaced in 2020 by the SDFP) seeks to obtain nonconcessional borrowing or when a country experiences a significant change in economic circumstances or borrowing assumptions (including because of conflict or natural disaster).

The Guidance Note specifies how IMF and World Bank staff are to coordinate in producing a DSA, based on their respective areas of expertise. According to the Guidance Note, the IMF “generally” takes the lead on medium-term macroeconomic projections (three to five years), whereas the World Bank takes the lead on longer-term growth prospects and, when required, on assessing the investment-growth relationship (IDA and IMF 2017a, 18). World Bank and IMF staff are required to agree on the broad parameters of a DSA, including growth and new borrowing, before the DSA draft is produced. If there are significant differences between IMF and World Bank projections, a dispute resolution mechanism is specified and can be used.

World Bank procedures for participating in the preparation and approval of the LIC-DSAs were adjusted in April 2021 to clarify the role of the World Bank in LIC-DSA preparation and to ensure that the process undergoes a sufficiently rigorous review. The updated Accountability and Decision-Making (ADM) framework guidelines clarified the relative roles of specific World Bank vice presidencies (Regions; Development Economics; Development Finance; Equitable Growth, Finance, and Institutions; and Operations Policy and Country Services) in the preparation and corporate review process and in coordination of the review with the IMF’s DSA preparation and approval process. The revisions established a more formal structure for DSA preparation, approval, and clearance within the World Bank and were intended to support greater internal contestability and integration of guidance from World Bank management. They could also potentially help strengthen data coverage and quality.

Realism Tools

The 2017 LIC-DSF reforms introduced a suite of four realism tools, intended to encourage examination of baseline assumptions by flagging differences from a country’s historical performance or cross-country experience. The tools include the following:

  1. Drivers of debt dynamics to analyze changes in debt over the past five years compared with projections over the next five years, including (i) a chart showing the evolution of 10-year projections of external and public debt to gross domestic product (GDP) ratios for three DSA vintages—the current DSA, the previous year’s DSA, and the DSA from five years past; (ii) decomposition of past (that is, previous five years) and projected (that is, over the next five years) drivers of external and public debt dynamics; and (iii) a breakdown of past debt forecast errors.
  2. Realism of planned fiscal adjustment to show how a country’s projected primary fiscal adjustment in the next three years compares with the distribution of observed primary fiscal balance adjustments over a three-year period for IMF-supported programs for LICs.
  3. Fiscal adjustment–growth relationship, which assesses the consistency of fiscal and growth assumptions in the medium term by comparing the baseline growth projection with growth paths that include the fiscal impact on growth calculated under a range of plausible multipliers.
  4. Public investment–growth relationship, which compares the contribution of public investment to growth over the historical period in the last DSA and in the current DSA for the next five years.

Realism tools signal when projections may require further justification or adjustment. Where projections in a DSA are significantly different from historical experience (for the specific country or similar countries), this should be discussed in the DSA write-up or motivate adjustments to the projection. All but two of the DSAs reviewed for this evaluation discussed the realism of planned fiscal adjustments. Realism tools are applied to medium-term projections; only the drivers of debt dynamics tool are applied over the longer term (that is, a 10-year horizon).

Country Classification and Debt-Carrying Capacity

The 2017 reform introduced a composite indicator score to classify countries according to their debt-carrying capacity into one of three categories (strong, medium, and weak). This classification represents the debt thresholds applied to the sustainability of debt under the baseline and stress tests. Before the 2017 reform, the LIC-DSF relied exclusively on the historical average of the country’s World Bank Country Policy and Institutional Assessment (CPIA) score to determine debt-carrying capacity. The composite indicator score uses a weighted average of the country’s CPIA score, real GDP growth, remittances, international reserves, and global economic growth. It captures both five-year historical averages and forward-looking five-year projections. Because the CPIA score is not projected, it is held constant for the purposes of the composite indicator.

There was an increase in the share of countries assessed as having strong carrying capacity immediately after the 2017 reforms. The share of countries classified as strong increased from 10 percent to 23 percent of the panel data set, whereas the share of those classified as weak decreased from 44 percent to 26 percent (figure 2.2). All things being equal, this suggests that the composite indicator approach—whereby real GDP growth, remittances, international reserves, and global economic growth are taken into account—presented a more favorable picture of the debt that LICs can bear compared with the CPIA score alone. This is not surprising given the relatively favorable outlook for these variables in 2018. This is consistent with the finding from the comparison of DSAs completed before and after the onset of COVID-19. After the COVID-19 pandemic started, there was a deterioration in the borrowing capacity of IDA-eligible countries according to the composite indicator. The share of countries classified as weak increased from 26 percent to 38 percent, whereas those rated as medium fell by 10 percentage points from 51 percent to 41 percent.

Figure 2.2. Debt-Carrying Capacity over Time (percent)

Panel a. Pie chart showing share of debt carrying capacity from 2015 to 2017 with 10 percent countries classified as strong.
Panel b. Pie chart showing share of debt carrying capacity from 2018 to 2019 with 23 percent countries classified as strong.
Panel c. Pie chart showing share of debt carrying capacity from 2020 to 2022 with 21 percent countries classified as strong.

Figure 2.2. Debt-Carrying Capacity over Time (percent)

Source: Independent Evaluation Group.

Note: CPIA = Country Policy and Institutional Assessment.

Tailored Tests and Customized Scenarios

As part of the 2017 reform, the DSF included three types of stress tests to gauge the sensitivity of projected debt indicators to changes in assumptions. The first type of test was a series of standardized stress tests that, through an adjustment to the DSA template, were automatically applied to all countries to assess the impact of temporary shocks on the evolution of debt burden indicators in both the external and the public DSAs. The second type of test is tailored stress tests, which consider risks that are common to only subsets of countries. The third type of test is an optional fully customized scenario that can capture idiosyncratic risks, if relevant.

Tailored tests are used for countries exposed to specific risks—for example, natural disasters, volatile commodity prices, and market-financing pressures. Although the DSA template provides a default shock, users can customize tailored stress test scenarios to the country context by considering the country’s historical experience with the specific types of shocks:

  • A natural disaster shock is applied to small states vulnerable to natural disasters and LICs that meet frequency criteria (two disasters every three years) and economic loss criteria (above 5 percent of GDP per year), based on the Emergency Events Database between 1950 and 2015.3 Some 36 percent of countries in the panel data set performed the natural disaster test.4
  • The commodity price test is applied to LIC-DSA countries for which commodities account for at least 50 percent of total goods and service exports over the previous three-year period. Some 42 percent of countries in the panel performed commodity price tailored tests.
  • A market-financing shock is applied to LIC-DSA countries with market access, including those that either have outstanding Eurobonds or meet the market access criterion for graduation from the IMF’s Poverty Reduction and Growth Trust but have not graduated because of serious short-term vulnerabilities. The market-financing shock assesses rollover risks resulting from a deterioration in global risk sentiment, temporary nominal depreciation, and shortening of maturities of new external commercial borrowing. Some 25 percent of countries in the panel applied the market-financing test.

The 2017 reform introduced use of customized stress tests to address specific risks not covered by the standardized tests. These tests allow users to fully customize debt paths for analyses that cannot be preprogrammed. This could include idiosyncratic risks, such as civil war or a major health crisis, large delays in investment projects that can adversely affect growth and fiscal revenues, contagion-related macroeconomic risks, or policy slippage, which could result in very different debt paths. Customized stress tests were used for 13 percent of countries in the sample panel data set.

Role of Judgment in Determining Risk of Debt Distress

The DSF guidelines allow for the use of judgment to arrive at a final risk rating. In particular, the guidelines note that judgment can help assess the gravity of threshold breaches and country-specific factors that are not fully accounted for in the model (IDA and IMF 2017a). They discuss several situations when judgment may be used, including in the interpretation of short-lived and marginal breaches; consideration of whether high-risk signals from public debt distress may affect the external risk rating; examination of risk signals from the market-financing pressures tool; assessment of the implications of nonguaranteed, private, external debt; availability of liquid financial assets; fragility, conflict, and violence; reserve pooling arrangements; availability of insurance type arrangements and state-contingent debt instruments; and level of confidence in the macro baseline.

Judgment can also be applied to long-term considerations. Although threshold breaches projected to occur in projection years 11–20 do not normally give rise to a rating downgrade, in exceptional circumstances, breaches may provide a rationale for changing a risk rating when “(i) such breaches are expected to be large, persistent and thus resulting in significant differences relative to historical averages; and (ii) [they] occur with a high probability despite occurring in the distant future. Such a situation could arise from trends that are not easily amenable to policy interventions, such as climate change, population aging, known changes in donor financing frameworks, or expected exhaustion of natural resources” (IDA and IMF 2017a, 46).

The application of judgment followed the LIC-DSF guidelines in the most recent DSAs. Across the 66 most recent DSAs for IDA-eligible countries, 14 (21 percent) applied judgment to modify the mechanical risk rating from the model. Of these, 8 were downgrades and 6 were upgrades (table 2.1). For 5 of these cases, the projection horizon was extended from 10 to 20 years to account for issues that were expected to arise only in the longer term—the impact of climate change (Haiti, Samoa, and Tuvalu) and a projected shift in financing toward debt (Afghanistan)—and to account for the implications of an expected fiscal cliff in FY23 (the Federated States of Micronesia). Other downgrades were due to high overall public debt vulnerabilities (Togo and Guinea-Bissau, where there were also substantial downside risks to the baseline) and potential deterioration in security issues or other fiscal pressures (Mali). Upgrades were related to removing the impact of the pandemic from historical averages for stress tests (Cambodia and Moldova), a breach being only two years and small (Benin), hydropower debt being akin to foreign direct investment (Bhutan), remittances being the key foreign exchange earner rather than exports (Nepal), and liquid assets being available in the Petroleum Fund (Timor-Leste).

Table 2.1. Application of Judgment in Most Recent Country Debt Sustainability Analyses


DSA Date

Application of Judgment Impact

Justification for Application of Judgment


May 2021


Extended to 20-year horizon due to projected shift in financing mix toward debt.


December 2020


Breach is only two years and small (and given Benin’s fiscal path).


April 2022


Majority of debt is linked to hydropower project loans from government of India akin to FDI, and the projected improvement of medium-term dynamics is because of hydro exports and revenues.


November 2021


Applies a customized stress scenario based on prepandemic stress test parameters to account for exceptional and transitory factors during 2020 related to the pandemic.


July 2021


Reflects vulnerabilities from high overall public debt and substantial downside risks to the baseline scenario.


December 2019


Extended to 20-year horizon and considers high probability of protracted and substantial threshold breaches from FY34 of the baseline scenario as well as Haiti’s institutional fragility and exceptional vulnerability to natural disasters.


February 2021


Customized scenario demonstrates Mali’s vulnerability to a change in security conditions or other fiscal pressures that could lead to larger fiscal deficits financed on nonconcessional terms.

Micronesia, Fed. Sts.

October 2021


The forecast horizon informing mechanical risk signals is extended to 20 years to take account of the longer-term implications of a possible fiscal cliff in FY23.


December 2021


Stress tests adjusted given the exceptional nature of the largely temporary impact of the pandemic, where 2020 was dropped from the calculations of historical average and variances.


December 2021


Judgment applied due to low ratios of present value of PPG external debt to GDP and PPG external debt service to revenue and how remittances, rather than exports, are the major source of foreign exchange to balance the current account and service external debt.


March 2021


Extended the projection horizon to 20 years given the high probability of a large and protracted breach under the baseline over the long run due to Samoa’s exposure to frequent natural disasters and the effects of climate change.


June 2021


Petroleum Fund is large relative to projected debt levels and debt service requirements, and its assets are liquid and accessible.


March 2020


Judgment was applied given high domestic debt vulnerabilities.

Source: Independent Evaluation Group.

Note: DSA = Debt Sustainability Analysis; FDI = foreign direct investment; FY = fiscal year; GDP = gross domestic product; PPG = public and publicly guaranteed.

  1. The Low-Income Country Debt Sustainability Framework has been jointly reviewed by International Monetary Fund and World Bank staff four times: in 2006, 2009, 2012, and 2017. The 2017 review was informed by a broad external consultation process, including dialogue with country authorities, staff of multilateral banks, members of the Paris Club, and civil society organizations. For the 2017 review, see IDA and IMF 2017b.
  2. The Guidance Notes also indicate how the International Development Association credit-grant allocation is usually determined based on the latest approved Debt Sustainability Analysis available as of the end of June.
  3. Emergency Events Database: The International Disaster Database, prepared by the Centre for Research on the Epidemiology of Disasters; see
  4. See appendix A for further details on the panel data set.